Managing out, humanely and quickly

The decision takes as long as it takes. What happens after it should take days.

That's the whole position, and the two halves are separate on purpose.

Getting to the decision honestly is slow work. You have to be sure it isn't a management failure. You have to have told the person clearly. You have to have given a real chance to change.

None of it should be rushed. Most of it usually is.

Once you know, though, the humane version is fast.

Speed isn't kind in the abstract. It's that the alternative, the six-week drift where you've decided and they haven't been told and the team can tell, is the cruellest arrangement available, and it's the one most managers land on by default.

Fast only works on top of honest

The speed argument carries a precondition I held for years without ever saying it out loud. It took getting the whole thing badly wrong to notice.

The best exit I've run didn't feel like an exit at all.

There had been honest conversations well before it. The kind where you say we're in this together, we're both looking at the same things.

By the time we reached the end, the conclusion was jointly held. A ceremony. None of it surprised either of us.

The worst one went like this.

Expectations I held and never stated. Hard feedback conversations I shied away from. The conversations I did have, undocumented.

Then avoiding the person, which is the step that tells you exactly what's happening if you're willing to look straight at it.

And at the end of all that, nothing left except a no-cause package, which came as a great surprise to everybody. HR. Legal. Finance. The individual. The team.

Look at what the avoidance actually did there. It removed my options.

With no record and no stated expectations, managing somebody out for cause isn't available to you, so the only instrument left is money.

Avoidance converts a performance conversation into a financial transaction. Quietly. Over months. While you tell yourself you're being kind.

So fast plus honest is clean. Fast plus undocumented is arbitrary, and what a team learns from arbitrary is that anybody can go without warning.

The claim that the team already knows holds when you've been talking. When you haven't, they don't, and your speed reads as a disappearance instead of a conclusion.

Three different problems that look identical

Work out which of these you actually have before anything else. The response differs for each, and getting it wrong is expensive.

A performance problem. The person can't do the work at the level required, and coaching hasn't moved it.

Real, and much rarer than it gets diagnosed.

A fit problem. The person is genuinely capable, and this is the wrong role, the wrong stage, or the wrong kind of company.

The excellent big-company engineer at a fifteen-person startup with no platform. The deep specialist in a role that needs range.

Nobody is failing here. Telling somebody they're underperforming when the truth is that the job is wrong for them is both false and needlessly damaging.

A management problem. Unclear expectations, no feedback, a project set up to fail, or a manager who decided in month one and stopped investing.

If you can't confidently rule this one out, you haven't earned the right to move to a decision.

The honest test: can you point to a specific conversation, on a specific date, where you told them plainly what needed to change. If that conversation doesn't exist, your first move is to go and have it.

Most of the situations that arrive on my desk as performance problems are one of the other two.

The conversation, once

When you've decided, the conversation is short and it doesn't hedge.

Say the decision in the first thirty seconds.

No context. No warm-up. No five minutes of praise before the turn, which everybody recognises and which makes the praise itself feel like a manipulation.

"I've made a decision about your role here, and I want to tell you what it is before anything else."

Then the reason. Once, in plain terms, without a list. One clear sentence about the gap.

A pile of evidence sounds like a prosecution and invites a debate about each item. You aren't there to win a debate. The decision is made.

Then stop talking.

The silence is unbearable and you have to let it happen. What comes next belongs to them. Anger, questions, sometimes relief. Quite often relief that surprises both of you.

Then the practical part, all of it worked out before you walked in. The date, the money, references, what gets said to the team, what they'd like said.

Having those answers ready is a form of respect. Improvising them signals that you decided this morning, or that they weren't worth the preparation.

One conversation.

The common alternative is a series of increasingly pointed hints across a month, which leaves somebody re-reading their own performance for weeks, trying to work out what's happening to them.

PIPs: honest and dishonest versions

A performance improvement plan is honest when there is a genuine chance the person keeps their job, when what "improved" means is specific and observable, and when the manager is actively invested in them succeeding.

Those exist. I've watched them work, mostly where the underlying issue was a skill gap with clear edges and the person genuinely wanted to close it.

The dishonest version is a documentation exercise. The decision is already made, the plan exists to create a paper trail, and everybody in the process except the person on the plan knows how it ends.

Common. Advised by lawyers. A spectacularly bad trade.

You buy some protection against a claim, and you pay for it with sixty days of somebody's life spent trying to save a job that no longer exists, plus whatever the rest of the team concludes once they see how the process works here.

If you're in a jurisdiction or a company where the process is mandatory, run it, and be honest inside it.

Tell the person where they actually stand. "I have to run this process, and I want you to know my honest read on the odds." That sentence costs you something. Worth it.

What you can't do is use the process to avoid forming a view.

If you genuinely don't know whether you'd keep them at the end, that's a real PIP. If you do know, the PIP is theatre, and you should own the decision instead of hiding behind a form.

What the team learns

Every departure teaches your team something, and they learn it from the timing and the manner. Never from what you announce.

They already know.

This is the part managers consistently get wrong. You're almost never the first to see it.

By the time you're weighing the decision, the person's peers have been covering, redoing, or working around for months, and your delay reads to them as either not noticing (you aren't paying attention) or not acting (the standard is negotiable).

Both are corrosive. The second is worse.

They watch how the person gets treated on the way out.

Dignity here is the single most-watched signal about what kind of place this is. Your strongest people are quietly asking what it would look like if it were them, and the answer decides how safe they feel taking a risk that might not work out.

And they notice what you say afterwards.

The correct amount of detail is very little. They've left, here's how the work is covered, I'm not going to discuss the specifics.

Anything more trades somebody's privacy for making yourself look reasonable, and the team registers that trade exactly as it is.

The blameless thing, and why it isn't a contradiction

I run mandatory blameless postmortems for every incident. I also move fast on people who aren't working out.

Those look inconsistent and they aren't. Worth stating the distinction plainly, because this is where "blameless culture" gets misread into paralysis.

Blameless is a claim about incidents.

When a system fails, hunting for the person who typed the command destroys the information you need and teaches everybody to hide. The system produced the outcome, and the system is what you fix.

It says nothing about individual performance being a system property, and nothing about standards being a form of blame.

A person who consistently can't do the job is a fact about a person. Treating that as a systemic mystery is a way of never having to say anything hard to anybody.

The connection between the two is honesty about causes. Same discipline, applied to a person instead of to a service.

The week after

Two things, both easy to skip.

Talk to the team about load before they ask. The work didn't disappear, and somebody is carrying it now, usually whoever was already covering informally.

And do your own postmortem, in writing, for yourself.

When did you first know. What did you tell yourself in the months between. What was the delay actually made of.

That document is worth more than any book on management, and nobody else ever needs to read it.