The first sign that a human resources function has been captured is rarely a scandal. In a decade of practice across Bangladesh and Australia, in workforces ranging from a few hundred staff to several thousand, the most reliable early signal I have found is a phrase. Some version of "we need to be commercial about this", or "we need to think about what is good for the business". The phrase is unobjectionable on its own. Spoken in the wrong meeting at the wrong time, it is the sound of a control function dissolving into a service function, and a profession quietly handing back its licence to operate.

I have been in too many of those meetings, in different rooms in different cities. The phrase is almost always uttered with something close to relief, as though a difficult question has finally been disposed of. The grievance against the favoured executive becomes a "performance conversation". The senior manager who needs to be moved on becomes a "settlement to keep things clean". The engagement survey result that should have triggered a board paper becomes a slide with three softening qualifications. None of this looks, at the time, like a failure of governance. It looks like a function doing its job in a complicated world. That is precisely the problem.

A control function, not a service function

The argument I want to make is structural before it is moral. Human resources, properly understood, is a control function. It belongs in the same family as internal audit, compliance, risk, and the company secretariat. Each of these functions exists because the modern organisation, and especially the modern regulated bank, is too complex, too fast-moving, and too prone to concentrated power for the executive layer to be trusted to police itself. Independence is the architecture that makes the corporation governable. Without it, the unitary board, the audit committee, the regulator, and the shareholder are all reading scripts written by the people they are meant to be supervising.

The Cadbury Report in 1992 made this point about the relationship between auditor and management. Higgs in 2003 extended it to non-executive directors. The OECD Principles of Corporate Governance, in their successive revisions, have applied it to every internal control function in turn. The post-2008 banking reforms, including the Senior Managers and Certification Regime in the United Kingdom and the equivalent regimes that followed in Europe, the United States, and parts of Asia, made it explicit that control functions sit inside the architecture of governance, not inside the operating line they oversee. The Three Lines of Defence model, articulated by the Institute of Internal Auditors and now embedded in the supervisory expectations of regulators including Bangladesh Bank, is the working vocabulary of any serious risk and control conversation. Where human resources sits within those lines is a question the profession has, on the whole, declined to answer with conviction.

The reluctance is partly self-inflicted. Dave Ulrich's business partner model, published in 1997 and refined repeatedly since, was a serious attempt to give HR strategic weight. In its purest form, the model assumed a partnership of equals, with a shared commitment to the long-term health of the organisation. In its degraded form, which is the form most commonly observed in practice, it has produced a generation of HR leaders who confuse proximity with partnership and access with authority. Patrick Wright's research at Cornell on the chief human resources officer has documented this drift carefully. The CHRO who is in the room is not the same as the CHRO who is heard. The CHRO who is heard is not the same as the CHRO who is independent. The conflation of the three is one of the great unforced errors of the modern people profession.

The mechanics of capture

An alliance with a powerful executive is rarely the result of a transaction. It is the result of a thousand small accommodations. The chief executive complains about a difficult senior manager and the head of HR, wanting to be useful, suggests a way to manage the person out without a formal process. The chief operating officer needs to make an appointment quickly and the head of HR, wanting to be commercial, agrees to compress the recruitment process and skip the panel. A complaint comes in against a favoured executive and the head of HR, wanting to be sensible, treats it as a performance conversation rather than a grievance. None of these moments looks, at the time, like a betrayal of the function. Each one looks like good judgment under pressure. The trade is small. The price is paid in instalments.

The pressures that produce these accommodations are familiar to anyone who has done the job. The HR leader is structurally weaker than the executives he or she is meant to supervise. The chief executive has hire and fire authority over the role in most jurisdictions, and even where the appointment is formally a board matter, the practical day-to-day is mediated by the executive layer. The HR leader who consistently says no to the chief executive is, in most organisations, on a clock. The HR leader who consistently says yes is on a different clock, but few people see it ticking, including, often, the HR leader.

This is the principal-agent problem in its purest form. Agency theory tells us that any agent whose monitoring depends on the cooperation of the principal being monitored will, in the absence of structural protections, drift towards the principal's interests. The drift is rational. It is also corrosive. The whole point of designing independent control functions is to remove the question of personal courage from the equation. A profession that depends on individual heroism for its integrity is a profession that has not been properly engineered.

What captured HR looks like from the inside

Inside the organisation, a captured human resources function performs a particular pantomime. The processes are still there. The grievance procedure is published. The disciplinary policy is on the intranet. The talent review cycle runs on schedule. The engagement survey is administered every year. From the outside, and to the regulator, the architecture looks intact. What has changed is what the architecture is for. The grievance procedure is now a mechanism for documenting why a complaint did not need to be upheld. The disciplinary policy is applied selectively, with discretion concentrated in the hands of the favoured executive. The talent review identifies the people the executive already wanted to promote. The engagement survey is interpreted with the kind of narrative care that a defence lawyer brings to inconvenient evidence.

In the disciplinary cases I have managed over the years, more than a hundred of them now, the strongest predictor of a clean outcome has not been the strength of the evidence, the seniority of the parties, or the complexity of the allegations. It has been whether the process was insulated from informal executive pressure from the start. The cases that produce zero successful legal challenges are not the easy ones. They are the ones in which the inquiry committee was constituted properly, the show cause was drafted on the evidence rather than the politics, and the recommendation went to the disciplinary authority on its merits. The cases that produce later trouble, including the ones that arrive years afterwards in the form of a writ petition or a regulatory question, are almost always the ones where someone, at some point, suggested that the matter could be handled more sensibly than the policy required.

The case literature beyond Bangladesh tells the same story at much larger scale. The Wells Fargo cross-selling scandal, which culminated in the regulatory settlements of 2016 and the years that followed, was not in the first instance a sales failure. It was a control failure, and a significant part of that control failure ran through the people function. Internal complaints about the unrealistic targets and the fraudulent account openings were treated, repeatedly, as performance issues for the complainants rather than systemic issues for the firm. The HR processes were intact. They were also, in effect, an enforcement mechanism for the very pressures that produced the misconduct.

The collapse of the Travis Kalanick era at Uber, triggered in part by Susan Fowler's blog post in February 2017, exposed an HR function that had become, by all credible accounts, a fortification around the senior leadership rather than a check on it. The pattern at WeWork in the months before the failed initial public offering of 2019 told a similar story, with people processes bent around the founder rather than applied to him. Theranos, in the testimony that emerged at the trial of Elizabeth Holmes, presented an even starker picture, with non-disclosure agreements and selective application of process used to suppress legitimate technical and ethical dissent. None of these organisations had no HR function. All of them had captured ones.

The academic literature anticipates these failures with some precision. Morrison and Milliken's work on organisational silence describes the conditions under which employees stop speaking up, and a captured HR function is one of the principal mechanisms by which those conditions are produced. Amy Edmondson's research on psychological safety makes the same point from the other direction, arguing that the willingness to raise concerns depends on the perception that the system will respond fairly. Charlotte Rayner and others working on workplace bullying have shown, repeatedly, that the role of HR in either enabling or interrupting persistent mistreatment is decisive, and that the variable is almost never the policy on the page. It is the independence of the people applying it.

The career arithmetic

The personal arithmetic is the part that rarely gets talked about, because it sits adjacent to a question most HR leaders prefer not to ask out loud, which is what they think they are doing the job for. In the short term, the alliance pays. The HR leader is invited to the off-sites. The bonus is generous. The reputation inside the executive bubble is one of being commercial, pragmatic, a safe pair of hands. The HR leader who maintains independence is, in the same period, having a harder time. He or she is on the receiving end of the chief executive's irritation. The bonus is smaller. The reputation in the executive bubble is one of being difficult, slow, a blocker.

Then the cycle turns. The chief executive leaves, often abruptly. Sometimes there is a scandal. Sometimes there is a regulator. Sometimes there is simply a new chair with a different view of the organisation. The HR leader who was the trusted ally is, suddenly, exposed. The new regime looks at the documentation, or at the absence of it, and asks why the grievance from two years ago was not properly investigated, why the senior manager who left received an enhanced settlement, why the engagement scores were presented to the board in their reassuring form rather than their disturbing one. The protection that came with the alliance evaporates with the executive who provided it. The HR leader who maintained independence, by contrast, finds that the new regime wants exactly what he or she has been quietly building. Independence, which was a liability under the old chief executive, becomes the most valuable thing on the curriculum vitae.

Over a five to ten year horizon, in my observation, the captured HR leader follows one of two trajectories. Either the pattern is detected, in which case the leader becomes unemployable in serious organisations and drifts down the market into firms with weaker governance, or the pattern is not detected and the leader retires comfortably, having presided over harm that was never properly attributed. The independent HR leader, by contrast, accumulates something more durable, which is a reputation among chairs, non-executive directors, and regulators that compounds over a career. The market for genuinely independent CHROs is, and will remain, far smaller than the supply of competent ones. That scarcity is the career arithmetic, and it favours, decisively, the path that looks harder in the short run.

What independence looks like in practice

Independence is not a stance. It is an architecture, supported by a discipline, expressed in a tone. The architecture is straightforward in principle and frequently neglected in practice. The CHRO should report on a dotted line to the chair of the nomination and remuneration committee, with direct, unmediated, and recorded access to the chair of the board and the chair of the audit committee. The performance review of the CHRO should not sit entirely with the chief executive. The compensation of the CHRO should not be set entirely by the executive whose conduct the CHRO may, one day, need to investigate. None of this is novel. It is the standard architecture for the chief audit executive in most listed companies, and increasingly the expectation for the chief risk officer in any regulated bank. The argument is simply that the CHRO requires comparable architecture, because the CHRO performs a comparable function.

Beyond the architecture there are the disciplines. Documentation, observed without exception. Every meaningful conversation about a senior individual is recorded in a contemporaneous note, filed properly, retrievable years later. In the disciplinary world, this is not a stylistic preference. It is the difference between a finding that holds and a finding that collapses on appeal. Process integrity, also observed without exception. The grievance procedure is run the same way for the protected executive as it is run for the junior administrator, and any deviation is itself documented and explained. Board access, treated as a recurring matter and not as a favour. The CHRO presents to the board on culture and people risk on a defined cadence, in person, without the chief executive mediating the message. The annual culture and conduct paper to the board is the CHRO's single most important deliverable, and it is written for the board, not vetted by the executive.

The tone is the part that cannot be written into a charter. The HR leader who, in his or her first encounter with a powerful executive, signals that the executive will be treated like everyone else, has set the terms for the rest of the tenure. The HR leader who signals the opposite, even once, has spent the rest of the tenure trying to claw back ground. For the newly appointed HR leader, the first ninety days are decisive. Three things must happen in that period. The reporting line and board access must be put in writing, ideally as part of the appointment terms but if not then as an early-tenure correction. One process must be run, or be seen to be run, at full integrity, even if the case itself is small. And there must be one private, calm conversation with the chief executive, the substance of which is that the partnership will be real, but it will not be an alliance. The distinction between those two words is, in the end, the whole of the job.

A return to the phrase

The phrase that signals captured HR will not stop being said. It is, in fact, a useful phrase. There are moments when commercial judgment and the people function genuinely need to align, and the HR leader who refuses to think commercially is just as much of a liability as the one who has been captured. The difference is how the phrase is used. In an organisation where HR is independent, "we need to be commercial about this" is the start of a conversation. In an organisation where HR has been captured, it is the end of one.

The job of the human resources leader, properly understood, is not to be liked by the executives, and not to be feared by the workforce. It is to be the function the organisation can rely on when it most needs to be told something it does not want to hear. The alliance with a powerful executive looks, from inside, like the path of least resistance. From the outside, and over time, it looks like exactly what it is, which is the slow surrender of a profession's reason for existing. There is nothing pragmatic about it. There is only the cost, paid first by the people the function was meant to protect, and last by the leader who agreed, one small accommodation at a time, to stop protecting them.