When someone on your team isn't working out

You've been noticing it for a while. Work comes back late, or thin, or you find yourself quietly redoing pieces of it. You keep telling yourself the next project will be different, and you keep not saying anything, because saying something would make it real and you don't want to knock someone's confidence over a rough quarter.

This article covers how to name the problem early, and how to work out what's actually causing it. It also covers what a genuine improvement plan looks like, and how to run the conversation if it ends anyway. Read the diagnostic section before you decide anything, because a lot of what gets called underperformance turns out to be a manager who never said clearly what good looked like.

The management thinking here comes from Matt Mochary, the CEO coach behind the Mochary Method curriculum, who works with founders at Coinbase, OpenAI, and Reddit; Kim Scott, who managed teams at Google and Apple and wrote Radical Candor; Alisa Cohn, an executive coach who works with founders on exactly these conversations; Elizabeth Stone, chief technology officer at Netflix; Rachel Lockett, an executive coach who was previously at Stripe; and Molly Graham, who led teams at Facebook, Google, Quip, and the Chan Zuckerberg Initiative.

Say something the first week you notice

Most of the damage happens in the months between noticing a problem and saying anything about it. Kim Scott tells a story about a direct report she calls Bob, who she liked, and whose work wasn't good enough for months while she said nothing about it. She eventually realized her best people were going to leave because they were spending their time fixing his output. By the time she sat down with him, it was ten months too late.

Scott has a name for the thing that kept her quiet. She calls it ruinous empathy, which is caring about someone and letting that stop you from telling them something they'd be better off knowing (6:25). She thinks 90% of managers make 90% of their mistakes there.

So the first move is one specific sentence in your next one-on-one, about one specific thing you saw. "The spec you sent Tuesday didn't have the migration risk in it, and I had to add it before I could send it on. That's the third time this quarter I've done that." Say the observation without attaching a verdict about the person, ask them what happened, and then stay quiet long enough for them to actually answer. If you want the wording for that, the feedback playbook covers how to say it so it gets heard.

Work out whether it's clarity, capability, or motivation

The Mochary Method curriculum has a diagnostic for feedback that isn't changing anything, written by CEO coach Celine Teoh. She splits the cause into three, and each one has a completely different response. Getting this wrong is expensive. Coaching someone who's confused won't help them, and neither will training someone who's simply lost interest.

Start with clarity, because it's the most common and the one you control. The curriculum gives a test worth memorizing: a video camera should be able to capture the behavior you're asking for. "Please be more punctual" fails the test. "Please text me if you're going to be more than a minute late to our meetings" passes it. If what you've been asking for is "more ownership" or "better judgment" or "more visibility," you've been asking someone to read your mind and then marking them down for guessing wrong.

Molly Graham puts the same point in numbers from her own experience running teams through fast growth.

She describes the check as two-sided. You say what you're seeing, and then you ask them what they think their job is and what number they were hired to move. When those two answers don't match, you've found something you can fix without anyone's job being at stake.

If clarity is fine and it's capability, the questions are practical. Do they need training. Can the work be broken into smaller steps they can succeed at. Do they need money, time, or your political cover to get it done. The curriculum suggests an exercise borrowed from submarine captain David Marquet: ask the person to send you an "I intend to..." note before they act, so you can see their thinking and correct it cheaply before they've spent two weeks going the wrong way.

Motivation is the hardest to read and the one people jump to first, usually wrongly. The curriculum's advice is to stop treating a lack of change as resistance and go find out what the person actually cares about, using the idea of currencies from Allan Cohen and David Bradford: some people are moved by learning, some by recognition, some by autonomy, some by being trusted with something important. It also flags that feedback itself can feel like a threat, using David Rock's SCARF model, and that someone who's worried their status or their standing on the team is gone will spend their energy defending themselves instead of improving.

Check whether the role is the problem

Plenty of people who are failing in one job are excellent in a slightly different one. The role changed under them after a reorg. They were promoted into managing when they were brilliant at building. The company grew and the job they were hired to do stopped existing. None of that is a character problem, and treating it as one wastes someone who could still do good work for you.

Molly Graham is blunt about how often hiring simply misses.

If half of all hiring decisions are wrong, then someone struggling in front of you is often a sign that the match was bad from the start. That's worth holding onto before you decide anything about the person. The Mochary Method's note on firing well makes the manager own that directly: either you didn't evaluate well during recruiting, or you didn't onboard and train well, and either way part of this is yours.

If an internal move is realistic, raise it as a genuine option rather than a soft exit. Say what they're good at, say what this particular job needs, and let them decide whether the other role interests them.

Rachel Lockett coaches leaders through exactly this call, and brings a question from her time at Stripe that cuts through a lot of circling.

The follow-up in that exchange matters as much as the question. A no doesn't mean you fire them, it means you have to do something, and Lockett lists the options: talk to them about it, put them on a performance plan, or move them into a different role (88:33). She gives an example she sees often, where a company builds a leadership team and a year later the job has changed under someone, so a finance leader who was right at the earlier stage now belongs in a narrower role rather than out of the building.

Write an improvement plan the person could actually pass

A formal plan usually has a name at your company, and HR will own the template. What HR can't write for you is whether the plan is winnable. That's the test to apply to your own draft: if this person did every single thing on this list, would you genuinely be glad to keep them? If the honest answer is no, you're not running an improvement plan, and you should stop and have a different conversation instead.

The Mochary Method curriculum's note on motivation adds something counterintuitive that's worth doing during the plan. Celine Teoh calls it cheering for learning to walk: notice every small step in the right direction and say so, and hold off on new constructive feedback for a few days. Most managers running a plan do the opposite and pile on corrections, which teaches the person that nothing they do will register.

Escalate in steps, so the stakes are clear each time

Faith Meyer's three strikes process in the Mochary Method curriculum is the simplest structure for the months before any of this becomes formal. It exists to get conflict-averse managers out of their own heads and saying things at the moment they happen.

Each step exists to give the person new information they can act on. By the third one, nobody is guessing about where they stand, and that's the whole purpose of doing it in stages.

Nobody should be surprised

Everything above is in service of this. If the person is shocked when you tell them their job is at risk, you've managed them badly, whatever the paperwork says. Alisa Cohn puts the responsibility squarely on the manager.

She goes on to describe what she hears when she asks clients whether they've been crystal clear about what they need (32:03). They make a wobbling hand motion, which means no. Ask yourself the same question before you decide someone has failed to respond to feedback they may never have received.

Cohn scripts the conversation that has to happen before any termination conversation, and the wording is worth stealing almost word for word.

Look at what that script actually contains. The specific behavior, the fact that it's been raised before, how long it's been going on, a deadline, the consequence, and then two sentences of genuine belief in the person. Cohn's view is that this conversation is more valuable than the firing conversation, because it might mean you never need the second one (34:33).

Netflix runs a different model, and it's worth knowing about

Most companies handle this through an annual review cycle and a formal plan. Netflix deliberately doesn't, and their version is worth understanding properly, because people who've only heard the phrase repeat it badly. Elizabeth Stone, their chief technology officer, describes the question managers ask themselves.

People repeating this model miss two things. It's a question managers ask themselves, with no ranking or quota behind it. And employees ask it back: Stone says people regularly ask her "am I passing your keeper test?", which she treats as another way of asking how they're doing (33:43). If you'd like a low-drama way to have that exchange in a one-on-one, that phrasing is a good import even if the rest of the model isn't.

Stone is also clear that the test only works because of what comes before it.

Even at a company built around it, she says the ideal is a mutual observation and admits it often isn't. That's a fair standard to hold yourself to: by the end, the two of you should broadly agree about what happened, even if only one of you wanted this outcome.

Running the conversation when the answer is no

Your HR partner will own the logistics, the timing, and the paperwork, and you should follow their instructions on all of it. What follows is the shape of what you say. The Mochary Method's note on firing well lays out the sequence, and the ordering matters more than the wording.

Managers skip that last step because they want the meeting over, and it does more than anything else to shape how the person remembers the whole thing. You don't have to fix their feelings or agree with their version of events. Listen properly, and show them you heard it.

Matt Mochary describes what the offer of help sounds like when he does it himself.

He gives himself a test for whether it worked, which is that he's stayed friends with everyone he's let go (21:04). Whether or not you'd want that, it's a useful thing to picture before you plan the conversation.

On the help you offer, the curriculum goes further than most companies do. It suggests severance of two months at an absolute minimum and realistically three or more, and that you actively work your network on the person's behalf: write a description of their strengths, post it with your endorsement if they're comfortable with that, and offer to send a recommendation to anyone who gets them past a first round. It also predicts the feeling that gets in the way, which is anger, and the urge to not give this person another penny. Its advice is to feel it and let it pass before you decide anything.

What to say to the team afterward

Your team will read how you treat someone on their way out as information about how you'd treat them. That's the real audience for the announcement, and the Mochary Method note is direct about how to handle it: announce the departure, and do not blame or criticize the person. Praise what they contributed, and take ownership yourself for not having matched their skills to what the company needed.

Expect the question "should I be worried?" from at least one person. Answer it honestly. If they're doing well, say so plainly and give an example. If someone else on the team is also struggling, they should already know that from you, which brings you back to the no-surprises rule.

Keep confidences you agreed to keep. People will ask for details, and the details belong to the person who left.

Looking after yourself through it

This is one of the harder things you do as a manager, and it's normal to sleep badly, to draft the conversation twenty times, and to feel guilty for months. Molly Graham describes a test she learned from her own manager for when she's stuck on a decision like this.

Her point is that the decision and the delivery are two separate problems, and mixing them is what keeps managers stuck for months. Work out what's right when you take the emotion out, then put all your care into how you deliver it.

Frequently asked questions

How long should I give someone before deciding?

Long enough for the change you're asking for to be visible in real work, which for most roles means a quarter rather than a month. Kim Scott's regret was ten months of silence. The fix for that is to name the problem earlier, and then still give a fair amount of time once you have.

Should I tell someone their job is at risk?

Yes, once the situation has genuinely reached that point, and check with HR on how your company words it. Someone who doesn't know the stakes can't make an informed choice about how hard to work at this or whether to start looking elsewhere. Faith Meyer's escalation process puts that sentence at the third step for exactly this reason.

What if the problem is that I don't enjoy working with them?

Separate the two. Write down the specific work outcomes that aren't happening. If you can't fill that list, the problem is a working relationship, and that's a different conversation, often about how you two communicate. If you can fill it, work the outcomes and leave your personal feelings out of the wording.

They've improved, but only a little. What now?

Go back to your own test: if they held this new level, would you be glad to have them? If yes, say so and close out the plan. If no, say that plainly now rather than extending quietly, because an extension you never explain tells them the bar wasn't real.

What if my own manager is pushing me to move faster than I think is fair?

Ask what's driving the timeline, since it's often budget or a reorg they haven't told you about. Then say what you need: the person hasn't had a clear ask yet, or hasn't had the support attached, and you want a specific amount of time. Bring a plan and a date rather than a request to wait.

Should I write everything down?

Keep notes of what you agreed in each conversation and share them with the person, so you both have the same version. Do that because shared notes make the next conversation honest, and because someone can't act on feedback they've misremembered. Your HR partner will tell you what your company additionally requires.

The checklist

If you're heading into a review cycle with someone whose performance is the open question, the performance review playbook covers the meeting itself. And if part of what you're dreading is telling your own leadership that a team commitment is going to slip because of this, the playbook on delivering that news has the structure.