If you work in finance or insurance, the annual planning routine is probably so familiar that you could describe it without the presentation. Financial objectives are established, business units receive their targets, and managers translate them into sales expectations, recruitment requirements, budgets and individual performance commitments. Results are reviewed monthly, sometimes weekly. Deviations require explanations, corrective measures are agreed, and the process repeats throughout the year.

By June, the conversation may sound remarkably similar to the one held the previous year. Sales are below expectations, recruitment has taken longer than anticipated, costs have risen, and customers are behaving differently from the assumptions used in the forecast. Finance wants revised numbers, while senior management expects the responsible managers to explain how the business intends to recover the shortfall. Some of the difficulties have been reported for months. Others have become significant only recently, and genuine execution problems may be among them.

Then comes a familiar reminder: We agreed these targets together. It is a reasonable objection. Annual plans would be meaningless if managers could disregard their commitments whenever circumstances became difficult. But agreeing to a target involves several judgements that are rarely visible in the final number. Someone estimated the market opportunity, assessed the available resources, anticipated the costs and concluded that the proposed results could be achieved. The people responsible for delivery may have contributed to those calculations, negotiated the target or accepted a number largely determined elsewhere.

When performance falls short, the organisation has to determine what went wrong. The familiar reporting process is designed to establish why actual results differ from the approved plan. That comparison alone cannot tell management whether the problem lies in execution, changed economic circumstances or the assumptions on which the original target was approved.

Where Did January’s Numbers Come From?

An annual plan rarely begins with a single calculation. There may be a profit expectation from shareholders, a strategic growth ambition, a forecast prepared by Finance, a market opportunity identified by an operating division and a set of assumptions based on the previous year’s results. These elements eventually have to become one financial plan.

Consider the difference between three figures that may appear in the same planning presentation. A forecast estimates what is likely to happen under specified conditions. A target states what the organisation intends to achieve. A budget allocates resources and establishes the financial framework within which managers are expected to operate. The relationship between them can be perfectly sensible. A higher target may encourage management to pursue opportunities that would otherwise be missed. New investment may make the original forecast unnecessarily conservative. A productivity improvement may allow more business to be handled without a proportional increase in costs. But the target itself does not establish that these opportunities exist. The organisation still has to explain how the additional result will be produced.

In December 2024, Allianz announced financial ambitions for the 2024–2027 period, including annual growth of seven to nine per cent in core earnings per share and a core return on equity of at least 17 per cent. Alongside these shareholder-oriented measures, the insurer described the business assumptions expected to support them. Its property and casualty insurance division, for example, was expected to achieve annual revenue growth of six to seven per cent, together with specified profitability and efficiency improvements. The figures were accompanied by assumptions about individual business segments, productivity, capital generation and customer satisfaction. Allianz was describing a proposed route towards the financial result, rather than presenting a desired return without an accompanying business rationale.

By August 2026, the company was reporting strong first-half results and continued confidence in its strategic targets. That experience is worth considering because ambitious growth is not inherently unreasonable. A company with a competitive advantage, expanding demand or genuine efficiency opportunities may be able to sustain substantial increases in earnings. The important consideration is the evidence supporting a particular target.

Even in insurance, profit growth and sales growth are not interchangeable. A company can improve earnings through more accurate pricing, better control of claims costs, investment income, changes in the mix of business or greater operational efficiency. It may also increase revenue while profitability deteriorates. The distinction becomes less visible as the financial plan travels down the organisation. A group earnings objective can become a business-unit revenue target. That revenue target may become a requirement to sell more policies, recruit more advisers or conduct more customer meetings.

For such a plan to work, each conversion from one measure to the next requires an assumption. How many additional customers are available? What proportion will buy? How long will new recruits take to become productive? What will it cost to acquire and retain the additional business? Will the resulting transactions generate an acceptable return? These are questions that can be investigated. Once the numbers have been distributed across the management hierarchy, however, the underlying assumptions may become considerably less visible than the targets themselves.

We Agreed the Targets Together

The annual planning meeting is often described as a collaborative process. Managers present their expectations, negotiate resources, challenge particular figures and eventually accept the result. That agreement has organisational value. It establishes responsibilities and enables different parts of the company to coordinate their decisions. It also creates a difficulty when performance subsequently falls short.

Imagine a regional manager who has repeatedly improved her team’s sales results. She has introduced better working methods, recruited capable advisers and exceeded targets that initially appeared demanding. At the next planning meeting, last year’s figures are on the screen and another ten per cent is proposed. She knows the market has become more competitive, several experienced employees have left, and bringing new advisers up to the required standard will take time. While she is still working out where the additional business might come from, her manager says: “You’ve done exceptionally well. We need another ten per cent next year, but given your track record, I’m confident you could deliver even more. Let’s be conservative and settle on ten.”

She recognises the sandwich technique; she has used it herself with her own advisers. The encouragement may be entirely sincere, and the target may ultimately be achievable. Meanwhile, she is still calculating how many additional customers she would need, who could recruit the new advisers and how much time their training would take. Questioning the number now means challenging something that has just been described as the conservative option. Six months later, her agreement may be recalled more readily than the questions she was still trying to answer.

Research on goal-setting provides a more precise starting point. In a 2011 study published in The British Accounting Review, Mahfud Sholihin and colleagues examined 54 managers within a British financial services organisation. They found that participation in goal-setting was positively associated with commitment to the resulting goals, with perceived procedural fairness playing an important role. There is a good reason for organisations to involve managers in planning, then. Participation can support commitment. But the research did not establish that a jointly agreed target was a more accurate estimate of future economic performance.

Theresa Libby’s 1999 research into participative budgeting examined another relevant distinction. A manager may be given an opportunity to express an opinion without having the final authority to determine the budget. Libby investigated how that opportunity, and the explanation provided when a manager’s input does not influence the decision, affect perceptions of fairness and performance. An agreement reached after genuine negotiation is different from one reached after participants have been heard but the final number remains largely predetermined. Both may be described afterwards as a participative process.

Past success introduces a further complication. In three experiments published in The Accounting Review in 2023, Christoph Feichter investigated whether supervisors’ previous performance in lower-level jobs influenced the targets they subsequently established for employees. Supervisors who had themselves performed well set higher targets than supervisors with weaker previous performance, even when both groups received the same information about employees’ potential. The effect appeared when supervisors had experience of the particular task they were assigning. The findings suggest that managers can overestimate the general relevance of their own successful experience. This is especially interesting because the resulting expectations need not originate in deliberate pressure or bad faith. A successful manager may sincerely believe that others can reproduce her results.

The Wells Fargo investigation shows how the formal distribution of decision-making authority can create a much more serious problem. In April 2017, Wells Fargo’s independent directors published the findings of their investigation into the bank’s retail sales practices. The report documented an organisation strongly committed to year-on-year growth in the number of banking products sold. Some regional sales plans were internally known as 50/50 plans. According to the investigation, senior management expected only about half of the regions to meet them.

The regions did not necessarily determine their own targets. The degree of input from regional banking leaders varied over time, while final responsibility rested with the senior leadership of the Community Bank. The goals were subsequently passed down through the hierarchy. Regional managers, branches and individual employees were evaluated against them. Rankings, compensation and career opportunities were affected by performance relative to the targets.

Several regional leaders challenged the expectations. They raised concerns about unrealistic sales volumes and the increasing number of accounts that customers did not need or use. The independent investigation found that Community Bank leadership frequently interpreted resistance as an attempt to obtain easier targets or as evidence of poor local management. The result was an important difference between receiving an objection and treating it as information about the business.

The Wells Fargo case ultimately involved widespread improper sales practices, which the investigation linked to the bank’s sales culture, management and control failures. Its implications extend well beyond annual budgeting. Nevertheless, the report documents a point directly relevant to target-setting: participation in a management structure did not guarantee that those responsible for delivering the goals had sufficient influence over them, or that their warnings about feasibility received appropriate consideration. A company’s statement that everyone agreed the numbers may be entirely accurate. To understand what that agreement establishes, we also need to know how the numbers were determined, which objections were raised and what authority the participants had to change the result.

When Last Year’s Success Sets This Year’s Standard

Repeated target increases can emerge without anyone explicitly deciding that growth must continue indefinitely. A business performs well. The following year’s plan starts from that higher result. Management expects further improvement. The new target is distributed through the organisation, and another round of performance measurement begins. There may be good reasons for this progression. Improved technology can permanently reduce costs. Better training can raise productivity. A new market may provide additional customers for several years. But previous performance also reflects conditions outside management’s control. A favourable market, an unusually profitable contract or a temporary reduction in costs may make one year exceptionally successful. Unless those benefits can reasonably be expected to continue, building them into the next year’s starting point creates an additional demand on the business.

Management accounting research describes one form of this process as budget ratcheting. In 2002, Andrew Leone and Steve Rock published an analysis of business-unit budgets within a large multinational corporation. They found that favourable deviations from the previous budget were followed by larger increases in subsequent performance budgets than the reductions associated with similarly sized unfavourable deviations. The researchers also examined how these expectations could influence managers’ accounting decisions. Their findings concerned one corporation, and they recognised that adjusting future budgets to previous results may sometimes serve legitimate forecasting and incentive purposes. The practical difficulty arises when a favourable result is treated as evidence of permanent earning capacity without adequately examining its causes.

Wells Fargo provides an unusually clear illustration of how the problem can accumulate over time. Its independent directors’ report found that earlier sales totals included substantial numbers of low-quality accounts, some of which customers neither needed nor used. Those figures nevertheless contributed to the starting point for subsequent growth expectations. A target calculated from the previous year’s volume could therefore contain sales that had little genuine customer value. Maintaining growth from that enlarged baseline required even more accounts to be opened. Regional leaders argued that the increasing targets encouraged further low-quality sales, but senior Community Bank management did not adequately address their concerns.

This is more specific than saying that excessive ambition creates pressure. The quantity being used to justify the next target had become an imperfect measure of the economic activity management actually wanted. The same general issue can arise without misconduct. An insurer may increase premium income because it has raised prices to reflect inflation. A software company may report higher subscription revenue after renewing contracts at higher prices, despite losing customers. A manufacturer may increase sales revenue while selling fewer products. Those outcomes may improve or weaken profitability, depending on the economics of the business. None establishes by itself that the organisation has increased its underlying productive capacity.

For an annual plan built on continuing growth, the origin of the previous year’s improvement is therefore important. It determines how much of that result can reasonably be carried forward and what additional conditions must hold for the next increase to occur.

The Information Was Already Changing

The consequences of an outdated economic assumption can be seen in the performance of Direct Line Insurance Group during 2022. The British insurer entered that year after reporting strong results for 2021. Its weather-normalised combined operating ratio had been 91.1 per cent, ahead of its medium-term target range.

The terminology is technical, but the economics are straightforward. An insurer collects premiums and expects to pay claims and operating expenses. The combined operating ratio measures those costs against the corresponding insurance revenue. A ratio of 95 per cent means that approximately £95 of claims and relevant operating expenses are associated with every £100 of insurance revenue. At 103 per cent, the costs exceed that revenue by approximately £3. This measure describes the underwriting result. Investment income can still contribute to the insurer’s overall profitability. The weather-normalised version adjusts for differences between actual weather-related claims and a normal level, making it easier to examine the underlying performance of the business.

In March 2022, Direct Line reaffirmed its expectation of achieving a weather-normalised combined operating ratio of 93–95 per cent for the year. The company had reasons to expect continued progress. It had invested in pricing technology, developed its digital capabilities and reported a strong previous year. By the first quarter, however, costs in the motor insurance market were becoming more difficult to predict.

When an insurer sells a motor policy, it must estimate the likely cost of future claims. Some expenses, including vehicle repairs and compensation settlements, may arise well after the policy has been priced. If parts, labour and replacement vehicles become more expensive than expected, the insurer may have accepted premiums that are insufficient to cover the eventual costs. The number of accidents does not even have to increase. The average cost of each claim can rise.

On 4 May 2022, Direct Line acknowledged that motor insurance market prices did not fully reflect claims inflation. Its trading update referred to continued pressure from used-car prices, supply-chain disruption and repair costs. The same statement reported trading broadly in line with expectations and progress in several areas of the business. Management retained the 93–95 per cent target.

It would be tempting to describe this as evidence that the company was unwilling to confront the emerging problem. The public information available at that point does not establish such a conclusion. Management had identified the risk and was taking action, including developing new pricing capabilities. By July, however, the accumulated claims experience had changed the outlook.

In a trading update published on 18 July 2022, Direct Line reported that motor claims inflation during the first half of the year had exceeded the assumptions incorporated into its prices. It revised the expected full-year weather-normalised combined operating ratio to 96–98 per cent and described price increases and improvements in risk selection intended to restore margins.

The deterioration continued. When Direct Line published its 2022 results in March 2023, its weather-normalised combined operating ratio for ongoing operations was 103.3 per cent. In motor insurance, the reported combined operating ratio reached 114.7 per cent. The motor figure reflected particularly severe inflation in the cost of claims. Other business lines had performed considerably better.

On a comparable basis, operating profit from ongoing operations fell from £590.3 million in 2021 to £32.1 million in 2022. Higher-than-expected weather claims, market volatility and regulatory changes also contributed to the overall deterioration. The company’s own assessment was that it had not navigated the difficult conditions as effectively as it would have wished.

The sequence gives us several distinct decision points. In March, management reaffirmed the annual objective. By May, adverse cost developments were sufficiently clear to be discussed publicly, although the company still expected to meet its target. In July, actual experience had moved far enough away from the pricing assumptions to justify a revised outlook. The final result was considerably worse.

We cannot infer from the year-end numbers alone exactly what Direct Line’s management should have decided earlier. That requires examining the information reasonably available at each point, including how quickly pricing changes could affect new and renewed policies and how much uncertainty remained around future claims. But the case illustrates the central problem of an annual financial commitment. Its feasibility depends on economic relationships that continue to develop after the original plan has been approved. A company may report its performance against a fixed objective every month. The underlying economics are not fixed merely because the objective is.

Why More Effort May Not Close the Gap

Direct Line could respond to deteriorating margins through pricing decisions, changes in the risks it accepted, expense management and improvements in underwriting. It could not simply instruct vehicle repair costs to return to their earlier levels. Increasing premiums might improve the expected profitability of new business, but customers could respond by selecting another insurer. Lower premiums might protect policy volumes while leaving too little revenue to cover the cost of claims. The result depends on the balance between customer demand, pricing, claims costs and competition.

A simple illustration makes the problem visible. Suppose an insurer collects £500 for a policy and expects claims and relevant operating expenses of £450. On these simplified assumptions, the policy contributes £50 before investment income and any other items excluded from the calculation. If expected costs rise to £540 while the premium remains £500, the insurer is now expecting a £40 underwriting loss. Selling more policies with the same economics increases premium revenue, but also increases the expected loss. In that situation, demanding more sales activity may worsen the financial result the company is trying to improve.

The numbers are illustrative. They are not Direct Line figures. Their purpose is to show why sales volume and profitability must be examined separately. There are also circumstances in which greater activity is precisely the appropriate response. If potential customers exist, the product is competitively priced and advisers are failing to contact them, increasing prospecting activity may produce more business. Better recruitment, stronger training and improvements in customer service can also raise sustainable performance. The effectiveness of those measures depends on the relationship between the activity and the outcome.

An insurance sales organisation might measure the number of customer approaches, completed appointments, offers presented and contracts concluded. Management may then calculate what proportion of offers becomes a sale, how much it costs to acquire the business and how many customers retain their policies. When a target is missed, these measures can help determine what has changed. If advisers conduct fewer meetings than planned, the organisation has an activity shortfall. If the number of meetings is stable but fewer suitable customers purchase the product, the problem requires a different explanation.

Customers may have less disposable income, face more attractive alternatives or decide that a discretionary insurance product is no longer affordable. The original forecast may have assumed a conversion rate that the current market no longer supports. Insurance products also respond differently to economic conditions. Compulsory motor cover, household insurance and long-term life insurance serve different customer needs. Rising costs may prompt customers to shop around for cheaper motor insurance, reduce optional cover or reconsider longer-term savings commitments.

In its 2023 analysis of inflation, the European Insurance and Occupational Pensions Authority identified increasing claims and expense costs as pressures on non-life insurers. It also warned that weaker household saving capacity could affect life insurance sales and lead to more policy lapses. Those differences make blanket activity targets particularly difficult to evaluate without reference to the product and customer market.

The principle extends beyond insurance. Imagine a software company whose annual plan assumes that additional sales calls will generate more subscriptions. Its salespeople meet the expected activity levels, but prospective customers have begun reducing their technology budgets. Fewer enquiries become contracts, and retaining existing clients requires larger discounts and more customer support. The company could still achieve growth by improving its product or reaching a different market. Requiring the same sales team to make more calls, however, does not establish that those opportunities are available.

In both businesses, an initially plausible relationship between effort and output has to be tested against actual results. A performance report showing that the sales team is behind target identifies an outcome. It does not tell management which part of the business model needs to change.

The Work the Plan Did Not Fully Account For

Even where market demand is sufficient, an annual plan may underestimate the work required to generate the desired result. Consider an insurance sales organisation planning to increase its adviser headcount and new business production in the same year. The staffing plan projects how many new advisers will be recruited. The commercial plan estimates how much additional business they will generate. Training and management responsibilities are assigned to existing team leaders. On paper, the individual objectives may appear reasonable. The difficulty emerges when the same managers are responsible for delivering them simultaneously.

I encountered this tension in a qualitative study I conducted in 2020 into recruitment practices within the Hungarian operations of a European insurance group. The research involved interviews with eight insurance advisers, six team managers and two regional managers. It examined one organisation, with a small sample, and cannot establish the prevalence of similar difficulties across the insurance industry.

The organisation had formal recruitment and onboarding procedures, supported by a CRM system that recorded activities and monitored progress. Several managers nevertheless reported that prescribed mentoring, administrative and supervisory requirements could not always be completed as intended alongside their sales responsibilities. The company could track each required activity separately. The managers still had to find the hours to perform them all. An experienced team leader was expected to maintain the performance of an existing sales team while recruiting, training and supervising new advisers. Where turnover was high, the recruitment and training process had to be repeated for replacements.

The additional people were not immediately equivalent to an increase in productive capacity. New advisers required time and attention before they could work independently. A manager who devoted additional hours to this month’s sales results might therefore have fewer hours available for developing the people expected to produce future results. The competing work could be recorded as separate responsibilities, each with its own target and reporting process. It still required time from the same individual.

My study documented managers’ descriptions of those conflicts. It did not independently measure their total working hours or demonstrate that the conflicts caused a particular subsequent sales shortfall. Nevertheless, the findings provide an instructive example of what annual plans can leave unresolved. Recruiting ten additional advisers may be achievable. Increasing sales production may also be achievable. A structured onboarding process may be well designed. Whether all three can be completed by the same managers, within the same period and with the resources available, requires another calculation.

A delay in one activity can affect the others. If recruitment takes longer than anticipated, the expected new business may arrive later. If training is compressed to protect immediate sales activity, advisers may need additional support in subsequent months. If experienced staff leave, the organisation loses current production while increasing the need for replacements and mentoring. These effects are conditional. They do not occur in every team, and a company with adequate managerial capacity or efficient onboarding systems may avoid them. But where they do occur, the financial plan’s assumptions about when additional resources become productive can be as important as the eventual number recruited.

Short-term adjustments add another layer. A manager postpones training to deal with a customer problem. Another takes on additional responsibilities while a vacant position is filled. A project is delayed so that a more urgent commitment can be met. Such decisions may be sensible at the time. The postponed work, however, often remains part of the operating system. If it must be completed before another task can begin, the original timetable becomes increasingly difficult to maintain.

This is a different mechanism from a decline in market demand. It produces the same visible symptom: results begin to diverge from plan. By the middle of the year, a business may therefore be carrying both financial assumptions that no longer reflect the market and operational commitments that no longer fit the capacity available. A report comparing actual results with the original annual target cannot, on its own, distinguish between these conditions.

Who Has to Explain the Original Plan?

There are legitimate reasons for a company to retain ambitious targets during periods of uncertainty. One disappointing month may have little significance. Some costs are temporary. A new employee may take longer than expected to become productive and then perform exceptionally well. A weak quarter can be followed by a strong one. Constantly reducing targets would undermine planning discipline and create incentives to explain away ordinary performance problems. The alternative is to establish what evidence would support each explanation.

Suppose a sales director has missed the first-half revenue target. Senior management asks for corrective action. The report shows that the team has completed the planned number of customer meetings, but fewer meetings have produced contracts. That evidence does not immediately resolve the issue. The quality of the meetings, the mix of customers and the suitability of the proposed products may have changed. Some advisers may be less effective than others. The team had delivered the planned number of meetings, so management first needed to understand why fewer of them resulted in contracts.

The company needs to examine conversion rates by product, customer segment and period. It needs to understand whether prices, customer affordability, competitive conditions or adviser performance have changed. If the conversion rate used in the original plan is no longer supported by the evidence, management must decide whether a different commercial approach can restore performance within the remaining months.

The same reasoning applies to capacity. Suppose an additional recruitment programme was approved because the business expected more productive advisers by June. If recruitment targets were met but the new advisers were not yet capable of generating the planned sales, the organisation should examine the productivity assumptions behind the staffing plan. How much business was expected from a new entrant, after how many months, and with what level of managerial support? Alternatively, if the recruitment activity itself was not carried out, responsibility may lie in execution. Both situations produce a lower-than-planned adviser contribution. They call for different decisions.

The uncomfortable part comes when the investigation reaches the original planning process. What if the forecast depended on a conversion rate that had already been weakening? What if the plan assumed that experienced managers could recruit and mentor additional employees without reducing the time available for their existing responsibilities? What if a previous year’s exceptionally favourable result was treated as the normal starting point for another increase? These assumptions may have been reasonable when approved. They may also have been inadequately examined.

The answer cannot be established by asking only whether the business-unit manager signed off the plan. That manager may have contributed to an optimistic forecast, failed to disclose known constraints or accepted a commitment without adequately assessing it. Such decisions carry responsibility. But someone also had to consolidate the different plans, assess whether their resource requirements were compatible, and judge the business case supporting the financial objectives.

There are circumstances in which the leaders who authorised the target possessed information that the people responsible for delivery did not. There are others in which local management had the better information but did not communicate it clearly. The Wells Fargo investigation illustrates an additional possibility: warnings can reach senior management and still be interpreted primarily as attempts to negotiate easier targets. Direct Line illustrates a different response. Its management publicly acknowledged changing economic conditions and revised its annual outlook when the reported evidence no longer supported the original expectation. The year-end results also show that recognising a problem and correcting it are separate challenges.

An accurate performance review has to reconstruct these decisions in their proper sequence. What did management know when the target was approved? What information became available afterwards? Who was in a position to recognise its significance? What changes were proposed, accepted or rejected? The responses determine whether the shortfall is principally evidence of weak execution, an unsuitable original plan, adverse changes in the business environment or some combination of the three. Without that examination, the approved target risks becoming the one element of the performance review whose own validity is taken for granted.

What June Can Tell Us About January

Return to the mid-year meeting. The financial director presents the difference between actual results and the approved plan. The sales director explains the decline in conversion. HR describes recruitment delays. Operations reports the additional work required to support existing commitments. Senior management wants to know whether the annual target can still be achieved.

There is enough information to begin answering that question, although it may not be enough to resolve it immediately. The company can examine whether the number of available customers matches the original forecast. It can determine whether sales activity has taken place, whether its conversion into profitable business has changed and whether the operating organisation has enough capacity to complete the work remaining. It can also revisit the original assumptions about pricing, costs, staffing and the expected contribution from different business units.

If those assumptions remain well supported, the pressure to improve execution may be entirely justified. If they have changed, management needs to understand whether the effects can realistically be offset through pricing, productivity, investment, reprioritisation or a different commercial approach. And if the original assumptions were never adequately justified, the investigation reaches further back than the current reporting period.

The fact that a target was approved does not make it immune to examination. Nor does an unfavourable result establish that the original decision was wrong. Forecasting always involves uncertainty, and some adverse developments become identifiable only after the business has committed its resources. The necessary judgement concerns what could reasonably have been known, when that information became available and what the organisation did with it.

For a company committed to recurring growth, this examination becomes increasingly important over successive planning periods. Another year of higher earnings may be entirely achievable, but the economic source of the increase must be identifiable. Improved productivity, a larger addressable market, a stronger product, more effective capital allocation and sustainable pricing each provide a different rationale for growth. Raising the target offers no substitute for establishing which of those opportunities can support it.

The June meeting can still end with a demand for better execution. In some businesses, that will be the correct decision. But if management cannot explain the original assumptions, the information that supported them and the reasons for continuing to rely on them, it has an incomplete basis for deciding whose performance requires correction.

A mid-year review may establish that managers failed to carry out activities they had agreed to perform. It may also reveal that those activities were completed, but the expected commercial results did not follow. In the second situation, another round of performance demands will achieve little unless management understands why the assumed relationship between effort and results has broken down. The original target, the evidence supporting it and the decisions taken as that evidence changed all belong in the same examination of performance.

References

Corporate documentation

• Allianz SE (10 December 2024). Allianz lifts financial ambitions at its Capital Markets Day 2024.

• Allianz SE (7 August 2026). Second-quarter and half-year 2026 results.

• Direct Line Insurance Group (8 March 2022). Preliminary results for 2021.

• Direct Line Insurance Group (4 May 2022). First-quarter trading update.

• Direct Line Insurance Group (18 July 2022). Trading update.

• Direct Line Insurance Group (13 March 2023). Preliminary results for 2022.

• Independent Directors of the Board of Wells Fargo & Company (10 April 2017). Sales Practices Investigation Report.

• European Insurance and Occupational Pensions Authority (5 October 2023). Report on the Impact of Inflation on the Insurance Sector.

Academic research

• Feichter, C. (2023). The Effect of Supervisors’ Prior Task Performance on Employees’ Targets. The Accounting Review, 98(1), 191–214.

• Leone, A. J., & Rock, S. (2002). Empirical Tests of Budget Ratcheting and Its Effect on Managers’ Discretionary Accrual Choices. Journal of Accounting and Economics, 33(1), 43–67.

• Libby, T. (1999). The Influence of Voice and Explanation on Performance in a Participative Budgeting Setting. Accounting, Organizations and Society, 24(2), 125–137.

• Sholihin, M., Pike, R., Mangena, M., & Li, J. (2011). Goal-Setting Participation and Goal Commitment: Examining the Mediating Roles of Procedural Fairness and Interpersonal Trust in a UK Financial Services Organisation. The British Accounting Review, 43(2), 135–146.

Author’s research

• Gerlach, L. (2020). Egy magyarországi biztosító társaság toborzási tevékenysége 2000–2018 között [Recruitment Practices at an Insurance Company in Hungary, 2000–2018]. University of Sopron, dissertation. Qualitative study involving 16 interviews within one insurance organisation.

If one manager can be biased, ask more people. Logical. After all, only the manager can be biased. The eight colleagues we ask obviously cannot.

There is good reason to collect those additional perspectives. Peer and subordinate ratings can contain relevant performance information that is missing from a manager’s rating. In one meta-analysis, both sources provided significant incremental information about objective performance measures beyond that provided by other rating sources. [1]

Involving more people can broaden the information available. At the same time, every additional rating brings another human judgement into the system – with its own observations, interpretations and possible biases.

When the result feeds into a performance rating, a talent decision, promotion or pay, it matters whether we have produced better evidence or simply collected more ratings. Still, at least it shows that we are taking the matter seriously: we have produced an impressive amount of data.

In two large samples of managers receiving multisource developmental ratings, individual-rater effects accounted for 62% and 53% of the variance in ratings. [2] Later work challenged aspects of the model, while still finding substantial individual-rater effects. [3]

So adding peers does not simply add information. It adds raters.

Peer ratings therefore do not, by themselves, remove the problems associated with human judgement. They may introduce additional rater effects without necessarily removing the problems in the manager’s judgement.

Then comes the question of what we do with the differences.

I will spare you the rest of the dry research findings, but the nerds will find them in the references.

Reading this back, it looks pretty much like a critique, so you would be justified in asking: what now? Should we stop asking other people? Should we ignore what colleagues who actually work with someone can see? How would I do it?

There is not enough space on this platform to address that question properly and with sufficient evidence. If you are interested in the answer, I discuss it in detail in Judging Strangers: The Readability Trap in Hiring – including how human observations can be turned into decision-relevant information without confusing more data with better evidence.

Sources

[1] Conway, J. M., Lombardo, K. & Sanders, K. C. (2001). A Meta-Analysis of Incremental Validity and Nomological Networks for Subordinate and Peer Rating. Human Performance, 14(4), 267–303. DOI: 10.1207/S15327043HUP1404_1.

[2] Scullen, S. E., Mount, M. K. & Goff, M. (2000). Understanding the Latent Structure of Job Performance Ratings. Journal of Applied Psychology, 85(6), 956–970. DOI: 10.1037/0021-9010.85.6.956.

[3] Hoffman, B. J., Lance, C. E., Bynum, B. & Gentry, W. A. (2010). Rater Source Effects Are Alive and Well After All. Personnel Psychology, 63, 119–151. DOI: 10.1111/j.1744-6570.2009.01164.x.

[4] Conway, J. M. & Huffcutt, A. I. (1997). Psychometric Properties of Multisource Performance Ratings: A Meta-Analysis of Subordinate, Supervisor, Peer, and Self-Ratings. Human Performance, 10(4), 331–360. DOI: 10.1207/S15327043HUP1004_2.

[5] Harris, M. M. & Schaubroeck, J. (1988). A Meta-Analysis of Self-Supervisor, Self-Peer, and Peer-Supervisor Ratings. Personnel Psychology, 41(1), 43–62. DOI: 10.1111/j.1744-6570.1988.tb00631.x.

[6] Mount, M. K., Judge, T. A., Scullen, S. E., Sytsma, M. R. & Hezlett, S. A. (1998). Trait, Rater and Level Effects in 360-Degree Performance Ratings. Personnel Psychology, 51(3), 557–576. DOI: 10.1111/j.1744-6570.1998.tb00251.x.

[7] Batista-Foguet, J. M., Saris, W., Boyatzis, R. E., Serlavós, R. & Velasco Moreno, F. (2019). Multisource Assessment for Development Purposes: Revisiting the Methodology of Data Analysis. Frontiers in Psychology, 9, 2646. DOI: 10.3389/fpsyg.2018.02646.

Subscribe now

Im Dezember 2025 sprach eine Bundesjury in Colorado Rehab Mohamed, einer ehemaligen Mitarbeiterin der Society for Human Resource Management, 11,5 Millionen US-Dollar Schadensersatz zu, nachdem sie zu dem Schluss gekommen war, dass Mohamed rassistisch diskriminiert worden war und Vergeltungsmaßnahmen ausgesetzt war. Davon entfielen 1,5 Millionen US-Dollar auf kompensatorischen Schadensersatz und 10 Millionen US-Dollar auf Strafschadensersatz. Der Fall lautet Mohamed v. Society for Human Resource Management, No. 1:22-cv-01625-GPG-KAS, vor dem United States District Court for the District of Colorado. [1]

Der Name SHRM dürfte in Europa weniger unmittelbar bekannt sein, deshalb lohnt es sich zunächst festzuhalten, von wem hier die Rede ist. Die Society for Human Resource Management hat fast 340.000 Mitglieder in 180 Ländern, zertifiziert HR-Fachkräfte, bietet Schulungen und fachliche Orientierung an und bezeichnet sich selbst als eine der führenden Berufsorganisationen, die die Arbeitswelt mitgestalten. [2] Es handelte sich also kaum um einen Arbeitgeber, für den Diskriminierung am Arbeitsplatz, Vergeltungsmaßnahmen oder ordnungsgemäße HR-Verfahren irgendein abgelegenes juristisches Spezialgebiet gewesen wären.

Die Einzelheiten des Falls lassen wir zunächst beiseite; wer sich dafür interessiert, findet sie in den Quellen. Auf Grundlage der Beweise kam die Jury zu dem Schluss, dass Mohamed rassistisch diskriminiert worden war und wegen ihrer Beschwerden über Diskriminierung Vergeltungsmaßnahmen ausgesetzt war. [1]

Nach dem Urteil beantragte SHRM ein neues Verfahren und focht zugleich den zugesprochenen Strafschadensersatz in Höhe von 10 Millionen US-Dollar an. Im April 2026 wies Bundesrichter Gordon P. Gallagher den Antrag zurück. „Wenn jemand um die Möglichkeit eines hohen Strafschadensersatzes bei Diskriminierung am Arbeitsplatz und böswilligen Vergeltungsmaßnahmen wissen musste, dann war es SHRM …“ Anschließend wies er darauf hin, dass SHRM selbst Unternehmen genau zu diesen Themen schult, einschließlich der damit verbundenen möglichen Haftung. [3]

SHRM legte Berufung ein, die juristische Geschichte ist also noch nicht abgeschlossen. [4] So weit der bisherige Sachstand.

Ein Urteil über 11,5 Millionen US-Dollar zugunsten einer Arbeitnehmerin, der Name SHRM und das Thema Diskriminierung haben zusammen genug Gewicht, damit dieses spezielle Risiko noch lange in den Kaffeeküchen der HR-Abteilungen die Runde machen dürfte.

Dass Diskriminierung am Arbeitsplatz vorkommen kann, dürfte niemanden völlig überraschen. Dieser Fall erzeugte jedoch genug Aufmerksamkeit, um weit über die Führungs- und Rechtsabteilungen von SHRM hinauszureichen. Er könnte praktisch die gesamte Branche erreichen, ebenso aber auch diejenigen auf der anderen Seite: die Beschäftigten. Und diese Gruppe ist praktisch unbegrenzt groß. Als Arbeitnehmerin oder Arbeitnehmer kann fast jeder irgendwann in eine Situation geraten, in der eine Diskriminierungsbeschwerde, eine Leistungsbeurteilung, ein Disziplinarverfahren oder eine Kündigung plötzlich persönlich wird. Die Information könnte sich deshalb weiter wie ein Virus verbreiten, eines, das später in dieser Geschichte noch eine wichtige Nebenrolle spielen wird.

Wenn das Unternehmensrisiko persönlich wird

Beginnen wir mit den HR-Fachleuten, die über Beschwerden, Leistungsprobleme, disziplinarische Maßnahmen oder Kündigungen entscheiden müssen. Natürlich wäre es wünschenswert, wenn ein Urteil wie dieses Diskriminierung und Vergeltungsmaßnahmen abschreckt. Es lohnt sich aber auch, eine mögliche Nebenwirkung zu untersuchen: Wie stark verändert die öffentliche Aufmerksamkeit rund um solche Fälle das Verhalten der HR-Fachkraft, die die nächste Entscheidung treffen muss?

Ein Leistungsproblem bleibt ein Leistungsproblem, ein Konflikt bleibt ein Konflikt und eine Beschwerde bleibt eine Beschwerde. Was sich verändert, ist das Wissen darum, dass ein schlecht gehandhabter Fall zu jahrelangen Gerichtsverfahren, einem Urteil über Millionenbeträge, öffentlichen Zeugenaussagen, einem Reputationsschaden und sehr unangenehmen beruflichen Konsequenzen führen kann. Dass all das ausgerechnet SHRM passiert ist, also einer Organisation, die andere genau im Umgang mit solchen Situationen schult und für die Nichtdiskriminierung zum beruflichen Alltag gehört, zeigt vielleicht auch, wie nachlässig wir sein können. Es lohnt sich deshalb zu fragen, warum wir manchmal glauben, dass a) das, was wir tun, gar keine Diskriminierung ist, b) es selbst dann niemand bemerken wird oder c) wir damit einfach durchkommen.

Sobald die Wahrnehmung des eigenen persönlichen Risikos Teil der Entscheidung wird, kann defensives Entscheidungsverhalten entstehen: An die Stelle der Option, die für die Organisation als besser angesehen wird, kann die Option treten, die für die entscheidende Person selbst sicherer erscheint. [5] Artinger und Kollegen befragten 950 Führungskräfte im öffentlichen Dienst zu den zehn wichtigsten Entscheidungen, die sie in den vergangenen zwölf Monaten getroffen hatten. Achtzig Prozent gaben an, mindestens eine solche Entscheidung getroffen zu haben. Im Durchschnitt waren 2,5 ihrer zehn wichtigsten Entscheidungen dieser Art. Das Phänomen zeigte sich auf allen untersuchten Hierarchieebenen. [5]

Die Wahrscheinlichkeit, dass Menschen die für sie persönlich sicherere Option wählen, ist keineswegs gering. In einem Experiment aus dem Jahr 2025 mit 315 Führungskräften wurde untersucht, unter welchen Bedingungen dieses Verhalten wahrscheinlicher wird. Geringe psychologische Sicherheit zusammen mit wenig authentischer Führung erhöhte die Wahrscheinlichkeit, dass sich die Teilnehmenden für die persönlich sicherere Option entschieden. [6]

Das bedeutet nicht, dass das SHRM-Urteil nun dazu führen wird, dass HR-Fachleute defensiver entscheiden. Dafür gibt es keine Belege. Aber die Frage verdient inzwischen Beachtung: In welchem Maß wird ein so öffentlich sichtbarer Fall Teil des Risikoumfelds, in dem die nächste HR-Entscheidung getroffen wird?

Und falls das passiert, müssen wir uns anschauen, wie sich das im Verhalten zeigen könnte. Wo endet berechtigte Vorsicht, und wo beginnt Selbstschutz?

Das andere Extrem

Wenn das eigene Risikogefühl der entscheidenden Person bereits Teil der nächsten HR-Entscheidung geworden ist, lohnt es sich, bei zwei Arten von Verhalten nach dem „anderen Extrem“ zu suchen.

Die erste besteht darin, immer mehr Schutzschichten um die Entscheidung herum aufzubauen: eine weitere Freigabestufe, die frühere Einschaltung von Juristen, mehr Dokumentation, noch eine Meinung, ein Gremium oder ein externer Berater. Der zusätzliche Nutzen dürfte vermutlich vor allem noch mehr Bürokratie erzeugen, obwohl davon ohnehin schon genug auf dem Tisch liegt.

Die zweite Richtung ist, die Entscheidung selbst zu vermeiden oder hinauszuzögern: Ein Leistungsproblem zieht sich hin, in einem schwierigen Fall will niemand als Erste oder Erster klar Stellung beziehen, oder eine Option rückt in den Vordergrund, weil die Person, die entscheidet, später persönlich weniger Risiko trägt. Für Beschäftigte, die ohnehin schon keinen ganz sicheren Boden unter den Füßen haben, kann das lähmend werden. Ganz zu schweigen von den geradezu wunderbaren – für den Arbeitgeber allerdings kostspieligen – Folgen der psychischen Belastung, die daraus entstehen kann.

Eine rechtliche Prüfung, Dokumentation oder eine zweite Meinung haben ihren Platz, wenn sie begründet sind. Entscheidend ist, ob diese zusätzlichen Schritte hauptsächlich dazu dienen, das persönliche Risiko der entscheidenden Person zu verringern, während sie immer weniger zur Qualität der Entscheidung beitragen. Dasselbe gilt für das Aufschieben einer Entscheidung. Dafür kann es fachliche Gründe geben. Es kann aber auch einfach dazu dienen, das Risiko nicht bei der Person zu belassen, die entscheiden muss.

Für dieses zweite Muster gibt es direktere Forschungsergebnisse. Eine Studie aus dem Jahr 2021 zu defensivem Entscheidungsverhalten brachte geringe psychologische Sicherheit und Arbeitsplatzunsicherheit mit einer wahrgenommenen Bedrohung persönlicher Ressourcen und einer höheren Wahrscheinlichkeit von Entscheidungsvermeidung in Verbindung. Die Forschenden untersuchten dabei getrennt, ob Personen eine Entscheidung aktiv angingen oder sie vermieden. [7]

Eine Verzögerung beweist für sich genommen natürlich noch nichts. Ein Experiment aus dem Jahr 2020 zeigte, dass es vernünftig und sinnvoll sein kann, eine Entscheidung aufzuschieben, etwa wenn weitere Informationen benötigt werden. Eine Verzögerung kann jedoch auch zur Entscheidungsvermeidung werden, wenn sie durch die Schwierigkeit der Entscheidung oder durch Unentschlossenheit aufrechterhalten wird. [8]

All das betrachten wir zudem in einem Berufsfeld, das ohnehin unter erheblichem Druck steht. Der HR Mental Wellbeing Report 2026 fasste Daten von fast 3.000 britischen HR-Fachleuten über einen Zeitraum von drei Jahren zusammen. In den jüngsten Ergebnissen erreichten 40 Prozent den Schwellenwert für klinisch relevante Angstsymptome, 44 Prozent für klinisch relevante Symptome einer Depression, und bei 62 Prozent wurde eine sehr hohe Wahrscheinlichkeit für Burnout festgestellt. Nur 13 Prozent waren der Ansicht, dass ihr eigenes psychisches Wohlbefinden am Arbeitsplatz ausreichend unterstützt wurde. [9]

Es handelt sich um eine Branchenumfrage. Sie liefert keinen klinischen oder kausalen Beleg dafür, dass HR-Fachleute deshalb schlechtere Entscheidungen treffen. Die Ergebnisse reichen jedoch aus, um Entscheidungsträger nicht als abstrakte, emotional völlig unberührte Akteure zu betrachten.

Auch die Wirtschaftsprüfung zeigt, was in einem beruflichen Umfeld passieren kann, in dem Entscheidungen dokumentiert werden müssen, später von außen überprüft werden können und Fehler erhebliche rechtliche Folgen haben können. In einem Experiment mit 118 Wirtschaftsprüfern verließen sich die Teilnehmenden stärker auf die Empfehlung einer Entscheidungshilfe, wenn das Risiko rechtlicher Auseinandersetzungen höher war. Wenn sowohl dieses Risiko als auch das Kontrollrisiko hoch waren, spielte es eine größere Rolle, wie gut sich die Entscheidung rechtlich verteidigen ließ. Gleichzeitig verließen sich die Wirtschaftsprüfer stärker auf die Entscheidungshilfe, während ihr Vertrauen in die Qualität des eigenen Urteils sank. [10]

In einer weiteren Studie änderten Wirtschaftsprüfer ihr Verhalten, nachdem eine der Banken in ihrem Kundenportfolio zusammengebrochen war. Gegenüber ihren übrigen Bankkunden wurden sie vorsichtiger. Die höheren Rückstellungen wurden jedoch nicht rechtzeitiger gebildet, erwiesen sich als weniger genau, wurden später wieder aufgelöst und schränkten außerdem die Kreditvergabe der Banken ein. [11] Ein schwerwiegendes negatives Ereignis kann also tatsächlich zu größerer beruflicher Vorsicht führen, ohne dass sich die Qualität der Entscheidungen entsprechend verbessert. Die betrieblichen Kosten entstehen dann an anderer Stelle.

Auch daraus folgt nicht, dass HR-Fachleute sich nach dem SHRM-Fall genauso verhalten werden. Wir haben inzwischen jedoch mehrere Teile des Mechanismus. Wir wissen, dass Bedrohungen persönlicher Ressourcen mit defensivem Entscheidungsverhalten verbunden sein können [7], dass das Aufschieben einer Entscheidung sowohl eine begründete Suche nach weiteren Informationen als auch Entscheidungsvermeidung bedeuten kann [8], dass eine dreijährige Umfrage unter fast 3.000 britischen HR-Fachleuten erhebliche psychische Belastungen festgestellt hat [9] und dass es in einem anderen stark regulierten Berufsfeld Hinweise darauf gibt, dass ein höheres rechtliches Risiko oder ein früheres schwerwiegendes negatives Ereignis das professionelle Urteil tatsächlich verändern kann. [10][11]

Zusätzliche Kontrollen, mehr Dokumentation oder das Aufschieben einer Entscheidung sagen für sich genommen deshalb sehr wenig aus. Wir müssten wissen, was sie zur Qualität der Entscheidung beigetragen haben, welches Risiko sie reduziert haben, wessen Risiko sie reduziert haben und welche Kosten dadurch bei der Organisation geblieben sind.

Ein Geflecht von Konflikten

Hier wird die Spannung zwischen den unterschiedlichen Interessen konkret. In derselben Situation hat der Beschäftigte ein Interesse daran, rechtmäßig und fair behandelt zu werden. Die Organisation wiederum hat ein Interesse daran, das betriebliche Problem zu lösen und sicherzustellen, dass die Arbeit dem Standard entspricht, auf den sich beide Seiten vertraglich geeinigt haben.

Und dann ist da noch die HR-Fachkraft. Ihr grundlegendes Interesse besteht darin, ihre Arbeit machen zu können – und sie auch machen zu dürfen. Dafür braucht sie Fachwissen, ausreichende Informationen und echten Entscheidungsspielraum. Gleichzeitig muss sie genau wissen, wer am Ende die Rechnung bezahlt und welchem organisatorischen Interesse sie in ihrer Rolle dienen soll. Sie sollte nicht zu moralischen Kompromissen gezwungen werden. Gleichzeitig sollte eine fachlich vertretbare Entscheidung sie nicht unverhältnismäßigen rechtlichen, beruflichen oder reputationsbezogenen Risiken aussetzen.

Im besten Fall passen Theorie und Praxis zusammen wie verschiedene technische Pläne für dasselbe Gebäude. Der Architekturplan ist für sich vollständig und zeigt Wände, Räume und Dach. Die Pläne für Tragwerk, Haustechnik und Elektrik beschreiben jeweils ihr eigenes System. Jeder steht für sich, und wenn man sie übereinanderlegt, sollten sie alle dasselbe Gebäude zeigen.

Im weniger günstigen Fall kann, wie auch der SHRM-Fall zeigt, das kleine Drama namens „Versäumnis“ die bestehenden Regeln und Vorgaben bis an ihre Grenzen belasten – mit einer erheblichen Zahlung als möglicher Folge.

Notwendigkeit, Verhältnismäßigkeit, Legitimität

Wenn diese drei Interessen tatsächlich im selben Entscheidungsraum zusammentreffen, muss etwas verhindern, dass eines davon die beiden anderen einfach verdrängt. Drei relativ einfache Kriterien bieten sich dafür an: Notwendigkeit, Verhältnismäßigkeit und Legitimität.

Bei der Notwendigkeit geht es zunächst darum, genau benennen zu können, welches Problem HR eigentlich lösen will. Ein Leistungsgespräch, eine Untersuchung, eine Verwarnung, eine Suspendierung oder eine Kündigung sollte einen klar erkennbaren Grund und Zweck haben. Wenn das unscharf wird, kann der Prozess anfangen, ein Eigenleben zu entwickeln, und aus der Risikotabelle wird nach und nach der Arbeitsplan: noch eine Untersuchung, noch eine Freigabe, noch eine Runde Dokumentation, weil der nächste Schritt auf diese Weise etwas sicherer wirkt. Dann lohnt sich die Frage, ob diese zusätzliche Ebene tatsächlich dazu beiträgt, den Fall besser zu verstehen oder die Qualität der Entscheidung zu verbessern.

Bei der Verhältnismäßigkeit kommt eine weitere Frage hinzu: Wie passt das Ausmaß der Reaktion zu dem, was wir tatsächlich wissen? Eine Beschwerde kann ernst sein, macht einen Beschäftigten für sich genommen aber nicht unantastbar. Ein Leistungsproblem kann real sein, rechtfertigt für sich genommen aber nicht jede beliebige Reaktion des Arbeitgebers. Die vorhandenen Belege, die Schwere des Problems, die Frage, ob es wiederholt auftritt, das Maß an Unsicherheit und die wahrscheinlichen Folgen bilden zusammen den Rahmen, innerhalb dessen beurteilt werden kann, wie weit die Reaktion von HR gehen sollte.

Legitimität ist eine persönlichere Frage. Was berechtigt gerade diese HR-Fachkraft dazu, in diesem Fall fachlich zu urteilen? Hat sie die notwendigen Informationen, das erforderliche Fachwissen und genügend Entscheidungsspielraum? Weiß sie, wo ihre eigenen Befugnisse enden und wann eine andere Form von Fachwissen hinzugezogen werden muss? Mit diesen Fragen habe ich mich ausführlicher in Who judges the judges? beschäftigt. Sie spielen auch hier direkt eine Rolle, denn die Person, die entscheidet, ist sich zugleich ihres eigenen Risikos bewusst.

Auch die spätere Überprüfbarkeit gehört dazu. Eine kompetente Person, die die Entscheidung später überprüft, sollte nachvollziehen können, welche Informationen damals vorlagen, welche Alternativen geprüft wurden und auf welcher Grundlage die endgültige Entscheidung getroffen wurde. Dokumentation ist nur ein Mittel dafür. Entscheidend ist, dass der Weg zur fachlichen Beurteilung später noch nachvollziehbar bleibt.

Wenn auch KI an der Entscheidung beteiligt war, ist das immer weniger nur eine Frage dessen, was Fachleute selbst für sinnvoll halten. Für bestimmte Hochrisiko-Systeme verlangt der EU AI Act Nachvollziehbarkeit, Protokollierung, menschliche Aufsicht und die Möglichkeit, der betroffenen Person eine klare und aussagekräftige Erklärung zu geben. [12] Auch der Zeitpunkt ist hier wichtig: Die Pflichten aus Kapitel III für Hochrisiko-Systeme nach Anhang III, die im Beschäftigungsbereich eingesetzt werden, gelten ab dem 2. Dezember 2027. Artikel 86 gehörte nicht zu den Vorschriften, deren Anwendung durch die Verordnung (EU) 2026/1744 verschoben wurde, und bleibt deshalb dem allgemeinen Zeitplan des AI Act unterworfen. [12]

Der Gerichtshof der Europäischen Union hat 2025 außerdem die Regeln der DSGVO zu automatisierten Entscheidungen weiter präzisiert: Die betroffene Person muss eine verständliche Erklärung darüber erhalten, nach welchem Verfahren und nach welchen Grundsätzen das sie betreffende automatisierte Ergebnis zustande gekommen ist. Dafür muss nicht der gesamte Algorithmus offengelegt werden. Die „Black Box“ kann aber auch nicht einfach mit der Erklärung abgetan werden, das System habe eben dieses Ergebnis geliefert. [13] Das gibt den Beteiligten einen recht dringenden Grund, den Entscheidungsweg nachvollziehbar zu machen, lange bevor später jemand diese Black Box öffnen muss.

Diese drei Kriterien können auch dabei helfen zu erkennen, wann der Selbstschutz von HR zu viel Raum in der Entscheidung einnimmt. Wenn ein Schritt unnötig oder unverhältnismäßig ist oder in den Händen einer Person liegt, der die notwendigen Informationen, die Kompetenz oder die Befugnis fehlen, geht es nicht mehr nur darum, ob das Verfahren ordnungsgemäß abläuft. Dann müssen wir auch fragen, wessen Risiko dadurch reduziert wird – und zu welchem Preis.

Nebenbemerkung: Wer hält sich selbst eigentlich für jemanden, der diskriminiert?

Unser Bild von der eigenen Unvoreingenommenheit ist kein besonders zuverlässiger Maßstab, wenn wir uns selbst überprüfen wollen. Das Phänomen des sogenannten Bias Blind Spot beschreibt unsere Tendenz, kognitive Verzerrungen bei anderen Menschen leichter zu erkennen als bei uns selbst. [14] Eine mögliche Erklärung dafür ist, dass wir uns selbst zum Teil nach unseren Absichten beurteilen, andere Menschen dagegen nach dem, was wir an ihrem Verhalten tatsächlich beobachten können. [15]

HR bildet da keine Ausnahme. In einer Studie mit 234 Schweizer HR-Fachleuten wurden die Teilnehmenden zu sieben Verzerrungen befragt, die bei der Personalauswahl auftreten können. Sie schätzten ihre durchschnittlichen HR-Kolleginnen und -Kollegen durchgehend als anfälliger für diese Verzerrungen ein als sich selbst. [16] Das bedeutet nicht, dass sie diskriminiert haben. Es zeigt aber, dass selbst Menschen, die solche Einflüsse in ihrer beruflichen Arbeit regelmäßig berücksichtigen müssen, ihre eigene Anfälligkeit für Verzerrungen nicht unbedingt richtig einschätzen.

Das Selbstbild, „jemand zu sein, der nicht diskriminiert“, reicht deshalb allein nicht aus, um festzustellen, ob eine konkrete Entscheidung tatsächlich frei von Verzerrungen war. Bei dieser Selbsteinschätzung ist besondere Vorsicht angebracht: Wir kennen unsere eigenen Absichten. Andere sehen dagegen, was wir tatsächlich tun und welche Folgen unsere Entscheidungen haben.

Die Möglichkeiten, wie Personalauswahl und menschliches Urteilen schieflaufen können, sind erstaunlich vielfältig. Sie hier alle aufzuschlüsseln, würde den Rahmen dieses Artikels deutlich sprengen. Wenn Sie damit zu tun haben, egal auf welcher Seite – als Bewerberin oder Bewerber, Führungskraft oder HR-Fachkraft –, gehe ich in meinem Buch Judging Strangers: The Readability Trap in Hiring deutlich ausführlicher auf diese Frage ein.

Quellen

[1] Mohamed v. Society for Human Resource Management, No. 1:22-cv-01625-GPG-KAS, U.S. District Court for the District of Colorado; jury verdict, 5 December 2025.

[2] Society for Human Resource Management. About SHRM; SHRM HR Certification; Educational Programs. SHRM.org.

[3] Mohamed v. Society for Human Resource Management, No. 1:22-cv-01625-GPG-KAS, post-trial order, U.S. District Court for the District of Colorado, 15 April 2026.

[4] Simpson, A. G. (2026). Insurer Says Virginia ‘Intentional’ Acts Law Blocks $10M Punitive Damages Claim. Insurance Journal, 31 July 2026.

[5] Artinger, F. M., Artinger, S., & Gigerenzer, G. (2019). C. Y. A.: Frequency and causes of defensive decisions in public administration. Business Research, 12, 9–25. https://doi.org/10.1007/s40685-018-0074-2

[6] Artinger, F. M., Marx-Fleck, S., Junker, N. M., Gigerenzer, G., Artinger, S., & van Dick, R. (2025). Coping with uncertainty: The interaction of psychological safety and authentic leadership in their effects on defensive decision making. Journal of Business Research, 190, 115240. https://doi.org/10.1016/j.jbusres.2025.115240

[7] Marx-Fleck, S., Junker, N. M., Artinger, F., & van Dick, R. (2021). Defensive decision making: Operationalization and the relevance of psychological safety and job insecurity from a conservation of resources perspective. Journal of Occupational and Organizational Psychology, 94(3), 616–644. https://doi.org/10.1111/joop.12353

[8] Berens, S., & Funke, J. (2020). A vignette study of option refusal and decision deferral as two forms of decision avoidance: Situational and personal predictors. PLOS ONE, 15(10), e0241182. https://doi.org/10.1371/journal.pone.0241182

[9] Baker, F., & Burrell, J. (2026). HR Mental Wellbeing Report 2026: A three-year analysis of HR mental wellbeing. Ultimate Resilience. Published in partnership with Everywhen.

[10] Gomaa, M. I., Hunton, J. E., & Rose, J. M. (2008). The effects of control risk and litigation risk on decision aid reliance. International Journal of Applied Decision Sciences, 1(1), 80–106. https://doi.org/10.1504/IJADS.2008.017953

[11] Hall, C., Judd, J. S., & Sunder, J. (2023). Auditor conservatism, audit quality, and real consequences for clients. Review of Accounting Studies, 28(2), 689–725. https://doi.org/10.1007/s11142-021-09653-1

[12] Regulation (EU) 2024/1689 (Artificial Intelligence Act), in particular Articles 12, 14, 26, 86 and 113, and Annex III; as amended by Regulation (EU) 2026/1744.

[13] Court of Justice of the European Union, CK v Magistrat der Stadt Wien (Dun & Bradstreet Austria), Case C-203/22, judgment of 27 February 2025, ECLI:EU:C:2025:117; in particular the interpretation of Article 15(1)(h) GDPR.

[14] Pronin, E., Lin, D. Y., & Ross, L. (2002). The bias blind spot: Perceptions of bias in self versus others. Personality and Social Psychology Bulletin, 28(3), 369–381. https://doi.org/10.1177/0146167202286008

[15] Pronin, E., & Kugler, M. B. (2007). Valuing thoughts, ignoring behavior: The introspection illusion as a source of the bias blind spot. Journal of Experimental Social Psychology, 43(4), 565–578. https://doi.org/10.1016/j.jesp.2006.05.011

[16] Thomas, O., & Reimann, O. (2023). The bias blind spot among HR employees in hiring decisions. German Journal of Human Resource Management, 37(1), 5–22. https://doi.org/10.1177/23970022221094523

Lilien Gerlach
Verhaltensanalytikerin & Autorin
liliengerlach.com

In December 2025, a federal jury in Colorado awarded $11.5 million in damages to Rehab Mohamed, a former employee of the Society for Human Resource Management, after finding that she had been subjected to racial discrimination and retaliation. Of that amount, $1.5 million was compensatory damages and $10 million punitive damages. The case is Mohamed v. Society for Human Resource Management, No. 1:22-cv-01625-GPG-KAS, before the United States District Court for the District of Colorado. [1]

Subscribe now

The SHRM name may be less immediately familiar in Europe, so it is worth establishing who we are talking about. The Society for Human Resource Management has nearly 340,000 members across 180 countries, certifies HR professionals, provides training and professional guidance, and describes itself as one of the leading professional organisations shaping the world of work. [2] In other words, this was hardly an employer for whom employment discrimination, retaliation or proper HR procedure were some remote legal speciality.

Let us leave the details of the case aside for now; anyone interested can find them in the sources. On the evidence, the jury concluded that Mohamed had been discriminated against on the basis of race and subjected to retaliation because of her complaints about discrimination. [1]

After the verdict, SHRM sought a new trial and also challenged the $10 million punitive damages award. In April 2026, federal judge Gordon P. Gallagher denied the request. “If anyone knew the possibility of high potential punitive damages for employment discrimination and malicious retaliation, it was SHRM…” He went on to point out that SHRM itself trains companies on precisely these issues, including the potential liability attached to them. [3]

SHRM appealed, so the legal story is not yet over. [4] That is the factual thread so far.

An $11.5 million award in favour of an employee, the SHRM name and the concept of discrimination together create a certain density: enough to keep this particular risk circulating in HR coffee-room whispers for quite some time.

Nobody is exactly blindsided by the revelation that discrimination can happen at work. This case, however, ended up making enough noise to travel far beyond SHRM’s executive and legal offices. It may reach virtually the entire industry, but also those standing on the other side: employees. And that population is, for all practical purposes, limitless. As an employee, almost anyone may one day find themselves in a situation where a discrimination complaint, a performance review, a disciplinary process or a dismissal suddenly becomes personal. And so the information may continue to spread like a virus – one that will have an important supporting role later in this story.

When Corporate Risk Becomes Personal

Let us start with the HR professionals who may have to make decisions about complaints, performance problems, disciplinary matters or dismissals. And while we would naturally want a judgment like this to deter discrimination and retaliation, there is another possible side effect worth examining: to what extent does the visibility of cases like this change the behaviour of the HR professional making the next decision?

A performance problem is still a performance problem, a conflict is still a conflict, and a complaint is still a complaint. What changes is the knowledge that a badly handled case can turn into years of litigation, a multimillion-dollar judgment, public testimony, reputational damage and some deeply uncomfortable professional exposure. The fact that all of this happened to SHRM – an organisation that trains others to handle precisely these situations and for which non-discrimination is part of the everyday professional vocabulary – may show just how careless we can be. It is also worth examining why we might think that a) what we are doing is not discrimination at all, b) even if it is, nobody will find out, or c) we can simply get away with it.

Once the perception of personal risk enters the decision, defensive decision-making may follow: instead of choosing the option considered better for the organisation, the personally safer one may come to the fore [5]. Artinger and colleagues asked 950 public-sector managers about the ten most important decisions they had made during the previous twelve months. Eighty per cent reported making at least one such decision, and on average 2.5 of their ten most important decisions fell into this category. The phenomenon appeared at every hierarchical level examined. [5]

And the chances that people will choose the option that feels safer for them personally are not exactly negligible. A 2025 experiment involving 315 managers examined the conditions under which this behaviour becomes more likely. Low psychological safety combined with low levels of authentic leadership increased the likelihood that participants would choose the personally safer option [6].

None of this means that the SHRM judgment will now cause HR professionals to start making defensive decisions. We have no evidence of that. But I think the question now earns itself a place at the table: to what extent does a highly visible case like this become part of the risk environment in which the next HR decision is made?

And if it does, we need to look at how that might become visible in behaviour. Where does justified caution end, and where does self-protection begin?

The Other Extreme

If the decision-maker’s own sense of risk has already become part of the next HR decision, then the ‘other extreme’ is worth looking for in two kinds of behaviour. The first is when more and more protective layers are built around the decision: another level of approval, earlier involvement of lawyers, more documentation, another opinion, a committee or an external adviser. The added value will, presumably, add still more to bureaucracy’s table, already groaning under the weight.

The other direction is avoidance or delay of the decision itself: a performance problem drags on, nobody wants to be the first to take a position in a contentious case, or an option comes to the fore because it will leave the decision-maker less personally exposed later. For an employee who is already not entirely sure-footed, this can become paralysing. Not to mention the wonderfully beneficial – though costly for the employer – effects of the resulting psychological strain.

Legal scrutiny, documentation or a second opinion all have their place when justified. The question to ask is whether the primary function of those additional steps has become reducing the decision-maker’s own exposure while contributing progressively less to the quality of the decision. The same applies in the other direction: postponing a decision may be professionally justified, or it may simply serve to ensure that the risk does not remain with the person who has to decide.

This second pattern has more direct research support. A 2021 study of defensive decision-making linked lower psychological safety and job insecurity to perceived threats to personal resources, alongside an increased likelihood of decision avoidance. The researchers measured decision approach and decision avoidance separately. [7] Delay, of course, proves nothing on its own. A 2020 experiment showed that deferring a decision can be reasonable and adaptive, for example when further information is needed, but it can also take the form of decision avoidance when it is sustained by the difficulty of the choice or by indecision itself. [8]

And we are hardly considering all of this in a profession that is free of pressure to begin with. The 2026 HR Mental Wellbeing Report brought together data from nearly 3,000 British HR professionals over three years. In the latest findings, 40 per cent met the threshold for clinically significant anxiety, 44 per cent for clinically significant symptoms of depression, and 62 per cent were assessed as very likely to be experiencing burnout. Only 13 per cent felt that their own mental wellbeing was adequately supported at work. [9] This is an industry survey, not clinical or causal evidence that HR professionals make worse decisions as a result. It is enough, however, to stop us imagining the decision-maker as an abstract, emotionally sterile actor.

Auditing also gives us a view of what can happen in a professional environment where decisions have to be documented, may later be scrutinised from the outside, and mistakes can carry serious legal consequences. In an experiment involving 118 auditors, participants relied more heavily on the recommendation of a decision aid when litigation risk was higher. When both litigation risk and internal control risk were high, legal defensibility carried more weight, reliance on the decision aid increased, while auditors’ confidence in the quality of their own judgement declined. [10]

In another study, auditors changed their behaviour after one of the banks in their client portfolio failed. They became more conservative with their remaining banking clients. Yet the larger provisions were no more timely, proved less accurate, were subsequently reversed, and also constrained the banks’ ability to lend. [11] In other words, a serious negative event can indeed be followed by greater professional caution without any corresponding improvement in decision quality, while the operational cost lands elsewhere.

None of this means that HR professionals will behave in the same way after the SHRM case. But we now have several pieces of the mechanism. We know that threats to personal resources can be associated with defensive decision-making [7]; that decision deferral can reflect either justified information gathering or avoidance [8]; that a three-year survey covering nearly 3,000 British HR professionals found substantial psychological strain [9]; and that in another highly regulated profession there is evidence that increased legal risk, or an earlier serious adverse event, can actually shift professional judgement [10][11].

From here, additional controls, more documentation or delaying a decision tell us very little on their own. We would need to know what they added to the quality of the decision, what risk they reduced, whose risk they reduced, and what cost they left with the organisation.

Cobweb of Conflict

This is where the point of tension between the different interests becomes tangible. In the same situation, there is the employee’s interest in being treated lawfully and fairly, and the organisation’s interest in resolving the operational problem and ensuring that the work is delivered to the standard both sides contracted for. Then there is the HR professional. Their primary interest is to be capable of doing their job – and to be allowed to do it. That requires professional knowledge, adequate information and genuine room for judgement; at the same time, they need to know exactly who pays the ferryman and which organisational interest they were hired to serve. On the one hand, they should not be forced into moral compromise; on the other, a professionally defensible decision should not expose them personally to disproportionate legal, professional or reputational risk.

In the optimal case, theory and practice fit together like the separate technical plans for the same building. The architectural plan is complete in its own right, with its walls, spaces and roof; the structural, mechanical and electrical plans each describe their own system in the same way. Each stands on its own, yet when overlaid they should all reveal the same building.

In the less optimal case, as the SHRM case also demonstrates, the little drama entitled Omission can strain the regulated framework beyond its limits – with a substantial payout as one possible consequence.

Necessity, Proportionality, Legitimacy

If these three interests really do occupy the same decision space, something has to prevent any one of them from simply overwhelming the other two. Three fairly simple checks suggest themselves: necessity, proportionality and legitimacy.

Necessity begins with being able to state precisely what problem HR is trying to address. A performance discussion, investigation, warning, suspension or dismissal should have an identifiable reason and purpose. If that becomes blurred, the process can begin to take on a life of its own, and the spreadsheet of risks may gradually turn into the work plan: another investigation, another approval, another round of documentation, because the next step looks a little safer that way. At that point, it is worth asking whether the additional layer actually contributes anything to understanding the case or improving the quality of the decision.

Proportionality adds the question of how the scale of the response relates to what we actually know. A complaint may be serious, but on its own it does not make an employee untouchable. A performance problem may be real, but on its own it does not justify any and every employer response. The available evidence, the seriousness of the problem, whether it is repeated, the degree of uncertainty and the likely consequences together provide the framework within which the scale of HR’s response can be assessed.

Legitimacy is a more personal question. What entitles this particular HR professional to exercise professional judgement in this case? Do they have the information, professional knowledge and decision-making latitude required? Do they know where their own authority ends and where another form of expertise needs to be brought in? I explored these questions in more detail in Who judges the judges? they carry directly into the present case, where the decision-maker is also conscious of their own exposure.

Reviewability is part of this as well. A later, competent reviewer should be able to reconstruct what information was available, what alternatives were considered and on what basis the final decision was made. Documentation is only one means of doing that. The point is that the path of professional judgement should remain traceable afterwards. If AI has also played a role in the decision, this requirement is becoming less and less a matter of professional preference alone. For certain high-risk systems, the EU AI Act requires traceability, logging, human oversight and the possibility of providing the person affected with a clear and meaningful explanation [12]. The timing matters here: the Chapter III obligations applying to Annex III high-risk systems used in employment apply from 2 December 2027. Article 86 itself was not among the provisions postponed by Regulation (EU) 2026/1744 and remains subject to the AI Act’s general application timetable [12].

The Court of Justice of the European Union further clarified the GDPR rules on automated decision-making in 2025: the person affected must be given an intelligible explanation of the procedure and principles by which the automated result concerning them was produced. The entire algorithm does not have to be laid bare on the table, but neither can the ‘black box’ simply be closed with the explanation that this is what the system produced [13]. That may rather urgently give the relevant parties a reason to make the decision path reconstructable long before someone has to open the box afterwards.

These three considerations can also help show when HR’s own self-protection begins to occupy too much space in the decision. If a step is unnecessary, disproportionate, or placed in the hands of someone who lacks the information, competence or authority required, the question is no longer simply whether the process is procedurally sound. We also need to ask whose risk it is reducing – and at what cost.

Sidenote: Who Actually Thinks of Themselves as Someone Who Discriminates?

The picture we have of our own impartiality is not a particularly reliable way of checking ourselves. The phenomenon known as the bias blind spot describes our tendency to recognise cognitive biases more readily in other people than in ourselves [14]. One possible explanation is that we judge ourselves partly by our intentions, while judging others by what we can observe in their behaviour [15].

HR is no exception. In a study of 234 Swiss HR professionals, participants were asked about seven biases that can arise in employee selection. They consistently judged the average HR colleague to be more susceptible to those biases than they were themselves [16]. That does not mean they discriminated. It does show that even people who routinely have to account for such influences in their professional work are not necessarily accurate judges of their own susceptibility to bias.

The self-image of being ‘someone who does not discriminate’ is therefore not enough, on its own, to establish whether a particular decision was genuinely free from bias. It is worth being especially cautious about that self-assessment when we know our own intentions, while others will judge the consequences of our decisions from the behaviour they can actually observe.

The range of ways in which recruitment and human judgement can go astray is impressively broad, and unpacking them here would stretch this article well beyond its limits. If you are involved on any side – as a candidate, a manager or an HR professional – I go much further into this question in my book, Judging Strangers: The Readability Trap in Hiring.

Sources

[1] Mohamed v. Society for Human Resource Management, No. 1:22-cv-01625-GPG-KAS, U.S. District Court for the District of Colorado; jury verdict, 5 December 2025.

[2] Society for Human Resource Management. About SHRM; SHRM HR Certification; Educational Programs. SHRM.org.

[3] Mohamed v. Society for Human Resource Management, No. 1:22-cv-01625-GPG-KAS, post-trial order, U.S. District Court for the District of Colorado, 15 April 2026.

[4] Simpson, A. G. (2026). Insurer Says Virginia ‘Intentional’ Acts Law Blocks $10M Punitive Damages Claim. Insurance Journal, 31 July 2026.

[5] Artinger, F. M., Artinger, S., & Gigerenzer, G. (2019). C. Y. A.: Frequency and causes of defensive decisions in public administration. Business Research, 12, 9–25. https://doi.org/10.1007/s40685-018-0074-2

[6] Artinger, F. M., Marx-Fleck, S., Junker, N. M., Gigerenzer, G., Artinger, S., & van Dick, R. (2025). Coping with uncertainty: The interaction of psychological safety and authentic leadership in their effects on defensive decision making. Journal of Business Research, 190, 115240. https://doi.org/10.1016/j.jbusres.2025.115240

[7] Marx-Fleck, S., Junker, N. M., Artinger, F., & van Dick, R. (2021). Defensive decision making: Operationalization and the relevance of psychological safety and job insecurity from a conservation of resources perspective. Journal of Occupational and Organizational Psychology, 94(3), 616–644. https://doi.org/10.1111/joop.12353

[8] Berens, S., & Funke, J. (2020). A vignette study of option refusal and decision deferral as two forms of decision avoidance: Situational and personal predictors. PLOS ONE, 15(10), e0241182. https://doi.org/10.1371/journal.pone.0241182

[9] Baker, F., & Burrell, J. (2026). HR Mental Wellbeing Report 2026: A three-year analysis of HR mental wellbeing. Ultimate Resilience. Published in partnership with Everywhen.

[10] Gomaa, M. I., Hunton, J. E., & Rose, J. M. (2008). The effects of control risk and litigation risk on decision aid reliance. International Journal of Applied Decision Sciences, 1(1), 80–106. https://doi.org/10.1504/IJADS.2008.017953

[11] Hall, C., Judd, J. S., & Sunder, J. (2023). Auditor conservatism, audit quality, and real consequences for clients. Review of Accounting Studies, 28(2), 689–725. https://doi.org/10.1007/s11142-021-09653-1

[12] Regulation (EU) 2024/1689 (Artificial Intelligence Act), in particular Articles 12, 14, 26, 86 and 113, and Annex III; as amended by Regulation (EU) 2026/1744.

[13] Court of Justice of the European Union, CK v Magistrat der Stadt Wien (Dun & Bradstreet Austria), Case C-203/22, judgment of 27 February 2025, ECLI:EU:C:2025:117; in particular the interpretation of Article 15(1)(h) GDPR.

[14] Pronin, E., Lin, D. Y., & Ross, L. (2002). The bias blind spot: Perceptions of bias in self versus others. Personality and Social Psychology Bulletin, 28(3), 369–381. https://doi.org/10.1177/0146167202286008

[15] Pronin, E., & Kugler, M. B. (2007). Valuing thoughts, ignoring behavior: The introspection illusion as a source of the bias blind spot. Journal of Experimental Social Psychology, 43(4), 565–578. https://doi.org/10.1016/j.jesp.2006.05.011

[16] Thomas, O., & Reimann, O. (2023). The bias blind spot among HR employees in hiring decisions. German Journal of Human Resource Management, 37(1), 5–22. https://doi.org/10.1177/23970022221094523

Lilien Gerlach
Behavioural Analyst & Author
liliengerlach.com

Leave a comment

Judging Hr

243KB ∙ PDF file

Download

Download

Over my morning coffee, I came across a post from an organisation that publishes on workplace issues and ethical AI: “If even Google doesn’t trust AI with hiring, why should we?”

The headline quickly digests the story and serves it back to the reader, with all the professional precision and intellectual humility one might expect from the subject. Or perhaps not.

The post points to an article in HR Portal, an HR publication, whose headline already knows the answer: “Google developers warn: AI wrongly filters out applicants” (HR Portal, 2026).

There is a real story behind it. Google DeepMind’s AGI Safety and Alignment team provided applicants with an additional form because, according to an internal document, there was a non-negligible chance that a CV could be incorrectly filtered out by the application system, or take too long to reach the team. The form was intended to ensure that an actual member of the team saw the application.

The original Bloomberg report also includes a response from a Google DeepMind spokesperson. The spokesperson disputed that the system was wrongly filtering out internal applicants and said that the additional form was designed to bypass recruiter review and send CVs directly to the team. So, in this particular story, one element of the human safety net involved bypassing another human (Love, 2026).

Well, that part remained rather well concealed behind the polished HR-expert costume.

AI can make mistakes, therefore we need a human — or perhaps we could pull ourselves a little closer to reality:

The quality of human judgement in selection is hardly self-evident. We have decades of research showing that the same candidate can be evaluated differently because of their name, sex, background, the impression they create in an interview, or simply the evaluator’s own preferences. Poor human selection decisions have a considerably longer history than AI. This does not rule out AI finishing the race badly too. If it does, the reasons need to be examined.

“Human-in-the-loop” guarantees very little by itself. Which human? What do they see? What criteria do they use? Does anyone check whom they reject incorrectly? Do we actually know how well a recruiter’s judgement predicts who will perform well in the job?

Mention AI and we immediately start talking about bias, error rates and auditability. Meanwhile, “AI” itself is allowed to remain some sort of mystical black box. A CV-ranking system? A generative model? An autonomous AI agent capable of carrying out multiple steps independently? Under one convenient label sit very different tasks, decision rights and sources of error. Perhaps the same degree of precision would be useful when examining human HR decisions.

Experts in the net

It is surprising that an HR publication needs this kind of net to catch readers. Look at major HR events and we appear to be overflowing with HR experts. At least according to the business cards. Often generously decorated with one academic qualification or another.

One HR Portal article is obviously not a representative survey of the state of HR. It is, however, quite useful for showing what level of content can enter and circulate through that environment without much difficulty.

Does anyone actually make use of all that at least presumed expertise? Or is a conversation in a nice outfit enough, with half the time spent praising each other’s achievements and the rest composing a creative platter from low-hanging fruit?

Because if all this expertise still fails to produce the question of how and to what extent human, AI-based or even autonomous AI-agent-led selection can be considered reliable, lack of experts does not look like the most likely explanation.

If someone calls themselves an HR expert, it seems a fairly rational expectation that they follow developments in their field, continue learning and have access to what can currently be known about their own subject. How well a genuinely articulate expert would complement the suit, make-up and business card: someone working from current knowledge, capable of drawing their own conclusions and thinking critically.

Are HR programmes in 2026 at the point where the ability to read international literature — and, even more importantly, the desire to do so — is considered part of professional intelligence? Anyone who gains access to current research only after somebody else has translated, summarised or retold it necessarily builds their knowledge through somebody else’s selection and interpretation.

Stories told by students or candidates can be interesting and current. They do not replace the literature. We do not always know how accurate they are, what has been omitted, or how much has been distorted by intention or interpretation.

Expert status presupposes access to original sources and the ability to check for yourself what your claims are built on before passing them on to others.

Pre-chewed food makes digestion easier, but credibility gets absorbed on the way down.

Are we at the point where someone publishing in an HR outlet is expected to read the original source, retain relevant counterpoints and refrain from making claims stronger than the source itself? Or do clickable headlines and easily digestible stories make a little distortion acceptable while the industry’s standards continue to sink?

The question is whether we are already discussing this from the basement.

And if, in a post by an organisation that publishes guidance and commentary on ethical AI, opinion, ethical judgement and fact are presented with the same degree of certainty, how confidently should people rely on the usefulness of that guidance?

A little off-topic for this particular post, but I do wonder how many HR experts — academics or practitioners — occasionally sit with this question honestly and with some serious self-criticism: when they assess other people’s expertise, suitability and preparedness, what sources, how current a body of knowledge, and how much self-scrutiny does their own expertise rest on?

Sources

HR Portal. (2026). Google fejlesztői figyelmeztetnek: az AI tévesen szűri ki a jelentkezőket. HR Portal – full article

Love, J. (2026, August 10). Google’s AI Team Tells Job Seekers Its HR Filters Are Unreliable. Bloomberg. Bloomberg article – syndicated by Yahoo Finance

Lilien Gerlach

Behavioural Analyst | Organisational Behaviour | Author | Speaker

Photo by Milan Czar on Unsplash

Fine. Let’s slow down for a moment and zoom in: beliefs vs. the hoped-for state vs. reality.

For a minute, let’s look inside an average family and see what kinds of questions come up. We can then think about which particular version of “family” the boss had in mind.

Are there favourites? Who is more equal than the others? Does the loyal family member actually feel appreciated for that loyalty? Because, after all, they could have sailed off across the career seas a long time ago.
What happens with conflict? After a serious disagreement, or one badly handled conversation, do we start building walls, end the relationship, or is it worth trying to repair it? What happens when someone no longer wants to be part of it? And is loyalty really always reciprocal?

Well, that is the “family” part of the famous sentence.
You might be one of those who would rather have real conversations about where things stand, the facts, the results and the plans.
Then you need solid foundations. Don’t design an ice palace for 40°C heat that is expected to last.

“We’re a family” will be the subject of my next dissection. I’ll look at what the research actually supports, and what remains from which the decent intention behind the idea can genuinely be built. So that our ice palace does not melt before the first meeting is over.

When you look ahead, the forest may still be thick with undergrowth. At least you can see where you want to go, and cut the path in that direction so the people coming after you can follow with confidence.

In the next piece, I’m examining what the research actually tells us about what happens when an organisation calls itself a family.

It will be research-grounded and fairly robust. If that sounds like your kind of reading, click Subscribe so you don’t miss it.

Lilien Gerlach | Behavioural Analyst Notes

https://unsplash.com/@cmilan1

It is becoming difficult to talk about the next few years in consulting without AI appearing within the first few sentences. There is a fairly tangible reason for that. If parts of analysis, research and presentable output can be produced substantially faster, the position of three different players changes at once.

It is a little like moving on from candles, losing our fear of electricity and finally daring to switch on the light. It does help, of course, if a competent electrician has checked the wiring first and made reasonably sure that using the switch will not electrocute us.

With AI, we are roughly at that stage. The light is unquestionably brighter, the switch is convenient, and everyone would quite like to use it. Someone still needs to understand what is running through the wall. The user also needs to know that if a little thing is sticking out of it, with yellow metal inside and plastic around it, picking it up and examining it while wondering hmm, what could this possibly be? is probably best avoided.

Consulting firms need to recalculate where the value sits for which they can continue charging a premium. The people working inside them need to work out which parts of their knowledge and labour will remain scarce. Clients would do well to become rather more precise about what they are buying: information, analysis, access, judgement, responsibility — and which of those they actually received.

An AUD 440,000 Warning

There is already a fairly tangible example.

In 2025, Deloitte Australia produced a 237-page assurance review for the Australian employment department for AUD 440,000. The report was later found to contain non-existent academic references and a quotation supposedly taken from a Federal Court judgment that had never actually appeared in it. The revised version disclosed that an Azure OpenAI GPT-4o-based tool had been used for part of the work. Deloitte ultimately refunded the final instalment of the contract.

The easy conclusion would be: there we are, AI hallucinated again. Except that for an AUD 440,000 expert report, the client was hardly buying advice from a language model. It was buying it from Deloitte.

A consulting firm can, of course, use AI. It can use search engines, databases, modelling software, Excel spreadsheets or anything else that makes the work faster and better. For the client, however, the name on the finished product carries the promise that somebody checked the wiring before the light was switched on.

In internal correspondence, the Australian department explicitly raised the possibility that the citation problems affected confidence in the report’s findings. At that point, AI has already moved beyond a technology discussion.

A model can make mistakes, as the accompanying exhibit rather helpfully demonstrates, and humans are hardly strangers to the activity either. At the end of the day, somebody still has to check the machine’s work, notice the error, put their own name to the result, and the client still buys somebody’s professional credibility.

So What Is a Partner For?

In a recent post, Ben Appleton lands on much the same issue through McKinsey’s new “distinguished partner” role. He argues that as AI reshapes consulting, domain expertise, commercial judgement, senior relationships and the ability to turn complex specialist knowledge into organisational change will become more valuable. His word for the next generation of partners is “translator”.

Rather than the approving I too think something impressively sophisticated comments that so often flood a thread, thoughtful professional objections and additions made consulting’s still-living reason for existence unusually visible.

Julius Hollmann thought much of this sounded rather familiar. A good McKinsey, Bain or BCG partner has always been expected to understand the client, grasp the problem, connect different disciplines and carry a decision through. He also argued that T-shaped, lateral and out-of-the-box thinking may become more valuable, while some specialist roles narrow. Samik Das approached the change from a similar direction: in his view, the partner’s role has essentially always involved change management, while the more immediate organisational consequences of AI may emerge closer to engagement-manager level.

That is an awkwardly good question for McKinsey’s new label. If the most important capabilities of the “partner of the future” turn out to be the same ones that made a partner good yesterday, there are at least two possibilities. Either the relative weight of those capabilities is genuinely changing, or the industry is repackaging an old expectation in AI-compatible language.

Here, “translator” already means rather more than rendering technical language comprehensible to a board.

Rolled into a single “future partner” profile, all of this could quite easily become another carefully wrapped competency list, lacking only “strategic mindset”.

The comments pulled the package apart. Luca De Angeli suggested “orchestrator” rather than “translator”: somebody has to connect domain expertise, strategy, technology and the operating reality of the business. Olivier Burnouf, referring to a McKinsey interview with Sven Smit, added the data constraint: much of the most interesting information used in larger research exercises simply is not sitting there waiting for AI. It may be proprietary data, newly created information, or knowledge held by people. Sridhar Peddisetty separated the production of intelligence from the decision itself. Vinod Venugopal put CFO-quality, traceable data underneath “qualified subjectivity”.

It becomes harder to describe the partner’s job as simply translating whatever the machine says into understandable human language. Someone first needs to know what the machine worked from, what was missing, where professional judgement entered the process, and how far the final product placed in front of the client can be traced back.

What Trust Does

Trust performs a very concrete function in this chain. A board receives an analysis, listens to the people behind it, asks questions, argues and eventually decides. It does not reconstruct every data source, rerun every model or spend three days checking footnote 184. If it had to, very few decisions would ever get made.

At some point, it accepts that other people have done work it is not going to repeat. A Deloitte, McKinsey or BCG logo forms part of that. So does the partner’s name, the method, the origin of the data, the checking process and the fact that, if something goes wrong, there is somebody available to answer the question: how exactly did you manage this?

Within such a system, trust saves time, permits division of labour and allows decisions to be made even where nobody can see the entire chain alone. Which is also why discovering that it was placed badly can become expensive.

In the Deloitte case, that assumption was damaged: that somebody had actually looked through what went into the finished product.

The operational consequences of trust become particularly visible when the issue is no longer the credibility of a report or a client relationship, but whether an institution gets money again the next morning.

Monday Morning, 2008

If you are experiencing a slight sense of déjà vu: yes, the time machine is taking us back to 2008.

For an entire weekend that September, bankers, regulators and political decision-makers tried — to put it more sophisticatedly — to assess the chances of the system surviving. Less sophisticatedly: who would still be standing on Monday morning. And, as a minor footnote, that particular Monday morning went on to shake rather a large part of the world.

Without taking another trip through the entire mechanics of Lehman’s collapse, I covered that in more detail in Too Big to Believe: The behavioural logic behind Lehman’s collapse and the role of confidence. Here, I am pulling out only the thread of trust. The length of this article is already making a reasonable attempt on the reader’s patience.

Lehman’s short-term funding continued to function for as long as enough counterparties kept saying yes to financing it for another day.

The bank relied heavily on short-term funding. In the repo market, it repeatedly had to convince the other side that the collateral was acceptable, the money would come back, and Lehman would still be there tomorrow. While that continued, Lehman could maintain its short-term funding. Once increasing numbers of counterparties no longer wanted to take the same risk for another day, Lehman ran out of time very quickly.

Nobody needed a perfect view of Lehman’s entire balance sheet for that to happen. It was enough for growing numbers of counterparties to become uncertain about assumptions they had previously been willing to make. Trust here was not a mood. It had funding consequences.

None of this makes AI the next financial crisis, nor does a consulting report operate like the repo market. Even my imagination would have to work rather hard to sustain that comparison.

The narrow common point is that in both cases somebody has to make a decision without being able to verify everything independently. Somewhere in the process, information, a person or an institution has to be judged reliable enough for the next step to be taken.

That bleak weekend in September 2008 shaped the financial world for years. Eventually, the system recovered. Would the consulting industry have as much time if its own ominous Monday morning really were approaching?

If the friend walking beside us has just stepped into a brown, suspiciously well-formed pile, and a few metres ahead there is another one directly in our path, analysis still has a certain practical use: we can walk around it.

On this still-hypothetical Monday morning, the trouble would look rather different. Clients would simply start calculating differently: what they are paying for, what they can produce themselves, and which parts still genuinely require the person or firm whose name, judgement and professional responsibility they have been buying.

When the Client Discovers How Easy It Is to Press Enter

One of consulting’s longstanding advantages has been information asymmetry. More markets, more benchmarks, more experts and more hours of human work sat on the other side of the table. Clients paid for that access and processing too.

The Enter key has already shaved a surprisingly large piece off it. You still have to ask a question first, of course, and how you ask it is not entirely irrelevant. But even a less tech-savvy executive can reach a usable first analysis within minutes that, only a few years ago, might have set at least a small project team in motion.

That does not make the executive a McKinsey partner. It does, however, give them something against which to compare what the McKinsey partner puts in front of them.

The client will probably continue not to recheck every reference in a 237-page report. That is part of what external expertise is bought for. The material arrives, questions are asked, a decision is made, and everyone moves on. If something later goes seriously wrong, the reconstruction can quite easily reach the lawyers’ desks.

At that point, the origin of the data suddenly becomes interesting. So does which model touched it, who checked the output, who approved it, what the contract promised and whose name appeared at the end.

In the Deloitte case, the name carried the weight: Deloitte delivered the report as a professional product, and the client was paying for that name as well.

AI is beginning to pull at the same business model from two directions. Expert-looking output can be produced increasingly quickly and cheaply. Checking it, standing behind it professionally, defending it, implementing it and still being there months later has not become a cheap pastime.

Clients, meanwhile, have more and more tools with which to reach a usable first answer themselves. A growing share of the consulting premium will therefore depend on what happens after Enter: whether somebody spots the missing data, stops the weak inference, stands behind their own professional judgement, and remains there when things go wrong.

If those elements are missing, an AUD 440,000 report eventually prompts a rather prosaic question:

Why, exactly, did it require a consultant?

Subscribe now

Subscribe now

Photo by Milan Czar on Unsplash

There is a small piece of truth in it. Something to point back to later. Something happened, something was said, someone had a role, made an appearance somewhere, achieved something minor. So you cannot simply dismiss the whole thing as a bare-faced lie. But by the time the story is finished, that little truth has been made into something far grander than the facts could possibly justify. Rather like a shell game. The damn little ball is in there somewhere, of course. We are simply no longer looking where it was first shown to us.

Gradually, the details produce a picture of someone more successful, more thoughtful, braver, more knowledgeable, more sensitive or more important than the facts warrant. A touch of mannered performance sometimes helps, just to make sure. Long, winding sentences that say remarkably little. A voice lowered almost to a whisper. Oddly stressed syllables. Performed modesty. Elaborate professional language with even less substance behind it than the dove in the Hungarian folktale of King Matthias and the clever girl. At least the dove was there before it flew away. Eventually, you begin to wonder whether the problem is yours, whether you are somehow failing to grasp how these sentences are supposed to fit together.

What is the point of making such a grand story out of an ordinary event? The listener is expected to draw conclusions about the person from the event: how modest, how thoughtful, how sensitive, how important. The event becomes evidence of qualities it was never capable of establishing.

Recognition, too, often seems to arrive entirely uninvited. Someone else reels off the impressive credentials, mentions the unexpected interest or remarks on how far people have travelled to attend. The person being celebrated looks pleasantly surprised. As though the world had discovered something extraordinary entirely of its own accord. Naturally, there is little mention of where all the information for the glowing introduction came from.

For a while, all this is simply tiresome. But hear enough of these stories and plain accuracy somehow begins to seem rather unimpressive. A sentence that describes only what actually happened sounds almost inadequate beside one that can turn every crumb of truth into a much grander life story.

Social circles gradually teach people what sort of stories are worth telling about themselves if they do not want to disappear into the background. First comes a detail that may well be true. It still seems a little thin, so another layer is added. A role becomes an important role, attendance becomes contribution, contribution becomes a professional turning point, and the professional turning point becomes a method of their very own.

If that works, on to the next event. Then a third. Eventually, the separate stories add up to a beautifully polished account of an extraordinary professional career. At least from a distance. Squint a little and you can barely see the difference between what actually happened and what was added to the story afterwards.

After so many carefully constructed stories, the only question left is whether all the effort is worth it. What happens if the audience decides to take a closer look at what is actually on offer?

There may be an enormous marketplace left at the end of it all. One solitary market trader, a stall groaning under the weight of everything on display.

But where have all the customers gone?

Based on a weekend observation.
Lilien Gerlach | Behavioural Analyst Notes

Subscribe now

Only recently did I realise that it now has a catchy English name, has become a trend and comes with an inspirational story: naked quitting.

Naked quitting means resigning before you have another job lined up. The current English term is new. The behaviour behind it, and the original name for it, are much older.

The term can be traced back to the Chinese word luǒcí. Its literal meaning is roughly “naked resignation”: leaving without another job, a secure transition or guaranteed income for the months ahead.

By late 2010, the term was already being discussed in publicly available sources as a Chinese workplace buzzword. One article from December 2010 explicitly asked whether resigning without another job was a liberating decision or a reckless impulse. Another listed luǒcí among the important workplace terms of 2010. In those articles, the risk to income, the expected length of the job search and the possibility that the decision had been made too hastily were still central to the phenomenon (Huaxi Metropolis Daily, 2010; Qilu Evening News, 2010).

Around 2011–2012, the English-language Chinese press was already calling it naked resignation. The definition had not changed: someone resigned before finding another job. Reports at the time mentioned work overload, dissatisfaction, changing employee expectations and shifts in China’s white-collar labour market. One person interviewed by China Daily had enough savings to last at least a year and was therefore willing to spend months waiting for the right job. Even in this early account, which emphasised autonomy, the financial background was a visible condition (Wang, 2012).

In 2015, a resignation letter containing just ten Chinese characters went viral. Gu Shaoqiang, a secondary-school teacher, wrote after eleven years: “The world is so big. I want to see it.” A photograph of the letter spread rapidly across Chinese social media and received tens of thousands of reactions and shares (Yao, 2015).

Here, resignation became a symbol of freedom, postponed dreams and taking back one’s own life. A later report described the teacher as the heroine of a generation of workers hungry for adventure. Leaving without a safety net was no longer presented simply as a risky labour-market decision. Attention shifted to the kind of person capable of making it: someone who recognises their own desires and has the strength to follow them (Chinanews.com, 2015).

In 2018, a China Youth Daily survey of 1,972 workers aged between 18 and 35 found that 23 per cent had resigned without another job offer. Nearly half had considered it but had ultimately not done it. This was a survey reported in the press, so it should not be treated as a precise national prevalence figure. It does show that naked resignation had by then become a recognisable and widely discussed workplace behaviour (Xinhua, 2018).

By 2024, the term naked quitting had appeared in the English-language press. On 17 March 2025, the Cambridge Dictionary language blog recorded it among its new words and defined it as leaving a job without having another one to go to. This documented a new use of language; it was not evidence that the behaviour itself began in 2024 or 2025 (Cambridge Words, 2025).

What Changed?

The early Chinese term named the absence of a safety net. In today’s Western articles and posts, resignation increasingly appears as proof of personal autonomy, self-respect and courage.

Someone had been miserable in a bad job for a long time. One morning, or after one particularly awful meeting, they decided they had had enough. They resigned without knowing what would come next. The first relief—the sense that one decision had ended the oppressive feeling in a single stroke—may have been followed by the panicked thought, “Oh God, what have I done?” By the time they had gathered their things, they were riding the rollercoaster of relief and a new kind of anxiety all the way home.

But once the story is told in public, it does more than record a feeling. It interprets the decision for an audience. It tells us how to see the person who resigned: as someone who was afraid, but acted anyway.

And now let’s name the elephant in the room: what do they live on until they have another job?

The Financial Conditions of Courage

Leaving without another job carries very different levels of risk for people in different circumstances. A few months without income may be manageable for someone with substantial savings, an earning partner, family support or a profession in which their skills are easy to sell. The same decision may create a housing or debt crisis for someone supporting a household alone.

Using Australian longitudinal data, Gerards and Welters (2020) examined how liquidity constraints affected unemployed people’s job searches and subsequent employment. Jobseekers facing serious financial hardship searched more intensively. It did not shorten their search, and the pay or hours in the job they found were no better. Those who found work while under severe financial pressure were less satisfied with their pay and hours, were more likely to think about resigning again, and had less stable employment over the medium term.

The study looked at unemployed jobseekers, not specifically at people left without work after naked quitting. It matters here because it shows what can happen when the financial room to manoeuvre runs out after resignation. A person may put more time and energy into the search without finding a job any faster or on better terms. As their reserves shrink, they also have less freedom to be selective or to wait.

A working paper by Clymo and colleagues (2022) looks directly at the relationship between wealth, quitting and being laid off. Using worker panel data, they found a U-shaped relationship: both low-wealth and high-wealth employees were more likely to move from employment into non-employment, but for different reasons. Their model suggests that workers with less wealth may accept riskier jobs to get out of unemployment more quickly. Some wealthier workers may leave voluntarily because they can afford more time out of work.

This has not yet been published as a peer-reviewed journal article, so it should be treated as preliminary evidence. It does not prove that every wealthier person resigns simply because they can comfortably afford to. It does support the point that wealth is not a minor background detail. It helps determine who can afford time outside work and who has to find a new income as quickly as possible.

That difference rarely appears in accounts of naked quitting. Courage gets a full paragraph. The financial runway gets, at most, a throwaway half-sentence.

A household budget is not merely the minimum required to stay alive. There may be children, a mortgage, insurance, transport and regular healthcare costs. People may also pay for subscriptions and activities that keep them connected to their cultural and professional world. They may need that one break a year that allows them to carry on. It is easy to dismiss these as optional comforts when you are not the one who has to give them up.

Someone who takes all of this into account and stays in a bad job for the time being may be making a rational assessment of risk. They may know perfectly well how much the work is harming them. They can also see the consequences of resigning.

An article about naked quitting therefore offers a useful picture only if it presents the financial conditions of the decision alongside the courage. How long would the savings last? Was there another income in the household? Were there dependants, loan repayments or substantial fixed costs? How likely was the person to find another job at a similar level?

Without those details, the account presents as personal virtue what is partly economic room to manoeuvre.

Why Do We Post It? Or, Lord, Let Me Share My Courage with the World

Let’s be honest: if the relief after resigning were enough in itself, it would be enough simply to feel good.

A social-media post serves another purpose. It turns the resignation into a public statement of identity. The text usually provides the preferred interpretation in advance: this was courage, choosing myself, drawing a boundary, breaking free. From the opening lines, the writer offers an interpretation in which the decision appears not as recklessness but as a personal achievement.

My reading is that many of these posts process the expected criticism in advance. The writer calls it courage before someone else can call it financial recklessness. Admitting the fear raises the moral value of the decision:

I didn’t know what would happen. I was terrified, but I did it anyway.

The sentence directly communicates fear. Indirectly, it also says that the writer did what many people cannot. They can present themselves as an example to follow without ever saying so outright.

This is where we can recognise one possible structure of humblebragging: self-promotion wrapped in a complaint or apparent modesty.

Across nine studies, Sezer and colleagues (2018) examined humblebragging. They defined it as self-promotion disguised as a complaint or modesty: the speaker wants to make a positive quality or achievement visible while avoiding the social cost of openly boasting. Participants judged humblebraggers as less sincere and less likeable, and in some studies as less competent. The strategy could work even worse than straightforward complaining, because at least the person complaining was perceived as more honest.

The research did not examine posts about resigning. It is relevant to naked quitting because of the structure of the communication. Admitting uncertainty and fear may ask for sympathy, while “I did it anyway” makes courage and agency visible.

I am not, of course, saying that every post about resigning is a humblebrag. Some people are asking for help, sharing information, making workplace abuse visible or simply announcing that their situation has changed. A single text cannot tell us the writer’s internal motivation with certainty.

The structure can still be observed. Uncertainty and fear provide the complaint or modesty layer. The underlying claim is that the writer managed to do what many others cannot. The account may ask for sympathy and admiration at the same time.

Naked quitting can therefore become a strange kind of status signal. The person posting appears to have jumped without any safety net. The wealth, partner’s income, family support or strong position in the labour market that made the jump possible may remain outside the account.

Does Someone Else’s Courage Make Me a Coward?

A bad job is already difficult enough. Social media can add comparison: someone else recognised their value, drew a boundary and had the courage to act. I am still sitting here.

Fukubayashi and Fuji (2021) used two studies to examine the relationship between career-related social-media content, social comparison and career frustration. The first involved 309 Japanese employees. The second used a seven-day experience-sampling method and collected 1,254 responses. Viewing positive career posts was associated with social comparison. Upward comparison—seeing the other person’s position as better than one’s own—was associated with greater career frustration. In the first study, comparison was also linked to the feeling that one ought to escape one’s current career situation.

The research did not specifically examine naked-quitting posts, and the two studies do not show that every resignation success story causes frustration. Applying the relationship here is defensible because these posts are also positive career content and can offer a direct basis for comparison: someone else was in the same bad situation and still left.

The reader gets the carefully constructed story of the person who acted. On their own side are the mortgage, the children, an uncertain labour market and savings that may last who knows how long.

They now have two problems. Their job is bad, and they may also feel that they are not brave enough to leave it.

And What About Those Who Could Leave, but Stay?

There are also people who have savings, a realistic chance of finding another job and no immediate financial obligation that makes leaving impossible. Yet they remain for years in a workplace that is deeply anxiety-inducing or damaging.

There may be perfectly rational reasons for this too. Mitchell and colleagues (2001) used the concept of job embeddedness to examine what ties people to their work and their wider environment. They distinguished three areas: links to other people, teams and groups; fit with the job, organisation and community; and everything a person would sacrifice by leaving. Job embeddedness helped predict intentions to quit and actual voluntary turnover beyond job satisfaction, organisational commitment, perceived job alternatives and job search.

This helps explain why being able to afford to leave does not mean someone will leave. They may fear losing their professional identity, status, colleagues, established position or the chance of returning to the same field later. As an eternal optimist, or by avoiding a direct reckoning with the situation—perhaps through endlessly reframing it—they may also hope that things will soon improve.

When those reasons no longer explain staying, and the situation involves persistent anxiety, paralysis, self-damaging adaptation or recurring relational patterns, therapy may be relevant. The question then is what keeps someone in a harmful environment even when leaving is realistically possible.

That, however, is a different story. It does not justify applying the same psychological explanation to people who stay because they are financially constrained.

The Missing Sentence

Resigning can be a brave decision. Leaving an oppressive or humiliating workplace may require considerable personal effort.

It is not fair to praise the courage while quietly leaving out the rather significant fact of financial security.

If someone tells us how afraid they were, the heroic account has room for the small print too: how they paid for housing, their children’s expenses and everyday life in the months that followed.

As long as we are shown only the decision itself, the reader or listener is left to imagine the background that made it possible. What each person reads into that, and how they interpret it, is probably material for another essay—also not a short one.

Subscribe now

References

Cambridge Words. (2025, March 17). New words – 17 March 2025. About Words – Cambridge Dictionary blog. https://dictionaryblog.cambridge.org/2025/03/17/new-words-17-march-2025/

Chinanews.com. (2015, July 25). Teacher’s resignation: A love story not an adventure. People’s Daily Online. https://en.people.cn/n/2015/0725/c90882-8926221.html

Clymo, A., Denderski, P., & Harvey, L. A. (2022). Wealth, quits and layoffs [Working paper]. SSRN. https://doi.org/10.2139/ssrn.4248008

Fukubayashi, N., & Fuji, K. (2021). Social comparison on social media increases career frustration: A focus on the mitigating effect of companionship. Frontiers in Psychology, 12, 720960. https://doi.org/10.3389/fpsyg.2021.720960

Gerards, R., & Welters, R. (2020). Liquidity constraints, unemployed job search and labour market outcomes. Oxford Bulletin of Economics and Statistics, 82(3), 625–646. https://doi.org/10.1111/obes.12345

Huaxi Metropolis Daily. (2010, December 9). Resigning without another job lined up: Carefree or impulsive? [Chinese-language article]. Sina Education. https://edu.sina.com.cn/j/2010-12-09/1001197016.shtml

Mitchell, T. R., Holtom, B. C., Lee, T. W., Sablynski, C. J., & Erez, M. (2001). Why people stay: Using job embeddedness to predict voluntary turnover. Academy of Management Journal, 44(6), 1102–1121. https://doi.org/10.5465/3069391

Qilu Evening News. (2010, December 14). 2010 workplace buzzwords: “Rubber people,” exam-job seekers and naked resignation make the list [Chinese-language article]. Sina Education. https://edu.sina.com.cn/j/2010-12-14/1529197322.shtml

Sezer, O., Gino, F., & Norton, M. I. (2018). Humblebragging: A distinct—and ineffective—self-presentation strategy. Journal of Personality and Social Psychology, 114(1), 52–74. https://doi.org/10.1037/pspi0000108

Wang, W. (2012, January 31). ‘Naked’ resignations on the rise. China Daily. https://www.chinadaily.com.cn/business/2012-01/31/content_14510459.htm

Xinhua. (2018, August 12). One in five Chinese youths resign without new job. China.org.cn. https://www.china.org.cn/china/2018-08/12/content_58518822.htm

Yao, Y. (2015, April 25). Quitting jobs to fulfill dreams no longer taboo in China. China Daily. https://www.chinadaily.com.cn/cndy/2015-04/25/content_20537275.htm

Lilien Gerlach | Behavioural Analyst Notes

Subscribe now

The manager is reading a version that has already been selected, processed, arranged, and described. The system has chosen which information counted. The recruiter may have accepted the summary. The platform may already have marked one profile as a cleaner match and another as lower fit. The hiring discussion may then begin from an order created outside the meeting and visible to everyone in it.

This article follows how AI-supported HR output becomes decision material, and where HR can still check whether a person has been read accurately enough to decide on.

On 19 May 2026, the European Commission published draft guidelines on the classification of high-risk AI systems under the EU AI Act and opened a targeted consultation. The consultation was originally due to close on 23 June and was later extended to 23 July 2026. The guidelines are intended to help providers, deployers, and market surveillance authorities assess whether an AI system should be treated as high-risk. For many organisations, the document will first move through legal, compliance, procurement, and technology teams. They will identify systems, review vendor material, classify tools, and prepare documentation.

For HR, the classification question reaches directly into the way an AI-supported result becomes a judgement about a person. Under Annex III of the AI Act, employment-related systems can fall into high-risk categories when they are used for recruitment and selection, including targeted job advertising, application filtering, and candidate evaluation. The same area also covers systems used for work-related decisions such as promotion, termination, task allocation, monitoring, and performance or behaviour evaluation.

Not every AI-supported HR tool will automatically be high-risk. Classification depends on the intended use, the system’s influence on the decision, and whether it involves profiling or materially affects access, evaluation, promotion, termination, task allocation, or monitoring.

In recruitment, selection, performance evaluation, promotion, internal mobility, and worker management, AI outputs can influence access to work, pay, status, future opportunity, and reputation. In these settings, a score or recommendation helps shape how a person becomes available for decision. The governance question for HR is practical: can the organisation reconstruct how that judgement was produced?

AI enters a process that already reduces people into formats

HR decisions have always relied on formats: the CV, application form, interview note, assessment report, performance rating, calibration grid, succession chart, and manager comment. Each one turns a longer working life into something that can be compared, discussed, stored, and moved into a decision.

Reduction is part of HR work. A hiring panel cannot read an entire career. A promotion committee needs evidence in a form it can compare. A performance review process needs categories. A succession meeting needs names, roles, risks, and readiness levels. The risk appears when the shortened version carries more authority than the evidence behind it can support.

AI is now being added to this existing system of reduction. It is used to parse CVs, score applications, summarise interviews, rank candidates, flag risk, analyse performance patterns, support workforce allocation, and identify potential. These tools may improve consistency, reduce administrative load, and help managers handle large volumes of information. They can also make a shortened version of the person look complete.

A parsed CV can give a messy career history an orderly shape. A behavioural score can make different people look comparable. An interview summary can make an incomplete exchange look processed. A risk flag can sound as if something has been detected. A ranking gives the meeting an order before anyone has discussed the work behind it.

The output can be useful while it stays tied to the evidence that produced it. Once the wording moves without its source, later decision-makers may discuss the wording as if they were discussing the person.

The judgement trail

An AI-supported recommendation reaches the manager after several prior choices have shaped it. A data source was chosen, a workflow approved, and the available information processed before a recruiter, HR business partner, or manager read the result. In many cases, someone then turned it into a note, shortlist, dashboard, promotion pack, or leadership update. The judgement trail follows how information about a person becomes a claim used in a decision.

It should show what data entered the system, what context stayed outside it, what the model produced, who reviewed the output, how the result was rewritten for HR or management use, where disagreement was possible, and which forum used it. A technical audit trail can show that the system processed data and produced an output. The judgement trail shows how that result acquired authority inside the organisation: who read it, who accepted or challenged it, which wording carried it forward, and how it affected the final decision.

The EU guidance gives this HR practice question a regulatory frame. Classification tells an organisation which systems may fall under high-risk obligations. The judgement trail shows whether the organisation can account for the way those systems affect actual people decisions. A tool can be classified correctly while the organisation still lacks a clear account of how its result entered a hiring, promotion, performance, or workforce decision.

Inside HR workflows, the issue appears in ordinary handovers. A recruiter reads a score and turns it into a shortlist note. A hiring manager reads an interview summary and asks fewer follow-up questions. A talent team uses a potential rating in a succession meeting. A line manager sees an attrition-risk flag and starts treating an employee as less committed. A performance dashboard shapes how a calibration discussion describes contribution. Each handover changes the weight of the original result, because each one gives it a new use.

The small steps that turn people into decision data

The shift begins in the ordinary materials of HR work: the CV field, the interview note, the transcript, the score, the summary, the shortlist. A candidate becomes a keyword pattern. A career history becomes a sequence of titles, gaps, tenure lengths, and sector labels. An interview becomes a transcript, then a summary, then a few behavioural tags. A manager comment becomes a performance indicator. A promotion case becomes a readiness rating. An employee’s activity becomes a productivity signal. A development need becomes a risk category.

These translations are useful while they remain connected to observed behaviour and relevant evidence. Trouble begins when later discussion treats the translated phrase as if it still carried the full case. Take a candidate with an irregular career path. A human interviewer may learn that the person moved sectors after a market contraction, took time out for caring responsibilities, joined a smaller firm after relocation, or accepted a lateral move to gain operational experience. A system may process the same history through tenure, title progression, keyword match, and gap patterns. The eventual output may appear as “inconsistent trajectory” or “limited role fit.” If that wording reaches the selection meeting without the surrounding context, the meeting may never examine the career logic behind it.

Inside organisations, the same mechanism can appear in performance analytics. A tool may show lower visible activity during a quarter. The missing context may be project recovery, client escalation, mentoring, cross-functional coordination, illness, restructuring, or a manager who failed to record contribution accurately. If the result later appears as “lower engagement” or “reduced productivity,” the employee may be discussed through a narrower version of the work they actually performed.

What AI adds to an existing compression

Executive search has long worked with a similar compression, even without AI. A long career becomes a shortlist paragraph. A credibility judgement becomes a few lines in a candidate note. Client discussions may turn on phrases such as “credible with investors,” “too operational,” “limited scale exposure,” or “difficult to read.”

An executive search consultant who writes “difficult to read” should be able to explain the observation: which part of the interview, reference conversation, career pattern, or client discussion produced that wording. That explanation may be incomplete. The client can still ask, the consultant can answer, and the judgement can be tested.

AI changes the scale and speed of the same compression. A phrase that once shaped one search process can now be generated, repeated, ranked, and carried across hundreds of profiles before a hiring manager begins the discussion. If similar wording appears in an AI-generated summary or ranking note, the organisation may not know whether it came from a relevant behavioural signal, a weak proxy, or a missing piece of context.

A phrase in one search process can be questioned by someone who remembers the interview, the client conversation, or the reason for the note. A generated label in a large HR system may pass through several screens and meetings before anyone asks what it was based on. By then, the organisation may have acted on it without treating it as a decision.

Where judgement can distort

Distortion first enters at data entry. Fields, keywords, documented outputs, historic ratings, recorded activity, and labelled achievements are easier to process, so they become easier to use. Work that is harder to document receives less weight. Many organisations depend on work that is relational, corrective, preventive, cross-functional, or politically difficult to name. That work often leaves a poor data trace.

Studies of algorithmic pre-employment assessment tools show how much rests on earlier choices: what data is collected, what outcome is predicted, how bias mitigation is claimed, and how validation is carried out. Raghavan and colleagues’ analysis of algorithmic hiring vendors is useful for HR leaders because it follows the design choices that shape the score. By the time the score appears, earlier choices have already narrowed what it can mean.

Context loss creates the next risk. Career breaks, migration, caring responsibilities, restructuring, role ambiguity, industry disruption, and poor previous management can all shape a person’s record. HR needs to distinguish between context that would be inappropriate to use and context without which the evidence will be misread.

Management language changes the output again. AI output often leaves its technical form and reappears in managerial language as “risk,” “fit,” “readiness,” “potential,” “trajectory,” “engagement,” or “evidence strength.” These words travel easily in leadership settings because they sound decision-ready. They also reduce the visible uncertainty in the original case. Once a person has been described as “low fit” or “high risk,” participants may discuss the category rather than the observations behind it.

Review is another weak point. Process documents often treat human oversight as the safeguard: someone reviews the output before the decision is made. In practice, a reviewer may accept a result without understanding it. A busy manager may use a ranking because it saves time. A recruiter may rely on an AI-generated interview summary and stop looking for what the summary left out. Research on automation bias has shown that human reviewers may over-rely on algorithmic advice, including in settings where other available information would support a different conclusion. Oversight has substance when the reviewer understands the output, can see the relevant evidence, and has authority to challenge the recommendation.

The final risk appears when the person affected by the decision cannot correct the reading before it becomes final material. A candidate or employee may receive the outcome without seeing how the judgement formed. They may be told they were not shortlisted, not selected, not promoted, not ready, or not meeting expectations. The process can be administratively clean and still leave the affected person with no clear way to correct the record.

Contestability before the decision

A people decision becomes much harder to repair after the final material has been locked. Before that point, HR can create review points where a score, summary, rating, or label is checked against the evidence it claims to represent.

In recruitment, that review point sits before the shortlist is finalised. A recruiter or HR business partner compares the AI-generated shortlist note with the original CV, application material, and any available human observation. A career break, lateral move, sector change, or missing keyword can then be checked before the wording reaches the hiring manager’s pack.

In selection, the review point sits before an interview summary is circulated. The interviewer can add a short note when the generated summary misses the main point of an answer, removes relevant hesitation, overstates certainty, or turns a qualified response into a cleaner claim than the candidate made.

In promotion and succession work, the review point sits before the pack goes to calibration. A readiness rating, potential marker, or mobility note should be checked by someone close enough to the work to see whether the generated description reflects actual contribution. If the summary narrows the case too far, the correction should travel with the pack.

In performance review, the line manager should state whether activity data reflects the work actually performed. Recorded activity may miss project recovery, client escalation, mentoring, cross-functional coordination, or work done to prevent a visible problem. If the dashboard under-represents contribution, the context should be added before the calibration discussion uses the signal.

These are evidence checks. They give the organisation a way to separate an accurate signal from a poor proxy before the proxy becomes part of the official record.

The practical question is where the organisation allows correction to happen. A candidate who receives a rejection after a ranked shortlist has already been accepted has little room to correct the reading. An employee who first sees a label during a final performance conversation is already arguing against a settled description. A manager who wants to override a risk flag needs a way to explain why the signal does not fit the case.

The meeting may never discuss the full case. It discusses the wording that reached the pack. Contestability has to exist before the wording becomes final decision material.

Contestability also affects internal behaviour. Employees learn which decisions can be questioned and which labels are treated as settled. A manager who sees “lower engagement” on a dashboard may begin the next conversation from suspicion. An employee who knows the context may have to repair a judgement that has already travelled through the system. The later the correction appears, the harder it becomes to change the decision material.

The EU guidelines move this beyond an internal HR quality discussion. If a system is used in employment or worker management, and its output affects access, evaluation, promotion, termination, task allocation, or monitoring, the organisation needs more than a general assurance that a person remained in the loop. It needs a record of how the loop worked.

What HR should map now

HR should begin by identifying every point where AI or AI-supported tools enter people decisions: targeted job advertising, CV parsing, application filtering, candidate scoring, interview transcription, interview summarisation, assessment interpretation, shortlist generation, performance analytics, promotion calibration, succession planning, internal mobility, learning recommendations, attrition-risk modelling, workforce allocation, and productivity monitoring.

Some of these uses fall squarely within Annex III employment and worker-management categories. Others become legally and organisationally significant when they materially influence access to opportunity, evaluation, promotion, task allocation, monitoring, or employment conditions.

The inventory should show how each output enters a decision, in addition to where the tool is used. For each point, HR leaders should ask five questions:

What human activity became data?
The aim is to see whether the system is reading evidence that belongs in the decision. A sales number, interview phrase, job title, response pattern, or activity log may be relevant. It may also be a poor proxy for capability, judgement, potential, or contribution.

What context was left out?
The missing context may be irrelevant, or it may change the meaning of the pattern. HR has to know which kind of absence it is dealing with.

Who reviewed the output?
Naming a reviewer is a thin control if the reviewer cannot see the underlying evidence or challenge the system’s recommendation. Oversight has to leave a trace of what was checked.

Where could someone disagree?
Correction has to be possible before the output hardens into a decision. Disagreement may come from a recruiter, candidate, employee, line manager, HR business partner, works council, legal reviewer, or decision panel. The organisation needs that path before a dispute appears.

How did it reach the final decision?
The same score carries different risk depending on its place in the workflow. It may remain background input. It may set the order of the shortlist. It may appear in a leadership pack. Someone may rewrite it into a recommendation. It may enter a calibration meeting as a category.

For example, an AI-supported shortlist entry should record more than the score. It should show which evidence the system used, which context remained outside the model, who reviewed the output, whether the reviewer accepted or challenged it, and how the result appeared in the hiring manager’s material. If a recruiter overrides a ranking because interview evidence changes the interpretation, that override and its reason should travel with the shortlist.

The leadership problem

High-risk AI in HR crosses several functions: procurement sees the vendor, IT sees the integration, legal sees the classification, HR sees the workflow, and the hiring manager may see only the ranked shortlist. The candidate or employee may be the only person who knows which part of the profile was misread.

Each function can do its own part correctly while the organisation still lacks a clear account of how the judgement moved from evidence to decision. The judgement trail gives leaders a shared way to examine what happened before the case reached the meeting. Managers need enough process knowledge to see how the output entered the decision, what evidence it carries, what it omits, who reviewed it, who could challenge it, and how it influenced the final decision.

Organisations adopt HR technology to increase consistency, reduce administrative load, and improve speed. These aims are understandable in high-volume HR work. AI may also correct some weaknesses of purely manual processes. It can apply the same screening rule across a larger candidate pool, surface patterns that busy teams might miss, and reduce some forms of informal inconsistency.

Those gains still leave an evidential test. A tool can process information consistently and still process a poor proxy. It can reduce administrative work and still give decision-makers a thin version of the person. It can produce a plausible recommendation while leaving the organisation unable to show why that recommendation deserved weight.

HR AI governance needs a way to separate processing efficiency from evidential sufficiency. A tool may process information efficiently. A decision may still require more context. A system may produce a plausible recommendation. A leadership team may still need to ask whether the person has been made readable in the right way for the decision being made.

The practical test

The European high-risk AI discussion gives HR leaders a dated regulatory reason to examine a problem that has existed for years: organisations often decide from shortened versions of people.

A practical test is to choose one AI-supported people decision and follow the path from input to output to human review to decision forum. Identify what became data, what disappeared, who reviewed it, who could challenge it, and how the output reached the final discussion. Then ask whether the final decision-maker saw enough evidence to justify the conclusion.

That path will usually show whether the gap is in compliance, evidence, managerial interpretation, or the handover between them. When a people decision is supported by AI, HR should be able to show how the judgement moved from input to output to human decision.

An organisation can have a policy, a vendor file, and a human sign-off, and still be unable to show how the person became readable enough to decide on. The judgement trail should show how that reading was produced, where it was checked, and whether it was accurate enough to justify the decision

.

Leave a comment

The scene is easy to file under generational anxiety. Young people, AI, jobs, fear, bad timing. Convenient categories. Too convenient.

Graduation is already a strange ceremony. It congratulates people for surviving one system of measurement, then sends them into another that has learned to smile while measuring them. School at least had the vulgar honesty of grades. Work prefers vocabulary.

There is something almost too Huxleyan about it now. The ceremony is polite. That is part of the problem. Nobody says: welcome to the invisible exam. That would ruin the catering.

The old white-collar promise was simple enough to survive for decades. Study, perform, enter a profession, and knowledge will become security, autonomy, status and progression. The promise was unevenly distributed, and many people knew that even while repeating it. It still shaped the story families, universities, employers and workers told about education and work.

A degree did more than certify learning. It suggested entry into a world where the mind would be used, judgement would matter, responsibility would grow, and performance would slowly become a professional life. The prestige of white-collar work depended on that sequence: knowledge, contribution, progression, security.

The pressure now begins where work struggles to prove what it does with a person’s knowledge.

Many office roles are full of activity. Meetings, preparation, status updates, summaries, presentations, alignment calls, careful wording. People work. Often they work a great deal. The question is what the organisation does with the judgement inside that work.

A person sees a problem early. The organisation thanks them for the input. By the next meeting, the problem has become a sensitivity. By the final deck, it is a footnote with no owner. The employee learns something from this. Accuracy is allowed to travel when it stops embarrassing the room.

This is where the subject connects to The Lies That Hold the System Together. The book examines the point where official organisational stories separate from daily experience. That separation rarely needs a dramatic lie. It can happen through polite language, diluted warnings, internal proof of activity, and decisions that never quite meet the information people already have.

Buy the Kindle edition

A company speaks about value while employees spend much of the week proving that work happened. It speaks about responsibility while the space for judgement remains narrow. It speaks about autonomy while people learn which observations arrive too early and which sentences cost more than the meeting can afford.

White-collar work still has to prove that knowledge can reach consequence.

That proof appears when a specialist sees a risk and the decision changes. It appears when a customer complaint moves as operational information, not reputation management. It appears when a manager receives bad news and keeps it intact long enough for someone to act on it.

That is the part of work people still want to believe in.

AI arrives at an uncomfortable moment because many white-collar jobs had already lost part of this route. When a role is largely made of text movement, internal coordination, summary production and status updates, generative systems make an old problem harder to avoid. They ask a blunt question many organisations preferred to leave in human fog: where, exactly, was the judgement the organisation needed?

This is why an optimistic AI speech can sound so misjudged at a graduation ceremony. The graduate has just left a system where evaluation had a visible form. They enter a system where performance, visibility, loyalty, judgement and organisational tolerance mix in less explicit ways. The speech asks them to be excited about the future. The labour market asks them to trust an older bargain whose evidence has become weaker.

People already inside white-collar work know this from the other side. Good work rarely travels by itself. It needs someone to keep it recognisable while it passes through the organisation. Sometimes it loses force because it has been made acceptable before it has been made useful.

The workplace rarely calls this an exam. It has more elegant methods.

The better version of white-collar work is easy to recognise when it happens. Someone sees something before it becomes expensive. The organisation lets the observation remain sharp enough to be useful. A decision changes. A process changes. A customer is handled differently. A risk is understood earlier.

The person can see that their work reached something.

That matters more than most corporate language about meaning. People can tolerate dull tasks, conflict and repetition for longer than motivational literature likes to admit. What corrodes judgement is the repeated experience of seeing something clearly and learning that clarity must be reduced before it can move.

After a while, people adapt. They observe more cautiously. They take less responsibility where responsibility has no consequence. They learn what the organisation rewards: the useful observation, the well-timed silence, the pleasant version of the truth, the visible performance of being busy.

The mistrust around the old career promise should be treated as information. Graduation language still draws from an older dictionary. It speaks as if education leads into a professional world where knowledge naturally becomes contribution. Many people inside that world already know how often the route breaks.

White-collar work still has to prove that knowledge can reach consequence. That proof is created in daily work: when an observation reaches the people who can use it, when a warning survives the meeting in which it first becomes inconvenient, when a person’s judgement changes what the organisation does next.

A degree may get you in. Office status may make the role look safer than it is. The evidence sits in the path between what a person knows and what the organisation does with it

.

Last week I decided to try something from The Lies That Hold the System Together, a book I had just finished. There was a part in the book about what happens to a message after a meeting, and it felt a bit too familiar.

Buy the book

So yesterday, at the end of our Monday meeting, I did something different. I skipped the usual motivational round and the reframing, and asked everyone to write down one sentence: what in today’s discussion do you think will need a decision? Honestly, they looked at me as if I had asked them to hand over their phones.

Then the answers came in. Three out of five people had written something different from what I had in mind. One wrote: ‘We’ll keep an eye on this.’ I had been thinking we needed a decision by Friday. Another wrote: ‘The supplier side is still taking shape.’ No name, no deadline, no cost.

Reading it was pretty grim, but at least I don’t have to rely on my hunches anymore. Hmm... Not bad. I caught where my message had been shortened. I noticed which sentence had lost the decision. It became clear who had heard risk and who had heard an admin task.

Right, people are already coming over... I’ll come back to this later. Looks like Monday’s meeting has an afterlife today.

P.S. Hard to put into words, but this is a strangely special Tuesday morning.

Have a similarly odd little Tuesday, all you leaders in the middle.

Lilien Gerlach — Notes from a behavioural analyst Author of The Lies That Hold the System Togethe

r

By Thursday morning, it appears in a short update. The client name is still there, but the decision request has become “needs further alignment”.

By Friday, it enters the pre-read for the leadership meeting. The client name is gone. The date is gone. The owner has become “the team”. The issue now sits under “delivery dependency”.

On Monday, leadership receives the pack. The issue is visible. It has a line, a status, and a short explanation. The version on the table still looks professional. It also carries less decision value than the first version.

That last passage before a decision is easy to underestimate. It is where information is shortened, renamed, moved under safer headings, and separated from the person who first raised it.

Blame turns this into a personal drama too quickly. The more useful question is traceability: can leadership still see who changed what, why it changed, and whether the original decision request is still attached to the issue?

Before the meeting, leaders can already be assessed by whether they ask what happened to the information on its way to the table.

Who removed the date?
Who changed the owner?
Who replaced the client name with a category?
Who changed the decision request into something to monitor?
Who kept the status colour unchanged while the wording underneath became more careful?

The last mile before a decision is not just preparation. It shapes what leadership believes it is deciding on.

The Lies That Hold the System Together will be published next Monday, 8 June 2026.

.

Subscribe now

Let’s review 8 behaviour signs and decide afterwards.

• The person who raised the concern disappears from the wording. It becomes “feedback”, “noise”, “a concern”, or “some comments”.

• The issue stays in private conversation for weeks before it enters a report, dashboard, risk register, or leadership pack.

• Dates, numbers, client names, costs, deadlines or ownership are replaced with general language.

• A warning is renamed as a “dependency”, “timing issue”, “alignment point”, “sensitivity”, or “area to monitor”.

• More evidence is requested at the point where the existing evidence would already require a decision.

• The status colour remains unchanged while careful explanatory text is added underneath.

• People describe the problem accurately at the coffee machine after the meeting, then present a cleaner version in the formal forum.

• The issue moves to the next meeting, next review, next quarter or next phase without naming the decision that was avoided.

Are there consequences? Definitely.

Can that affect your life in an organisation? Probably much more than you can imagine.

Just to mention a few: budgets, project priorities, accountability, careers, customer relationships, and the quality of the decision itself.

The Lies That Hold the System Together will be published next Monday, 8 June 2026.

I will share the link here on Monday.

Subscribe now

Most people who have worked in an organisation will recognise the sentiment. Perhaps the boss is narcissistic, another colleague is passive-aggressive, or someone in HR is simply a toxic person. Such descriptions rarely require much explanation. The listener may immediately think of someone who fits.

Their appeal is understandable. Difficult working relationships leave people with unanswered questions. Why does a manager take credit for other people’s work? Why does a colleague appear supportive in meetings but undermine decisions afterwards? Why does someone behave pleasantly towards senior management while treating subordinates quite differently?

People want to understand what motivates those around them, particularly when another person’s behaviour affects their work, reputation or livelihood. Psychological language offers an attractive explanation. A few familiar characteristics appear to fit, previously confusing experiences begin to make sense, and someone else’s behaviour suddenly seems predictable. There is considerable satisfaction in believing that we can finally see behind the curtain, and an equally considerable market for explanations that promise to lift it.

The commercial appeal of psychological explanations

The workplace well-being industry has access to an almost inexhaustible supply of familiar human difficulties. Books, conference presentations, coaching sessions, masterclasses and corporate training programmes address experiences that practically everyone recognises. A speaker hardly needs to persuade participants that difficult people exist at work. Most arrived with a shortlist.

Psychological terminology makes these experiences especially interesting because it appears to reveal something about the person responsible. The manager who reacts badly to criticism becomes a narcissist. The colleague who expresses resentment indirectly becomes passive-aggressive. Someone whose behaviour repeatedly causes difficulties becomes a toxic personality.

These expressions have different meanings and degrees of scientific precision. Narcissism is a recognised personality construct, with narcissistic personality disorder representing a distinct clinical diagnosis. Passive-aggressive behaviour describes a pattern of indirect resistance or hostility, while toxic person is a considerably looser everyday judgement. Yet all three can serve a similar purpose in workplace conversations: they turn observations about someone’s conduct into an explanation of who that person is.

Nick Haslam’s work on concept creep examines how psychological concepts associated with harm and pathology have expanded to include a broader range of experiences and behaviours. His analysis concerns concepts such as bullying, trauma, abuse and mental disorder. It demonstrates how the boundaries of psychological terminology can change, sometimes helping people recognise previously neglected harms while also increasing the risk of pathologising ordinary experiences. [1] The popularity of workplace personality labels raises a related concern, although Haslam’s research does not establish how frequently professionals misuse them.

Someone observes a few instances of behaviour, recognises characteristics associated with a familiar psychological category and concludes that the category explains the individual. The interpretation becomes especially convincing when it appears to account for several unpleasant experiences at once. What began as a description of particular actions develops into a broader judgement about personality and motives.

Professional settings can strengthen that conviction. An organisation invites someone to speak about leadership, mental health, organisational behaviour or employee well-being because it expects relevant expertise. Participants reasonably assume that the material has been selected and interpreted by someone who understands the subject. That expectation gives professional education much of its value. It also creates an opportunity for questionable explanations to acquire credibility.

Some non-clinical providers present themselves as knowledgeable about human behaviour while making claims about people’s personalities or motives that exceed both their professional competence and the available evidence. Coaches, workplace well-being trainers, mental health consultants and speakers may have legitimate expertise within their respective fields. Their qualifications do not necessarily establish the competence required to assess the psychological characteristics of individuals they have never examined.

A speaker may refer casually to a narcissistic manager while discussing workplace conflict. A coach may describe a difficult relationship through the language of toxic personalities. A trainer may use recognisable character descriptions to make a presentation more engaging. None of these examples necessarily involves an explicit invitation to diagnose colleagues. Participants may nevertheless apply the descriptions to people they know, sometimes leaving with greater confidence in their ability to interpret personalities than the material justifies.

When an explanation begins to control the evidence

Consider a hypothetical workplace situation. An employee prepares financial and operational analyses for several important management presentations. Her manager presents the findings to senior executives without clearly acknowledging her contribution. Initially, she assumes the omission was accidental, but something similar happens during another project.

She raises the issue directly with her manager. The conversation is uncomfortable, and she receives no satisfactory explanation. During the following weeks, she is excluded from meetings she previously attended. Her next performance assessment contains criticism of her attitude and cooperation, although earlier reviews had been positive.

The employee begins to suspect that these events are connected. She discusses the situation with a colleague, who suggests that her manager is a narcissist. The colleague may have encountered the description in a workplace training session, a coaching conversation or one of the many presentations explaining how to recognise difficult personalities. The description appears to fit. He takes credit for other people’s achievements, reacts badly to criticism and seems willing to damage an employee’s reputation to protect his position. The employee now has an explanation for behaviour that previously seemed confusing.

She may even be correct about some of his motives. The difficulty is that her explanation has travelled considerably further than the available evidence.

She knows which analyses she prepared and how they were subsequently presented. Emails, document histories and meeting records may establish whether her contribution was acknowledged. The timing of her exclusion from meetings can be examined, as can the reasons given for changing her responsibilities. Her performance assessment can be compared with previous reviews and evidence of her actual work. Those records could support a serious complaint about the appropriation of work, unfair treatment or retaliation following a legitimate concern.

The manager’s underlying personality remains a separate question. The observations are insufficient to establish a clinical condition, and they do not necessarily reveal the motives behind every decision. The manager might also offer a different account of the presentations, the meeting changes or the performance review. Some explanations may be legitimate, while others may prove inconsistent with the records. A fair investigation would examine them.

Once the employee adopts the personality label with sufficient confidence, subsequent events may be interpreted through that explanation. A critical remark becomes further evidence of narcissism. A friendly interaction with senior management becomes calculated impression management. A disagreement over project priorities becomes a deliberate attempt to undermine her.

Some interpretations may be correct. Others may have ordinary explanations. The label makes those possibilities harder to distinguish because new observations are increasingly assessed for their consistency with an existing judgement.

Raymond Nickerson’s review of confirmation bias describes how established beliefs and expectations can influence the selection and interpretation of evidence. The research concerns a general feature of human reasoning, with implications extending well beyond workplace relationships. It provides a plausible explanation for how an initially uncertain interpretation can become increasingly resistant to contradiction. [2]

Suppose the employee approaches HR and reports that her manager is a narcissist who has been deliberately trying to destroy her career. HR may question her ability to make such an assessment. The manager may argue that she has developed a personal hostility towards him and is interpreting routine management decisions accordingly. The discussion can then become a dispute about her interpretation of his character, while the allegations concerning his actual conduct remain unresolved.

A competent investigation should examine the work she prepared, the presentation records, the meeting exclusions and the deterioration in her performance assessment. In practice, however, an unsupported personality judgement may give the manager an opportunity to challenge the employee’s credibility before the underlying conduct has been properly investigated.

An employee with a potentially legitimate complaint may therefore have made her position more difficult to defend. The explanation that initially helped her make sense of the situation may now complicate her attempt to establish what happened.

Who carries the consequences?

Employees do not enter workplace conflicts from equal positions. A manager may influence assignments, career opportunities, performance assessments and professional reputation. An employee considering a complaint must weigh the likelihood of being heard against the possibility of damaging an existing working relationship or jeopardising future employment.

Elizabeth Morrison’s review of research on employee voice and silence examines the conditions influencing whether employees raise concerns, including expectations about how others will respond and the consequences of speaking up. Her work helps explain why organisational relationships and perceived risks influence the decision to report problems. [3]

Introducing a compelling psychological explanation into this situation may increase the employee’s confidence without improving their ability to substantiate a complaint. The employee may believe that the terminology gives their account greater credibility. When that assumption is challenged, they discover that recognising a familiar description provides little protection.

The professional who introduced the concept can leave the session and move on. The employee must return to work with the person they have labelled, decide whether to raise a concern and defend their account if challenged. The professional’s involvement may have ended before the practical consequences of the interpretation become apparent.

The same problem can develop in coaching relationships. A coach may use a psychological label to acknowledge a client’s difficult experience. The client may find the explanation reassuring, particularly after struggling for months to understand another person’s behaviour. Repeatedly discussing the relationship through that label can reinforce an interpretation that has never been adequately tested.

A coach may understand the formal limits of their professional role and still fail to observe those limits consistently in conversation. When a client describes several troubling incidents, a coherent explanation can be attractive to both parties. An apparently tentative suggestion may gradually become the accepted account of the other person’s behaviour, especially when subsequent sessions return to the same interpretation. The client’s confidence can increase without any corresponding improvement in the evidence.

Deliberate manipulation or financial exploitation is unnecessary for this to happen. A provider may sincerely believe that the explanation is helpful. They may have encountered similar situations, read extensively about the subject or received positive feedback from previous audiences. None of these circumstances establishes that their interpretation of a particular individual is reliable.

Professional explanations can influence decisions long after the original conversation has ended. That possibility deserves particular attention when the person receiving the explanation is dealing with a workplace conflict in which their employment, reputation or professional relationships may be affected.

Professional expertise comes with obligations

Coaches, mental health consultants, workplace well-being trainers, organisational consultants and conference speakers work with related psychological concepts, but their qualifications and professional responsibilities differ considerably. Some have formal education in psychology or counselling-related disciplines. Others have completed shorter professional programmes or developed their expertise through management, coaching or training experience.

A corporate trainer may be competent to explain a psychological concept and discuss its possible relevance to workplace behaviour. A coach may be competent to help a client examine experiences and consider alternative interpretations. Neither activity, by itself, establishes the competence to make confident assessments of another person’s personality, motives or psychological condition.

Professional boundaries may be well understood in principle and considerably less visible in practice. Providers can acknowledge the limitations of their qualifications while communicating interpretations with a level of authority that exceeds those limitations. Participants may hear an explanation of an individual’s behaviour without recognising how little evidence supports the conclusion.

The World Health Organization’s 2022 guidelines on mental health at work provide a relevant example. Their recommendations support training managers to recognise mental health concerns, respond appropriately and improve working conditions. The guidance states that such training does not qualify managers to diagnose or treat mental disorders. [4] Although this recommendation concerns mental health training, it illustrates the importance of distinguishing knowledge of psychological concepts from competence to assess particular individuals.

A presentation explaining narcissistic characteristics, for example, should distinguish observations of behaviour from conclusions about the person displaying it. Repeated credit-taking may warrant concern. Hostile reactions to criticism may provide further information about how a manager handles disagreement. Patterns across situations can strengthen an assessment of conduct.

Such observations may establish that the behaviour is unacceptable. Determining whether it reflects a stable personality characteristic, a particular motive, organisational incentives or circumstances unknown to the observer requires further evidence. Omitting these limitations can leave participants with a misleading understanding of what psychological terminology allows them to conclude.

A provider may encourage an audience to identify familiar personality types among colleagues and subsequently remind participants that they should not diagnose other people. A disclaimer may be appropriate and useful when the presentation has consistently explained the limits of psychological interpretation. The difficulty arises when participants have repeatedly been encouraged to identify psychological characteristics in people they know, while those limitations receive only passing attention.

If most of a session encourages participants to match recognisable behaviours to psychological categories, a concluding reminder may have little influence on interpretations already formed. The examples may have been memorable precisely because participants recognised people they know. By the time the disclaimer appears, some may already have accepted the suggested explanations as descriptions of those individuals.

A provider who repeatedly encourages these connections cannot reasonably treat their occurrence as entirely unexpected. Calling an example illustrative does not establish that the audience understood its evidential limitations. Professional responsibility requires attention to the conclusions that a presentation is likely to encourage and the confidence with which participants may subsequently use them.

Commercial incentives can complicate this responsibility. A presentation promising to help employees understand difficult people has immediate appeal. A session explaining the limitations of personality inference, the ambiguity of behavioural evidence and the difficulty of establishing motives demands more patience from the audience. The first promise is easier to market and can be delivered through memorable descriptions of recognisable characters.

That commercial advantage provides little justification for leaving important limitations unexplained. Professionals are paid for expertise their audiences may lack. When a provider knows that a psychological category can easily be misapplied, making that category memorable while leaving its limitations obscure represents a failure of professional responsibility.

The organisation still has a responsibility to investigate

The employer’s responsibility remains even when employees express their concerns through questionable psychological terminology. If someone reports that a manager repeatedly claimed credit for their work, the organisation can examine the project records. If meeting invitations were withdrawn after a complaint, the timing and reasons for those changes can be investigated. If a previously positive performance assessment deteriorated, HR can examine the evidence supporting the new evaluation.

The manager should have an opportunity to respond, and the investigation should consider information that contradicts the employee’s account. An allegation does not become true because the employee describes it convincingly. Equally, attaching an unsupported psychological interpretation to an allegation does not relieve the employer of the responsibility to examine the reported conduct.

Personality explanations can also influence how organisations understand recurring problems. A manager’s conduct may be influenced by individual characteristics, but it may also be enabled by poorly designed incentives, weak oversight, ambiguous responsibilities or a reporting structure that discourages complaints.

Labelling the manager as toxic can encourage an organisation to treat the problem as an isolated personal failing. That interpretation may be convenient when examining the surrounding management arrangements would require more difficult decisions.

An organisation may have rewarded the manager’s performance for years, overlooked complaints or accepted practices that contributed to the problem. Removing the individual may be justified, but the organisation still needs to establish whether the conditions that enabled the behaviour have changed.

Similar difficulties arise when employees are encouraged to manage problematic relationships through personal coping strategies while the employer leaves the underlying working conditions untouched. Psychological support can help employees articulate their experiences and recognise when they need assistance. Employers still require procedures capable of establishing what happened, assessing the relevant evidence and addressing misconduct or organisational failures. A well-being programme cannot fulfil those responsibilities on the employer’s behalf.

The responsibility behind the promise

Psychological explanations can help people describe confusing experiences and recognise behaviour that deserves closer examination. Their usefulness, however, depends partly on understanding what those explanations can reasonably establish.

The consequences of unjustified certainty can extend beyond the person applying the label. A manager may face an unsupported accusation. An employee with a legitimate complaint may undermine their own position. An organisation may become distracted by competing personality judgements while failing to investigate the events that gave rise to the conflict.

These are foreseeable risks arising from a plausible sequence of interpretation and action. The research discussed here does not establish how frequently that sequence occurs or how often it results in harm. The possibility nevertheless gives professionals reason to examine the confidence their material creates and whether that confidence is justified.

I find it difficult to reconcile a declared commitment to employee well-being with practices that encourage consequential judgements about colleagues while leaving the limitations of those judgements insufficiently explained.

I have little patience for professional explanations that make other people’s personalities appear easy to read while leaving the audience to deal with the consequences of getting it wrong. When someone’s reputation, employment or well-being may depend on the judgement, that is an unacceptable standard of professional care.

References

[1] Haslam, N. (2016). Concept creep: Psychology’s expanding concepts of harm and pathology. Psychological Inquiry, 27(1), 1–17.

https://doi.org/10.1080/1047840X.2016.1082418

[2] Nickerson, R. S. (1998). Confirmation bias: A ubiquitous phenomenon in many guises. Review of General Psychology, 2(2), 175–220.

https://doi.org/10.1037/1089-2680.2.2.175

[3] Morrison, E. W. (2023). Employee voice and silence: Taking stock a decade later. Annual Review of Organizational Psychology and Organizational Behavior, 10, 79–107.

https://doi.org/10.1146/annurev-orgpsych-120920-054654

[4] World Health Organization. (2022). Guidelines on mental health at work. World Health Organization. Recommendation 4, Manager training for mental health.

https://www.who.int/publications/i/item/9789240053052

Lilien Gerlach
From a behavioural analyst’s notebook

.

Who finds out first? Who does the information go back to? Who writes their performance review? Who assigns the next shift? Who might decide on a possible transfer? The employee uses these questions to calculate what may happen the next day on shift, in the next review, in a transfer decision, or in the first leadership discussion.

A recent case gave this question a concrete point of reference: the Tata Consultancy Services back-office case in Nashik.

According to Reuters’ report of 11 May 2026, India’s National Commission for Women (NCW) found a “toxic workplace environment” at TCS’s Nashik unit. The Commission identified harassment, bullying, and breaches of India’s rules on workplace sexual harassment. The unit under review had around 150 employees, within a company that employs more than 584,000 people globally.

In a global company, the presence of central regulation seems self-evident. The local employee still reads the situation from their own environment. They see who can be approached. Who responded before. Which issue stayed verbal. Which complaint was followed by a change in someone’s shift pattern. Who received protection, and who was left alone. They learn how the complaints route works from these cases.

According to the investigation reported by The Times of India, the Nashik office lacked the local information needed for complaint handling. Employees could not clearly see whom they could contact, how the members of the internal committee could be reached, and what route a report would follow.

The final detail is this: the investigation also found that several female employees had wanted to complain, but held back because of fear, social pressure, family stigma, and concern about transfer or dismissal.

The organisational life of a complaint

For the employee, the existence of a procedure, a committee and a legal obligation is not enough. Before submitting a report, they need to see how the system works in their own workplace situation.

Who takes over the case? Who writes it down? Who follows it up? Who prevents the information from returning to the same local circle whose conduct the employee was trying to report?

An organisation can handle such a situation in two ways. In the first, the report triggers the prescribed process: the case is recorded, the complainant can see the next step, and local actors cannot informally reverse the whole matter. In the second, the employee sees that the report exists as a procedural option, while the consequences appear first in their own working day.

This is where the official route of the case separates from what the employee expects to happen the next day. Fear of retaliation rarely appears as an open threat. Employees often fear consequences that are difficult to prove later with a single document: a worse shift pattern, exclusion from a task, a more distant leadership response, or a weaker review under the language of collaboration or attitude.

These consequences are not always visible from the outside. They still shape the decision, because the employee is not thinking in legal categories. They are thinking about the next working day.

A complaint is also a social event. Others watch what happens to the complainant: whether the issue becomes a written case, whether the procedure starts, whether the complainant’s work changes, whether they receive protection, or whether the whole matter slides back into the familiar local way of working.

Silence in this setting is learned risk calculation. Employees use previous cases to assess what they risk if they speak. If, in the past, a complainant received worse shifts, was left out of a project, or their case circulated verbally for weeks, others learn from that. The mere existence of a policy is not enough.

Why this case caught my attention

TCS employs more than 584,000 people. From the outside, an employee in a company of that size appears to be part of a large system: policies, internal committees, legal frameworks and compliance processes surround them. At the moment of complaint, the risk moves very close. The employee does not have to work the next day with the whole company. They have to work with the people who may see the report, shape their schedule, influence their tasks, review their work, or signal informally to others how they should be treated.

Even in a company with hundreds of thousands of employees, the consequence of a report may appear first in the employee’s next-day workplace environment, not in the central system.

The complaints channel is tested where the employee returns to work after submitting the report

Subscribe now

.

Subscribe now

Leadership research has long tried to define what makes a leader credible. The classic answers are familiar enough: be reliable, communicate clearly, understand the area you lead, be consistent, show empathy, and do not turn authenticity into an over-produced performance. Transformational leadership organises the question around inspiration, role modelling and a shared purpose (Bass, 1985; Bass & Riggio, 2006). Leader–member exchange theory focuses on the quality of the relationship between the leader and the employee: where there is more trust, respect and reciprocity, commitment and cooperation can become stronger (Graen & Uhl-Bien, 1995). The model of authentic leadership emphasises self-awareness, transparency, acting in line with internal values and balanced decision-making (Walumbwa et al., 2008).

These concepts are useful, though they can easily sound as if the company were a moral training programme and leadership credibility were a matter of a few competencies to develop. Organisations are more stubborn than that. They operate through shareholder expectations, market pressure, internal status games, bonus systems, headcount limits, information hierarchies and risk-averse decision routines. The leader tries to remain credible inside this space. Sometimes that requires personal integrity. Sometimes political judgement. Sometimes the ability to turn a message arriving from a spreadsheet into a human sentence without stripping all reality out of it.

Leadership credibility therefore cannot be treated purely as a personality trait. Employees often read the system through the leader. They watch what is actually backed, and what is not. Which value lasts only until it costs money. Which promise survives an uncomfortable decision. Which piece of information suddenly becomes confidential just as responsibility for the decision would become visible. From a behavioural analysis point of view, this matters because credibility is formed not only by what leaders say about themselves, but also by what their repeated decisions allow people to infer.

Trust as an economic resource

It is easy to talk about trust as if it were a soft human factor. Companies like trust for very hard reasons. Trust speeds up cooperation, reduces the need for control, makes information flow more easily, and lowers the internal energy loss created when everyone starts looking for a second meaning behind every sentence. A classic model of trust argues that trust is built, among other things, on perceptions of the other party’s ability, benevolence and integrity (Mayer et al., 1995). In a leadership context, this becomes very concrete: does the leader understand what they are talking about; do they take employees’ situation into account; do they stand by what they previously represented?

Organisational trust is therefore more than a matter of mood. It reduces operating cost. Where employees believe that the leader is not manipulating important information, fewer defensive routines are needed. Fewer extra checks, fewer cautious people copied into emails, fewer corridor explanations in the background. The organisation does not become paradise. It simply burns less energy protecting itself from itself.

Then comes the less celebratory part. Many organisations are happy to consume trust, but less willing to pay the cost of maintaining it. Producing trust requires time, consistency, explanation, predictability and sometimes short-term loss. Quarterly targets are a simpler genre. They do not enjoy long explanations. They prefer numbers, ideally before month-end.

This is one of the basic tensions of leadership credibility. Organisations often want leaders whom people trust while running incentives that reward behaviour that damages trust. Fast cost-cutting, reports polished on their way upwards, conflict-avoiding decisions, optimism communicated too early, problems named too late. None of this is automatically malice. Much more often, it is adaptation. The trouble begins because employees adapt as well. They say less, ask more cautiously, and pay closer attention to which truths are still worth bringing into the system.

Share Hidden Signals Analytics

The middle manager as a translation point

Leadership credibility becomes especially interesting at middle-management level. Senior leadership often speaks in strategic language; teams live with operational consequences. The middle manager stands between the two. They translate cost pressure, shareholder expectations, restructuring, headcount limits, AI implementation or an “efficiency improvement programme” into everyday sentences. That is already a difficult role. During translation, something is always decided: how much reality remains, and how much becomes manageable corporate language.

“We are increasing operational efficiency” can be an accurate business statement. It can also be a polite way of saying that the same work will need to be done by fewer people. A “more flexible operating model” can mean better decision speed. It can also mean that uncertainty has finally been given a friendlier name. The lovely thing about corporate language is that almost anything unpleasant can appear as a development opportunity if the font size in the presentation is large enough.

This is where the middle manager’s credibility becomes fragile. Their personal view of the decision is only one part of the matter. The way they handle the responsibility of translation also counts. If every uncomfortable decision is packaged as if it will bring long-term benefits for everyone, the team learns that the leader’s language does not inform; it cushions damage. If, on the other hand, the leader comments cynically on every senior decision, they start dismantling their own leadership role. Credibility often shows in how long a leader can remain accurate in a situation where the system encourages vagueness.

This is not a simple moral test. Middle managers often genuinely cannot say everything. There are legal, commercial, personal and strategic limits. Employees usually understand that. What they tolerate less well is unnecessary fog. The leader speaks imprecisely not because they are not allowed to be precise, but because precision would be more uncomfortable.

Transparency where information is power

Transparency is one of the favourite words of modern corporate language. Few organisations describe themselves as closed, opaque and selectively honest. Most companies promise open communication. Over time, employees learn that transparency can mean real explanation, and it can also mean carefully rationed information.

Information inside an organisation is not neutral material. It gives access, status, protection and influence. Those who know earlier what is changing can adapt earlier. Those who see the real numbers interpret official optimism differently. Those who understand where a decision came from can also see where questions might be asked. For that reason, transparency is more than a communication style. It also touches power relations.

Leaders often face genuine dilemmas. Information shared too early, while still incomplete, can create panic. Information shared too late can damage trust. Too much detail can drag attention away from the point. Too little detail can sound empty. Employees do not expect every strategic document to appear immediately on the shared drive. Given the usual state of shared drives, that would be a fairly modest danger anyway. What they watch is how accurately the leader speaks within the room they actually have.

“This information cannot currently be shared” is therefore a sensitive sentence. It may be entirely legitimate. It may be justified by data protection, commercial reasons or legal constraints. Credibility depends on how often it appears, in what situation, and whether it is later followed by a meaningful explanation. If the sentence appears whenever a question becomes uncomfortable, it stops functioning as a confidentiality boundary and becomes an organisational reflex. After a while, employees no longer hear the answer. They hear the pattern.

Empathy under budget pressure

Research on supportive and attentive leadership behaviour generally links it to higher trust, satisfaction and willingness to cooperate. For this article, the general value of empathy is less interesting than the practical room an organisation gives leaders to behave empathetically.

The organisational value of leadership empathy is not self-evident. A leader can be personally attentive while operating inside a budget that does not allow empathy to have practical consequences. They listen to an overloaded team, then continue with the same target and the same headcount. They acknowledge the risk of burnout, then remain unable to change the deadline. They speak in a human tone while the operating conditions stay the same.

In these situations, employees do not necessarily doubt the kindness. They look at its backing. Does anything follow from what the leader has heard? Does anything change in workload, priorities, expectations or the decision process? If nothing changes, empathy becomes a communication gesture. It makes the situation more civilised, but it does not change the conditions producing the situation.

That does not make empathy worthless. Leadership coldness rarely improves organisational functioning; it merely looks simpler. The problem begins when empathy becomes the cheaper substitute for system change. Listening without resources. Recognition without capacity. A caring tone alongside unchanged overload. This combination can reduce tension in the short term, but over time it teaches the organisation that human language and operational logic work in two separate departments.

Competence in uncertain conditions

Competence is one of the least romantic parts of credibility. Employees generally dislike following a leader who visibly does not understand the situation in which they are making decisions. Yet leadership competence today is rarely limited to technical expertise. Leaders often have to give direction in areas where they themselves are working with uncertain information: technological change, AI, market decline, geopolitical risk, supply chains, regulatory change, labour shortages and cost pressure.

In this environment, all-knowing leadership confidence quickly becomes suspicious. Employees notice when a presentation contains more certainty than reality does. No special training is needed. A few reversed decisions, three over-optimistic all-hands meetings and one strategic document containing the phrase “robust execution model”, while the hardest part of the project plan still has no owner, will usually do the job.

Credible competence here does not mean infallibility. A leader appears competent when they understand the nature of uncertainty, can distinguish between fact, estimate and hope, and do not confuse communication confidence with an information base. Research suggests that leadership effectiveness cannot be reduced to one single trait; technical ability, relational behaviour, decision quality and context all shape the result (Judge & Piccolo, 2004; Harms & Credé, 2010).

Competence becomes a credibility issue when the leader also has to handle their own uncertainty. If everything is presented as certain, and reality keeps forcing the leader to withdraw their own statements, employees do not only trust the specific decision less. They begin to recalibrate the leader’s ability to read the situation. This is one form of quiet trust loss.

Authenticity inside an organisational role

Authentic leadership theory argues that a leader builds lasting trust when they act with self-awareness, transparency and alignment with their values (Walumbwa et al., 2008). This is a strong idea, but it is easy to misunderstand inside an organisation. A leader is never present purely as a private person. They carry a role. They represent responsibility. Sometimes they have to communicate a decision they did not initiate. Sometimes they have to stand behind a target they do not personally see as ideal, because their leadership role requires them to implement it.

Authenticity, then, cannot mean unlimited self-expression. A leader cannot say everything simply because it feels honest. The organisation is not a therapy room, even if after some meetings there may be a case for one. Leadership authenticity is better measured by how far the leader can speak without distortion inside the role they hold. They do not over-dramatise their own honesty, they do not perform being the team’s friend while holding decision power over them, and they do not pretend every uncomfortable decision is a shared adventure.

Employees often sense this difference very accurately. They do not expect the leader to be completely personal in every situation. They expect the leader not to build false closeness where the relationship is in fact asymmetrical. A leader can say that they are unable to discuss certain things. They can also say that the team has to work with a decision, even if it is difficult. That is usually more credible than packaging every structural constraint as personal enthusiasm.

What does the employee learn from this?

Inside organisations, employees constantly learn which behaviours are worth it. They rarely describe it this way. They do not sit down in the morning and decide to update their model of leadership credibility. They watch simpler things. What happens to the person who brings bad news? Does the person who asks a clarifying question receive an answer? Is there any consequence when a leader promises something? Does the same rule apply at higher levels? How long is honesty treated as a value, and when does it become “a negative attitude”?

Research on employee voice and silence suggests that people weigh up whether speaking up about a problem, idea or criticism may carry personal risk, and whether it is likely to have any meaningful effect (Morrison, 2023; Milliken et al., 2003; Detert & Edmondson, 2011). This connects directly to leadership credibility. If a leader says they are open to feedback, then handles uncomfortable feedback with defensiveness, dismissal or silence, the team learns. Next time, less information arrives. It may not appear as visible resistance. It often appears as subtle editing.

This is one of the key points in organisational behaviour: loss of credibility does not always produce conflict. Often, it produces professional silence. People do not slam doors, hold revolutionary meetings in the tea kitchen, or write a manifesto for the internal newsletter. They simply become more cautious. They document more. They commit less verbally. They leave more questions inside their own heads. From the outside, the organisation continues to function. Less reality reaches the places where decisions about that reality could be made.

Why is it difficult to be credible today?

The conversation around leadership credibility often suggests that the answer lies in better self-reflection, better communication, more empathy and greater transparency. These are genuinely necessary. The picture remains incomplete, however, unless we examine the system in which these behaviours have to be practised. Leaders behave differently where short-term financial targets override everything else. They behave differently where uncomfortable information is punished on its way upwards. They behave differently where leadership bonuses are tied to cost reduction while the company values include respect for people. The latter is at least linguistically harmonious. Organisationally, less so.

The organisational conditions of credibility are therefore at least as important as the leader’s personal qualities. A leader may be fair, but if they regularly have to represent decisions for which there is no real explanation, their voice starts to wear thin over time. They may be empathetic, but if they have no room to reduce workload, the backing behind their empathy weakens. They may be competent, but if the organisation prevents the real risk from being named for political reasons, that competence cannot become fully visible.

This does not absolve the leader. Organisational constraint does not remove individual responsibility. It makes the question more precise. Leadership credibility is shaped by what the leader does with the room they have been given. How much they distort, how much they clarify, how much reality they allow through, and when they choose the accurate sentence over the more comfortable one. This is rarely spectacular heroism. More often, it is a series of small decisions from which employees gradually work out how much it is worth believing the leader.

A behavioural analyst’s comment

Everyday workplace half-sentences often show the state of an organisation more accurately than official culture surveys. Not because they are major pieces of evidence on their own, but because they are tied to repeated situations. Employees learn which sentences were followed by real change, when consequences failed to appear, which promise had operational backing, and which remained a communication gesture.

This connects to the earlier view of trust. The backing behind trust first becomes visible not in the leader’s intention, but in what the organisation does with accurate information. If bad news, an uncomfortable question or raw feedback regularly creates risk for the person who says it, employees will not necessarily start arguing. Often, they start editing. They speak more briefly, write more cautiously, leave more traces behind them, and become increasingly accurate at judging which sentence may carry what price.

This is one source of professional silence. From the outside, it does not always look like resistance, because the form remains the same: the meeting takes place, the report is produced, the status update enters the system. The content changes. There is less early warning, less half-formed concern, fewer uncertain but important signals. The organisation is not simply losing information. It is weakening its own ability to recognise patterns.

The behavioural reading therefore does not look for moral judgement. It looks at the adaptation that appears when the backing behind leadership statements repeatedly proves weak. Loss of trust often becomes visible before any major rupture. It appears first in how employees speak, document, ask questions and take risks.

Closing

Leadership credibility today is more than a leadership development topic. It is an economic, organisational and behavioural question. Behind a leader’s behaviour stand incentives, information hierarchies, shareholder expectations, decision risks and cultural patterns. Employees do not describe it in those words. It is unlikely that someone by the coffee machine will say: “the political-economic conditions of managerial credibility are problematic.” Fortunately.

The sentences are usually simpler. “We’ve heard this a few times now.” “Let’s see what actually happens.” “I wouldn’t say that now.” “Put it in an email as well.” These sentences are small, but they reveal a great deal about the state of the organisation. Where leadership statements regularly have weak backing, people often avoid open dispute and adjust their own behaviour instead: they phrase things more cautiously, document more, and bring less raw information into the shared space.

The organisation still works. Meetings continue, reports get written, targets enter the next quarterly plan. Meanwhile, more and more people learn that the accurate sentence has a price, while the vaguer sentence often carries less personal risk.

Leave a comment

Sources and further reading

Bass, B. M. (1985). Leadership and performance beyond expectations. Free Press.

Bass, B. M., & Riggio, R. E. (2006). Transformational leadership (2nd ed.). Lawrence Erlbaum Associates.

Detert, J. R., & Edmondson, A. C. (2011). Implicit voice theories: Taken-for-granted rules of self-censorship at work. Academy of Management Journal, 54(3), 461–488.

Dirks, K. T., & Ferrin, D. L. (2002). Trust in leadership: Meta-analytic findings and implications for research and practice. Journal of Applied Psychology, 87(4), 611–628.

Graen, G. B., & Uhl-Bien, M. (1995). Relationship-based approach to leadership: Development of leader–member exchange theory of leadership over 25 years: Applying a multi-level multi-domain perspective. The Leadership Quarterly, 6(2), 219–247.

Judge, T. A., & Piccolo, R. F. (2004). Transformational and transactional leadership: A meta-analytic test of their relative validity. Journal of Applied Psychology, 89(5), 755–768.

Mayer, R. C., Davis, J. H., & Schoorman, F. D. (1995). An integrative model of organizational trust. Academy of Management Review, 20(3), 709–734.

Milliken, F. J., Morrison, E. W., & Hewlin, P. F. (2003). An exploratory study of employee silence: Issues that employees don’t communicate upward and why. Journal of Management Studies, 40(6), 1453–1476.

Morrison, E. W. (2023). Employee voice and silence: Taking stock a decade later. Annual Review of Organizational Psychology and Organizational Behavior, 10, 79–107.

Walumbwa, F. O., Avolio, B. J., Gardner, W. L., Wernsing, T. S., & Peterson, S. J. (2008). Authentic leadership: Development and validation of a theory-based measure. Journal of Management, 34(1), 89–126

.

A job interview does more than test whether someone can answer questions under pressure. It gives an organisation an early reading of how that person handles disagreement, hierarchy, and interpersonal risk. Research on employee voice and silence is useful here. Voice concerns the discretionary act of raising ideas, concerns, or objections that matter to the work. Silence concerns the withholding of those concerns. Employees often remain silent because they expect negative consequences, because they assume speaking up will not change anything, or because they have absorbed tacit rules about when candour is unsafe [1–3].

Picture a job interview. A professionally competent, promising candidate takes their seat. The usual questions come one after another, interrupted from time to time by questions that sound interesting at first glance, but seem secondary.

• Tell us about a situation, within whatever details you are able to share, where you disagreed with your colleagues on a professional matter.

• You have to work from a report, and it contains an error made by your highly regarded line manager. What do you do?

Questions like these are already doing more than checking polish. They sample how a candidate narrates disagreement, whether they can name an error tied to status, and how much of their judgement remains intact when the answer may carry a social cost. Interviews are strongly social-evaluative settings, and the research literature treats impression management as a central part of what happens inside them. Applicants do not simply report who they are. They actively shape what is visible, and that shaping can affect interview ratings [4–6].

Then, at certain points in the interview, the candidate starts to weigh up how far their answer should reflect what they actually think. In an effort to avoid conflict, the answer often loses some of its original shape and force: the reservation becomes milder, the problem is named less precisely, and the professional judgement is not expressed in full. It may also be that the candidate wants the job badly enough to try to give the answer they assume an ideal candidate would give [6–8].

Compromise in an interview rarely begins with a blatant invention. More often, it begins with truncation. The candidate removes the part of the answer that would create friction. A substantive disagreement becomes a matter of communication style. A firm professional judgement becomes a cautious preference. An error made by a senior person becomes a delicate ambiguity that “needed clarification”. Research distinguishes honest from deceptive impression management, which matters because not every polished answer is false [7]. The pressure to manage the impression remains real, and interview anxiety is one of the factors that can push candidates further in that direction [7, 8].

In the short term, this can even look advantageous. A smooth, easy-to-work-with manner is welcome in many places. In the interview itself, however, that adjustment also changes the signal that becomes available. Less friction is created in the room, but less of the candidate’s professional judgement is made fully visible.

That short-term advantage is easy to understand. Candidates know that likeability, composure, and fluency can influence interview evaluations. Meta-analytic work on impression management shows that applicants’ self-presentational tactics do shape ratings, so softening an answer can feel like a rational adjustment to the setting [4, 5]. The difficulty is that the adjustment changes the signal the organisation receives. It does not only lower friction in the room. It also lowers the visibility of judgement.

In that sense, the interview can function as an early behavioural sample of how a person handles professional judgement under social pressure. Detert and Edmondson’s work on implicit voice theories shows that people carry taken-for-granted rules about when speaking up is risky or inappropriate [3]. Those rules are often enacted before a person is formally inside the organisation. The interview can be the first moment when the candidate demonstrates, in public, whether they preserve their view under pressure or convert it into something easier to receive.

From there, the reading needs to stay narrow. A compromised answer in a job interview can show that softening, partial withholding, or careful editing of professional judgement is available to the person under social pressure.

None of this is to say that interviewers themselves do not shape the interaction. Their own clarity, consistency, and tolerance for disagreement also affect whether a candidate’s compromise is a choice or a response to a perceived demand.

A job interview concentrates evaluation, hierarchy, acceptance and risk into the same short exchange. In that setting, a clear professional view can carry a social cost. The person responds by diluting, editing or holding back part of what they actually think. That response may be occasional. It may also belong to a broader pattern. The interview shows the response. It does not show its full range.

That reading also has support in the organisational literature. Managers do not respond equally to every form of speaking up. Burris found that employees who used more challenging forms of voice were viewed as worse performers and had their ideas endorsed less than employees whose voice was more supportive in form [9]. The finding comes from a specific organisational context and is not a universal law, but it illustrates a pattern that has been observed across multiple studies. A candidate who senses this possibility during an interview is not imagining a social risk that exists nowhere else. They are responding to a real organisational pattern. What the interview can show is that, under that kind of anticipated risk, the person may choose adaptation over full signal accuracy.

There is, however, an important qualification. Preserving professional judgement does not require theatrical bluntness. Research on applicant self-verification suggests that accuracy in self-presentation can work in favour of strong candidates. Moore and colleagues found that candidates with a stronger drive to present themselves accurately were, among high-quality candidates, more likely to receive an offer, partly because they were seen as less inauthentic and less misrepresentative [10]. This does not mean that every direct answer is rewarded. It means that faithful signalling is not automatically a disadvantage. In some cases, it reads as credibility.

If this pattern appears repeatedly in working life, it can reduce how much of a person’s actual judgement reaches shared decisions. The effect does not come from one compromised answer in one interview. It appears when softening, withholding, or careful editing becomes a recurrent way of handling professional risk. How much this matters, however, will vary across organisations, sectors, and national cultures – the cost of compromise is not the same everywhere.

That is where the broader organisational cost can begin. Teams learn through reporting errors, asking difficult questions, and surfacing concerns while there is still time to act on them. Edmondson’s work on psychological safety showed why this is difficult: behaviours that help teams learn often carry embarrassment or interpersonal threat [11]. Later meta-analytic work linked psychological safety to learning behaviour, task performance, and constructive contribution [12]. When the weakening of one’s own signal becomes a repeated way of handling professional risk, the organisation loses part of its access to what that person knows [1, 11, 12]. The loss is practical before it is moral. It affects what enters discussion, what gets corrected, and what remains invisible.

That also raises another question: out of these versions — not saying it, not saying it fully, and saying something quite different — which, if any, still falls within acceptable limits?

The most useful distinction here concerns signal accuracy. There are situations in which not saying something yet is understandable: the judgement is still forming, the facts are incomplete, or the timing makes the message impossible to hear properly. The decisive change often happens one step later, when the person does speak but weakens the answer enough that the receiver hears less concern, less certainty, or less independence than is actually present. Saying something quite different goes further still. At that point, the organisation is being given a misleading preview of how that person is likely to operate once hired. That matters because recruitment is, among other things, an act of inference under uncertainty. The organisation is not only selecting a skill set. It is estimating what kind of judgement will be available to the work later on [1, 10].

So what is the cost of compromise? And is it worth it?

Lilien Gerlach, behavioural analyst

Subscribe now

References

[1] Morrison, E. W. (2023). Employee voice and silence: Taking stock a decade later. Annual Review of Organizational Psychology and Organizational Behavior, 10, 79–107. https://doi.org/10.1146/annurev-orgpsych-120920-054654

[2] Milliken, F. J., Morrison, E. W., & Hewlin, P. F. (2003). An exploratory study of employee silence: Issues that employees don’t communicate upward and why. Journal of Management Studies, 40(6), 1453–1476. https://doi.org/10.1111/1467-6486.00387

[3] Detert, J. R., & Edmondson, A. C. (2011). Implicit voice theories: Taken-for-granted rules of self-censorship at work. Academy of Management Journal, 54(3), 461–488. https://doi.org/10.5465/AMJ.2011.61967925

[4] Levashina, J., & Campion, M. A. (2007). Measuring faking in the employment interview: Development and validation of an interview faking behavior scale. Journal of Applied Psychology, 92(6), 1638–1656. https://doi.org/10.1037/0021-9010.92.6.1638

[5] Peck, J. A., & Levashina, J. (2017). Impression management and interview and job performance ratings: A meta-analysis of research design with tactics in mind. Frontiers in Psychology, 8, Article 201. https://doi.org/10.3389/fpsyg.2017.00201

[6] Melchers, K. G., Roulin, N., & Buehl, A.-K. (2020). A review of applicant faking in selection interviews. International Journal of Selection and Assessment, 28(2), 123–142. https://doi.org/10.1111/ijsa.12280

[7] Bourdage, J. S., Roulin, N., & Tarraf, R. (2018). “I (might be) just that good”: Honest and deceptive impression management in employment interviews. Personnel Psychology, 71(4), 597–632. https://doi.org/10.1111/peps.12285

[8] Powell, D. M., Bourdage, J. S., & Bonaccio, S. (2021). Shake and fake: The role of interview anxiety in deceptive impression management. Journal of Business and Psychology, 36(5), 829–840. https://doi.org/10.1007/s10869-020-09708-1

[9] Burris, E. R. (2012). The risks and rewards of speaking up: Managerial responses to employee voice. Academy of Management Journal, 55(4), 851–875. https://doi.org/10.5465/amj.2010.0562

[10] Moore, C., Lee, S. Y., Kim, K., & Cable, D. M. (2017). The advantage of being oneself: The role of applicant self-verification in organizational hiring decisions. Journal of Applied Psychology, 102(11), 1493–1513. https://doi.org/10.1037/apl0000223

[11] Edmondson, A. C. (1999). Psychological safety and learning behavior in work teams. Administrative Science Quarterly, 44(2), 350–383. https://doi.org/10.2307/2666999

[12] Frazier, M. L., Fainshmidt, S., Klinger, R. L., Pezeshkan, A., & Vracheva, V. (2017). Psychological safety: A meta-analytic review and extension. Personnel Psychology, 70(1), 113–165. https://doi.org/10.1111/peps.12183

Conflict of interest is usually described as though it begins with a moment of private moral failure. Someone sees a clear choice, recognises that one path serves the organisation and the other serves them, then quietly backs themselves. It is a satisfying story because it is easy to understand and even easier to condemn. It also flatters the organisation, because it suggests the system was sound and only the individual went astray.

Subscribe now

Inside organisations, the reality is often less dramatic and more interesting. Conflict of interest frequently appears long before anyone experiences it as a conflict at all. A buyer is rewarded for cost savings. A quality lead is recognised for keeping complaints low. A divisional head advances on the strength of their own unit’s results. A consultant is more likely to be retained when a project continues rather than concludes. None of these arrangements looks especially sinister in isolation. In most companies they would barely raise an eyebrow. They would be filed under performance management, accountability or commercial discipline, which is one of the reasons the problem is so easy to miss.

The difficulty is that these arrangements do more than reward behaviour. They organise perception. They tell people what matters, where the visible risk sits, which outcomes will be noticed, and what a sensible professional is supposed to worry about. The buyer learns to see immediate cost with unusual clarity. The quality lead becomes finely attuned to failure rates, defects and the downstream consequences of short cuts. The divisional head sees threats to their own area quickly and the wider cost more faintly. After a while, people do not merely respond to incentives from the outside; they begin to reason from within them.

That matters because conflict of interest inside organisations is very often sustained by people who can give a perfectly sincere and perfectly plausible account of what they are doing. The buyer is not usually sitting there thinking, “I know this will hurt the business, but the bonus is the bonus.” They are more likely to think they are showing proper commercial judgement. The quality lead does not experience themselves as an obstacle placed in the path of progress. They think they are preventing avoidable damage that somebody else seems oddly relaxed about. The divisional head who protects their own numbers can sound every inch the responsible operator. Everyone has a case. Everyone has a metric. Everyone has a vocabulary for seriousness. This is what makes the subject behaviourally rich and organisationally awkward.

Once you look at it this way, conflict of interest stops being just a question of motive and starts to look like a question of structured partiality. People do not need to be cynical in order to become predictable. They do not need to be corrupt in order to become skewed. They simply need to occupy a role for long enough, under a set of incentives that makes one slice of reality more salient than the rest. Over time, that slice begins to feel like the whole picture. The person then experiences their own judgement as objective, because within the boundaries of their role it is objective enough. The problem lies in the jump from locally reasonable to organisationally sound. Companies make that jump far more casually than they should.

This is one reason the standard ethical framing is often too thin. Of course there are cases involving concealment, self-dealing, undisclosed relationships and outright abuse. Those cases matter. They are also the easiest ones to recognise, because they fit the familiar script. The more consequential organisational problem tends to be quieter. It lives in ordinary operating arrangements that allow one team to collect the benefit of a decision while another absorbs the cost, or that reward short-term visible gains while pushing longer-term damage somewhere less photogenic. Nobody needs to lie for that to happen. A spreadsheet and a bonus plan will often do the job perfectly well.

Take a simple example. Procurement is rewarded for reducing cost. Quality is judged on complaints, defects and reliability. Finance likes the savings this quarter. Operations will be less cheerful six months later if failure rates start creeping up. By then, the original choice may already have been presented internally as a disciplined commercial decision, complete with numbers, approvals and the solemn tone organisations adopt when they would prefer nobody ask a second question. At each step, the decision can look rational. Across the system, it can still be wrong. That is what built-in conflict of interest looks like in practice: no melodrama, no villain, just a trail of locally defensible choices that add up to collective distortion.

The same logic appears well beyond procurement. Sales functions are often rewarded for volume, sometimes with only faint regard for what happens after the contract is signed. Consultants can find entirely respectable reasons for extending work that ought to be narrowed or finished. Leaders are praised for defending their teams, right up to the point where cross-functional co-operation would have served the firm better. In each case, the actor can remain sincere. In each case, the organisation may still be teaching them to interpret success too narrowly.

This is why ethics training, on its own, rarely gets very far. There is nothing wrong with teaching principles, clarifying standards or explaining why conflicts of interest matter. The problem is that employees are usually excellent readers of the organisation’s real priorities. They learn from patterns, not slogans. They notice what gets promoted, what gets excused, which trade-offs are quietly tolerated, and whose definition of “good judgement” wins when departments collide. If the system continues to reward local optimisation, it will not help much to assemble people in a meeting room and remind them to think holistically. They will nod politely, take the biscuits, and return to the same incentive structure afterwards.

From a behavioural perspective, the more useful question is not simply whether people have integrity, though integrity obviously matters. The sharper question is where the organisation has built competing interests into everyday work without making the trade-offs explicit. Where does one function gain from a choice whose cost emerges later or elsewhere? Where do performance metrics reward behaviour that another department must later correct, absorb or explain away? Where have role definitions quietly turned partial interests into apparently neutral judgement? These are better diagnostic questions because they focus attention on the machinery that keeps producing the problem, rather than on the occasional individual who becomes visible enough to blame.

There is also a credibility issue here, and organisations tend to underestimate it. Employees notice when the formal language says “act in the company’s best interests” while the operating model rewards narrower behaviour. They notice when collaboration is praised and territorial optimisation is promoted. They notice when leaders speak about long-term value but celebrate short-term wins that merely move cost across internal boundaries. People are generally much quicker at reading these contradictions than senior management would like to believe. In many firms, the credibility problem begins long before the compliance problem does.

That is why conflict of interest inside organisations should be treated as a serious matter of design and judgement, not just a niche ethics topic for policy documents and annual refreshers. It sits at the intersection of incentives, attention, role-based interpretation and institutional habit. It shapes what people notice, how they justify, what they defend and what they genuinely fail to see. Once that becomes clear, a good deal of ordinary organisational behaviour looks different. The issue is no longer limited to the rare case where somebody knowingly crosses a line. It is present in the far more common situation where the line was blurred by the system long before the individual arrived at the decision.

If an organisation wants to take conflict of interest seriously, it has to look beyond declarations of principle and into the structure of everyday work. It has to ask where it is rewarding one part of the system for creating costs in another, where it is calling something “objective” that is actually role-bound, and where its own measures of success are quietly teaching people to become partial in highly professional ways.

That is a less comforting view than the usual morality play. It is also much closer to how organisations actually function.

Leave a comment