How do we reduce the risk of a critical appointment?

When a senior appointment goes wrong, the post-mortem almost never concludes that the person could not do the job. They usually could. What went wrong was the interaction between a capable individual and a specific organisational context: an executive committee that did not accept them, a mandate that was described differently in the interview than it turned out to be, a shareholder whose expectations nobody had made explicit.

Failure is a fit problem wearing a competence mask

The reason failure is usually attributed to competence is that competence is the only thing most processes actually measure. Interviews probe experience, achievements and technical judgement, because those are legible and comparable. Fit — the question of whether this person will be effective with these people under these constraints — is assessed impressionistically, by whether the panel liked them.

Liking a candidate is data, but it is weak data, and it is heavily biased towards people who present in a familiar register. A structured process does not eliminate that bias. It does force it to compete with evidence.

Interviewing is not assessment

An unstructured interview is a conversation in which both parties perform. The candidate has rehearsed their narrative; the panel asks whatever occurs to them; the notes afterwards record impressions rather than observations. Two panels interviewing the same candidate frequently reach opposite conclusions, which tells you the method is measuring the panel as much as the candidate.

Structured assessment means agreeing the success criteria before meeting anyone, asking every candidate the same behavioural questions against those criteria, recording specific evidence rather than impressions, and producing a written assessment that a director who was not in the room can evaluate.

Psychometric instruments have a place within this, but a narrow one. They are useful for generating hypotheses to test in interview and for identifying derailment risk under pressure. They are not a substitute for evidence of what someone has actually done.

References are usually designed to succeed

The standard reference process asks the candidate for names and then calls them. The candidate has selected people who will speak well of them, which they duly do. The exercise confirms what everyone already believed and adds nothing.

Useful references are the ones the candidate did not nominate: a former peer who competed with them for resources, a direct report who left during their tenure, a board member who saw them under genuine pressure. These conversations require a network to reach and judgement to interpret, and they are where most of the real information sits.

The questions matter as much as the source. "Was she effective?" produces nothing. "Describe a decision where you disagreed with her and how she handled it" produces something you can use.

Decision architecture

Even with good evidence, the final decision is frequently made badly — in a meeting where the most senior voice speaks first, or by consensus that conceals a serious reservation nobody wanted to be the one to raise.

Two mechanics help disproportionately. Ask each decision-maker to record their assessment independently, in writing, before the group discusses. And ask explicitly what would have to be true for this appointment to fail, and whether anyone has seen evidence pointing that way. Dissent that is invited is expressed; dissent that is not invited turns up eighteen months later.

Make the mandate honest

A significant proportion of appointment risk is created during the sell. Organisations describe the role as they wish it were: the mandate for change is oversold, the resistance is understated, the shareholder's real timeline is left vague.

Candidates who accept on that basis discover the difference in month three, and the appointment starts from a deficit of trust. Describing the difficulty accurately loses some candidates. The ones it loses are generally the ones who would have left.