The Question That Tells You It's Time to Let Someone Go

A founder told me last year that she'd been saying the same sentence about one engineer for eighteen months: "Let's give it one more quarter." Four quarters later, she was still saying it. The engineer wasn't malicious, wasn't lazy, wasn't even obviously bad. He just wasn't getting better, and everyone around him had quietly adjusted their expectations of what he'd ship, who would review his code twice, and who would pick up the parts he'd drop.

That adjustment is the tell. When a team reorganizes itself around someone's limitations without anyone deciding to, the decision has already been made. It's just being made by everyone except the manager.

The delay isn't kindness

Managers tell themselves the wait is generous. It's usually avoidance wearing generosity's clothes. Firing someone is unpleasant, admitting a hiring decision was wrong is worse, and "one more quarter" costs the manager nothing today. It costs the team something every day between now and then: the best people notice that mediocre work gets protected, and they either lower their own bar to match or start looking for a place where it doesn't.

The honest version of kindness is a fast, clear answer — even a hard one — instead of a slow, ambiguous one.

The wrong questions

Most managers ask themselves questions that keep them stuck:

• "Are they trying hard?" Effort isn't the same as output, and rewarding effort over results teaches everyone the wrong lesson.
• "Have they been here a long time?" Tenure creates loyalty in the manager, not competence in the employee.
• "Do I like them?" Likability makes the conversation harder to have and easier to avoid — which is exactly why it shouldn't be in the decision.
• "Are they bad enough to fire?" This sets the bar at catastrophe. You end up waiting for a fireable offense instead of noticing a non-viable trajectory.

The right question

Ask this instead: would I hire this person again today, for this role, knowing everything I know now?

Not "is this person a bad person" — most aren't. Not "could they succeed somewhere else" — probably. The question is narrower and more useful: given their actual output, their actual trajectory, and the actual role in front of you, would you make this hire again with today's information.

If the honest answer is no, you already have your answer about staying. The only decision left is timing and process.

What people confuse for signal

A few things feel like evidence but aren't:

• A single bad quarter. Everyone has one. Look at trajectory across two or three cycles, not a snapshot.
• A personality clash with one manager. Move them, don't fire them, until you've ruled this out.
• A skills gap that's genuinely closeable in the time you have. A junior engineer who's slow at systems design in month four is not the same problem as someone who's slow at systems design in month sixteen.

And a few things are signal even when they don't feel dramatic:

• Other people start doing this person's job without being asked, and don't complain about it — they've just absorbed it as normal.
• The person's 1:1s are full of explanations for why work didn't happen, and the explanations keep changing.
• You find yourself giving them easier problems, not harder ones, to protect the roadmap.

That last one is the clearest tell of all. If you're managing someone by giving them less responsibility over time, you've already answered the keeper question. You just haven't said it out loud.

If the answer is genuinely unclear

Sometimes it really is unclear — the person is new to the level, the role changed underneath them, or you haven't given clear enough feedback to know if they can respond to it. In that case, run a real improvement plan, not a theatrical one. A real one has:

1. Two or three specific, observable behaviors to change — not "be more proactive," but "flag blockers within 24 hours instead of at the weekly sync."
2. A fixed timeline, typically four to six weeks, agreed on in the room, not extended informally later.
3. Weekly check-ins where you say plainly whether you're seeing the change, not just whether you had a nice conversation.
4. A decision at the end that actually gets made — stay, with the plan closed out, or go.

The theatrical version has none of these. It has a vague document, no real check-ins, and an ending date that quietly slides. Everyone involved knows the outcome already; the plan just delays saying it. If you're writing a PIP mainly to protect yourself legally rather than to genuinely find out if someone can turn a corner, be honest with yourself about that before you start.

Saying it out loud

Once the answer is no, the hardest part isn't the decision — it's the conversation. Don't soften it into ambiguity ("this isn't really working out for either of us") that leaves the person confused about what happened. Say what you observed, over what period, and what it means. People rarely fall apart at getting bad news clearly. They fall apart at getting bad news vaguely, because it robs them of the ability to do anything about it — including finding a role where they'd actually succeed.

The manager who told me about her eighteen-month engineer eventually let him go. Her regret wasn't the decision. It was that she'd known the answer to the real question by month six, and spent a year telling herself a different, easier question instead.