How do I run a retrospective after a hire goes wrong?

After a departure inside the first year, after a hire the team clearly regrets, after losing a finalist you badly wanted, and periodically for roles you hire repeatedly. Timing matters. Too early and you are reacting to a rough first month that many strong hires have. Too late and nobody remembers the evidence. A few weeks after the outcome is confirmed is usually right. Keep the session to forty-five minutes with the people who actually took part: the hiring manager, the interviewers and the recruiter. This is a working review of a process, not a performance discussion about a person who is no longer there to explain themselves. Book it as a standing item for the role families you hire repeatedly, so it happens without anyone having to call for it.

The written record, not recollections. The scorecards from every round, pulled from your applicant tracking system with the ratings and evidence attached. The original intake notes and requirements. The sourcing data and where the person came from. The debrief record and how the decision was reached, including any dissent that was overruled. Then the outcome: what actually went wrong, in behavioural terms. Read the scorecards before discussing them. A pattern shows up strikingly often, where one competency was rated on thin evidence, or a reservation appears in someone's notes and never made it into the conversation. Retrospectives run from memory converge on whoever is least popular, which teaches the group nothing. Circulate the documents a day ahead so people arrive having read them rather than reconstructing events live.

Four, asked in order. Did we assess the right things, meaning did the scorecard describe the job as it turned out to be. Did we get real evidence for each competency, or did we accept a story. Where did the signal first go wrong, and was it visible at the time. And what would have caught it: a different question, a work sample, an extra reference, a different interviewer. Then commit to one change and write down who owns it. One change that survives is worth more than a list of six that nobody implements. Update the scorecard template or the structured interview questions immediately, while the reasoning is fresh.

Separate the process from the people out loud at the start, then hold that line. The question is where the system produced a wrong signal, not who was fooled. That framing is what makes interviewers admit they had a doubt and did not raise it, which is usually the most valuable sentence in the room. Honesty also means naming causes outside recruiting: an unclear role, a manager who had no time to onboard, a team change after the offer, or a compensation constraint that limited the pool. Many bad hires are onboarding or role-definition failures rather than selection failures. Recording that is more useful than adding another interview round. Name that cause explicitly in the write-up, because a finding recorded in the wrong place never gets fixed.