Why senior hires fail in the first year

He was the easiest yes I have been part of.

The CV was excellent. He interviewed with real confidence, answered everything, and left every person in the process nodding. There was no debate afterwards. Nobody had a reservation to put on the table. We made the offer quickly because it felt obvious.

Within a few months it was clear something was wrong. Not the knowledge — the knowledge was exactly as advertised, and stayed that way. What never arrived was the part we had assumed came with the seniority. He would not decide anything without being told to. Given a problem with no clear brief, he waited. Given two imperfect options, he escalated rather than picked. Every technical thing we tested, he could do. The job was not a technical job.

What we actually measured

I have gone back over that process more than once.

We spent almost all of it establishing what he knew. Depth in his stack, systems he had worked on, decisions he had been near. He answered all of it well, because it was all true.

We spent almost none of it on how he worked. Nobody asked him about a time he had to move without enough information. Nobody made him choose between two things he wanted to keep, and then argue for the one he dropped. Nobody pushed back hard on an answer to see whether he would hold his position or fold. Nobody gave him something under-specified and watched what he did with the gaps.

Those are the things the job turned out to be made of. We did not check any of them, and then we were surprised.

Why "culture fit" is the wrong post-mortem

When a senior hire does not work out, the two explanations that show up are culture fit and onboarding. I have used both myself.

They are convenient because they are unfalsifiable. Nobody can prove a culture fit assessment would have caught it. Nobody can prove better onboarding would not have. Both point at something after the decision, which is comfortable, because the alternative is looking at the decision.

The honest version, in our case, was earlier and duller. We did not have a clear enough account of what the role actually required before we started interviewing for it, so we defaulted to interviewing for what is easy to interview for. Technical depth is easy to interview for. It is legible, it feels rigorous, and everyone in the room knows how to score it. Judgment under ambiguity is none of those things, so it quietly falls off the list.

Onboarding cannot fix that. Onboarding can shorten someone's ramp. It cannot install the thing you did not check for. If the wrong person is in the seat, a better first week produces a better-informed wrong person.

And the role itself is usually where it starts — not with the candidate at all.

The part that should have worried us

The unanimity.

Everyone agreed, immediately, with no dissent. At the time that read as confirmation. Looking back it reads as a group of people who had all evaluated the same dimension and found it strong, which is not the same as a group of people who had evaluated different dimensions and agreed.

A process where four interviewers ask four versions of the same question will always produce a confident answer. It will just be a confident answer about one thing.

I now treat a fast unanimous yes on a senior role as a reason to slow down and ask what nobody covered. Usually somebody, when asked directly, has a small unease they did not think was worth raising against the consensus. That is worth more than the fourth confirmation of technical depth.

It fails in the other direction too, and nobody notices

The uncomfortable half of this is that the same process turns good people away.

An interview weighted towards what is legible rewards a particular kind of candidate: quick, fluent, comfortable being assessed. Those are real qualities. They are also, largely, interview qualities. The person who thinks slowly in a room, asks three clarifying questions before answering, and is visibly unimpressive for the first twenty minutes is often the person who would have decided well on the actual job.

We never find out. A bad hire announces itself within months and gets discussed for years. A wrongly rejected candidate produces no evidence at all — they simply go and do good work somewhere else, and nobody in the room ever learns they were wrong. Every hiring process gets continuous feedback on one of its two failure modes and none on the other, which is exactly the condition under which a process stops improving and starts hardening.

What I would do differently

Decide what the role genuinely demands before writing a single question, and be specific about the parts that are not technical — how much of this job is deciding without enough information, how much is disagreeing with people more senior, how much is picking up work nobody assigned. Then build the process so each round establishes something different, and no two people are covering the same ground.

There is one question I would not run a senior process without now. Tell me about something you decided without asking anyone.

It sounds soft. It is not, because almost nobody has a prepared answer for it, and the follow-ups are where it does its work. What did you not know at the time? Who disagreed, and what did you do when they did? What would have changed your mind? Where did you tell people afterwards rather than before?

Someone who operates with ownership answers this immediately and specifically, usually with something that went partly wrong, because people who decide things regularly have a stock of decisions that did not fully land. Someone who waits for direction gives you a decision that was really an approval — they will describe consensus, or a choice that was obvious, or a moment where they recommended and somebody else decided. Nobody can perform their way through four follow-ups about a decision they did not actually own.

Our candidate, I am fairly sure, would not have got past the second one.

None of that is exotic. It is mostly just refusing to interview for what is convenient to interview for.

He was not a bad engineer. He was a good engineer we hired into a job that needed something else, on the strength of evidence that never touched the thing that mattered. That one is ours, not his.