Managing an Underperformer

Leading Change Without Losing People was about moving a whole team through something hard. This page zooms in on one of the loneliest situations a manager faces: a single person who isn’t performing, and the quiet dread of what to do about it. Most managers handle it badly — not because they’re cruel, but because they wait too long, then act too fast, and skip the one step that actually matters: figuring out why.

This page gives you a way to think clearly instead of anxiously. You’ll learn to diagnose the real cause before reacting, to open the conversation honestly instead of vaguely, to build an improvement plan that’s fair to everyone, and to tell the difference between someone turning it around and someone quietly stalling. And you’ll learn that a humane exit — done early and with dignity — is not a failure. It’s sometimes the kindest outcome for everybody, including the person leaving.

First, a definition that keeps you honest

“Underperformance” means a real, sustained gap between what the role needs and what the person is delivering — not that they annoy you, not that they work differently from you, not that they had one bad month. Before you do anything else, force yourself to write the gap down in plain, specific terms.

Vague (useless):   "Sam isn't pulling their weight."
Specific (usable): "Sam's tickets take about 3x longer than the team
                    norm, and the last two shipped with errors a peer
                    caught, not Sam."

If you can’t write the specific version, you don’t yet have an underperformance problem — you have a feeling, and acting on a feeling is how good people get pushed out unfairly and morale quietly rots.

The three root causes: skill, will, or situation

Almost every case of underperformance traces back to one of three roots. The mistake nearly everyone makes is assuming the cause they feel — usually “they don’t care enough” — without checking. Each root needs a completely different fix, so guessing wrong wastes months.

Skill — they can’t (yet)

The person wants to do well but lacks a specific ability, knowledge, or experience the role requires. A nurse new to a ward doesn’t yet know its protocols. A support agent hasn’t learned the product deeply enough to answer without escalating. This is the most fixable root, because skill is teachable — through training, mentoring, pairing, or simply time.

Will — they won’t (right now)

The person can do the work but isn’t. This is the root people jump to first and understand worst. “Won’t” is rarely laziness. It’s usually demotivation with a cause: they feel unrecognized, they’ve lost belief in the mission, they’re burned out, they resent a decision, or something outside work is consuming them. Will problems are fixed by understanding the cause — which means a real conversation, not a lecture about attitude.

Situation — the setup is broken

The person could and would perform, but the environment won’t let them. This is the root managers miss most, because the fix points back at themselves. Common situational causes:

• Unclear expectations. They don’t actually know what “good” looks like, because no one defined it.
• Bad fit. A brilliant analyst put into a client-facing sales role. The skill is real; it’s aimed at the wrong target.
• Blocked resources. They’re waiting on tools, access, information, or a decision only you can unblock.

Symptom you see:          Likely root:            Fix that works:
"They can't do it well"   Skill                   Teach / pair / time
"They're not trying"      Will                    Understand the why
"It never quite works"    Situation               Fix the setup (often yours)

The common trap — labeling a situation problem a will problem

The single most expensive mistake here is reading a situation failure as a will failure. You conclude someone “doesn’t care” when in truth you never told them what you wanted, or you starved them of a resource, or you put a square peg in a round hole. You then manage them out — and repeat the exact same setup with their replacement, who fails the same way. Before you ever attribute poor performance to the person, prove it isn’t the setup. Ask: Have I made the target unmistakably clear? Do they have what they need? Is this even the right seat for them?

The evidence — attribution error

Social psychologists have long described the fundamental attribution error: we tend to explain other people’s failures by their character (“lazy,” “careless”) while explaining our own by circumstance (“I was swamped”). Managers are not immune — arguably they’re more exposed, because a situation problem often implicates their own choices, which is uncomfortable to face. Naming this bias out loud is a practical safeguard: it forces the situational check you’d otherwise skip.

Your first move: the direct, specific, non-accusatory conversation

Once you’ve written down the gap and honestly considered the three roots, your first action is a conversation — not a plan, not a warning, not a complaint to your own boss. And it is a particular kind of conversation.

Most managers get this wrong in one of two directions. They go vague and soft — “I just need you to step up a bit, yeah?” — which the person can’t act on because it names nothing. Or they go hard and accusatory — “Your attitude is a problem” — which triggers defense and shuts down the truth you need.

The move that works is direct about the gap, specific with evidence, and generous about the cause.

Not this:  "You need to step up."   (vague — nothing to act on)
Not this:  "You clearly don't care." (accusatory — triggers defense)

This:      "I want to talk about something I've noticed, and I want to
            understand what's going on. The last two reports were a week
            late and each had errors Priya caught before they went out.
            That's a change from earlier this year. What's happening?"

Notice the shape. You state the specific, observable gap with evidence. You avoid character words (“careless,” “lazy”) entirely. And you end on a genuine question — because you have a diagnosis to make, and the person almost always holds the missing piece. Maybe they say “I’ve been drowning since we lost Dev and never said anything” (situation). Maybe “I honestly don’t know how to do the new reconciliation step” (skill). Maybe they go quiet and eventually admit they’ve checked out since being passed over for a promotion (will). Each answer sends you down a different road. If you’d opened with an accusation, you’d have gotten none of it.

In practice — the question that changed the diagnosis

A cafe manager was ready to write off a barista as slow and indifferent — orders backed up on her shifts, customers grumbled. Instead of a warning, the manager sat down and asked what her shifts felt like from the inside. It turned out the espresso machine on that side had a sticky group head that took twice as long to lock in, and she’d assumed everyone knew and just dealt with it. A twenty-minute repair fixed the “performance problem.” The manager had been one accusation away from losing a perfectly good employee to a hardware fault.

Building an honest improvement plan

If the conversation confirms a real gap that the person needs to close, you move to a plan. This is where being kind and being clear stop being in tension — the clearest plan is the kindest one, because it gives the person a genuine, fair shot instead of a fog they can’t navigate.

A real improvement plan has four parts. Skip any one and it collapses into either vague pressure or false hope.

1. Concrete, measurable expectations. Not “improve quality” but “reports go out on the due date, with no more than the occasional minor error, for the next six weeks.” The person must be able to tell, on their own, whether they’re meeting the bar.
2. A timeline. A defined window — often four to eight weeks depending on the work — long enough to show real change, short enough that it doesn’t drag. “Ongoing” is not a timeline; it’s a slow-motion trap.
3. The support you’re offering. State what you will do: pairing them with a strong peer, weekly check-ins, removing a competing priority, funding a course, unblocking a resource. A plan that demands change while offering nothing is a setup for failure, and everyone can smell it.
4. What happens if it doesn’t improve. Say it plainly and kindly: “If we get to the end of this and we’re not there, then this probably isn’t the right role, and we’ll talk about what’s next.” Withholding the stakes isn’t mercy — it robs the person of the information they need to decide how hard to fight for the role.

IMPROVEMENT PLAN — one page, shared, no surprises

Gap:          Reports late and error-prone since March.
Expectation:  On-time delivery, clean enough that no rework is
              needed, for the next 6 weeks.
Support:      Weekly 30-min check-in; paired with Priya for 2 weeks;
              I'll take the vendor calls off your plate.
Timeline:     6 weeks — review together on the 15th.
If we're not there:  We'll talk honestly about a different role or
              a respectful move on. No surprises either way.

Write it on one page. Share it. Both of you should be able to look at the same thing and agree on what “working” means. The goal of the plan is not to build a paper trail to fire someone — if that’s your real goal, you’ve already decided, and you should skip to the exit. The goal is to give a real chance, honestly.

How it connects

An improvement plan is feedback with structure and stakes attached — the same principles of specificity and evidence, formalized because the situation is serious. The everyday version of this skill lives in your regular one-on-ones and coaching; by the time you need a formal plan, the ordinary conversations should already have happened. If they haven’t, the plan will feel like an ambush, and that’s on the manager, not the person.

Telling genuine progress from stalling

Once the plan is running, your hardest job is honest reading. Managers err both ways — some declare victory at the first good week (relief makes us gullible), others can’t see improvement because they’ve already written the person off (frustration makes us blind). Watch for the difference between real change and its imitations.

Genuine progress looks like:

• Improvement across most of the measures, not one cherry-picked win.
• The person owns it — they name what they changed and why, rather than crediting luck or your leniency.
• It holds up over weeks, not just the first anxious fortnight.
• They’re asking better questions and needing less scaffolding, not more.

Stalling looks like:

• A burst of effort right after the conversation that fades within two weeks.
• Activity without results — lots of visible busyness, same outcomes.
• Every miss has an external reason; nothing is ever theirs to fix.
• You’re doing more of the work of the plan than they are.

Why does it matter to call this early and honestly? Because dragging out a hopeless case harms everyone. It harms the person, who’s kept in a role they’re failing at, their confidence eroding month after month, when they could have found a fit that suits them. It harms the team, who see the standard isn’t real and quietly resent covering the gap — your best people notice first, and it’s exactly the ones you most want to keep who lose faith. And it harms you, by consuming energy and credibility you owe to seven other people. Kindness that avoids the hard call isn’t kindness; it’s the postponement of a harder one, paid for by others.

The common trap — the endless extension

The most common failure of nerve is the rolling extension: the six-week plan quietly becomes twelve, then “let’s see how the next project goes,” then a year. Each extension feels compassionate in the moment. Together they add up to cruelty — the person invests another year in a role that was never going to work, while the team’s trust in the standard bleeds out. Decide in advance what genuine progress would look like, and hold yourself to reading the evidence at the date you set. One honest extension, with a clear new bar, can be fair. A pattern of them is avoidance wearing kindness as a costume.

The humane exit as a legitimate outcome

Here is the reframe that changes everything: an exit is not the failure state. A person kept for years in a role they can’t succeed in — that’s the failure. When skill can’t close the gap, will can’t be rekindled, and the situation is genuinely a bad fit, letting someone go is often the most respectful thing you can do for them.

Think of it from the person’s side. Almost nobody wants to spend their working life failing, dreading Mondays, sensing they’re disappointing people, watching their confidence drain. Freeing someone from a role that’s slowly grinding them down — and doing it early, before years are lost — is a gift, even though it never feels like one at the moment it happens.

A humane exit rests on a few principles:

• Early beats late. Ending it after three honest months is far kinder than dragging it to three years. The longer it runs, the more it damages the person’s confidence and finances and the harder the eventual landing.
• No surprises. If you ran the honest conversation and the plan, the exit is a conclusion the person saw coming, not an ambush. Surprise is what turns a hard moment into a betrayal.
• Dignity throughout. How you end it is remembered far longer than the performance issue. Private, direct, respectful. Help with the transition — a fair reference where you can honestly give one, time to look, no public shaming.
• Honesty about the fit, not the person. “This role wasn’t the right fit for your strengths” is usually both true and kind. Most underperformers are not bad people or even bad workers — they’re the wrong match for one particular seat.

The mechanics of doing this well — the conversation itself, the legal and human care it requires — are their own subject, covered in Layoffs and Letting Someone Go With Dignity. What matters here is accepting the outcome as legitimate. If you treat every exit as a personal defeat, you’ll dodge them, and your dodging will cost the very people you’re trying to protect.

Case study — Netflix’s “keeper test” (culture deck, 2009)

When Netflix published its culture deck in 2009, one idea drew both praise and criticism: the “keeper test” — managers asking themselves whether they’d fight to keep each person, and, if not, offering a generous exit rather than a slow decline. The controversial part is real and worth naming: pushed too hard, that ethos can feel ruthless and can make people fearful rather than free. But the defensible core is the one this page argues for — that clarity about fit, acted on early and paired with a genuinely generous, dignified exit, is often kinder than the quiet limbo most organizations default to. Take the honesty and the dignity; leave the fear. The lesson isn’t “cut people fast” — it’s “don’t leave people stranded in a role that’s failing them while pretending not to notice.”

In practice — the exit that was a relief

An engineer had been carried on a team for over a year — kind, well-liked, and clearly out of his depth in the role. Everyone knew; no one said it. When a new manager finally had the honest conversation and helped him move on, the surprising part was his own reaction weeks later: relief. He’d been exhausted by the daily performance of competence he didn’t feel. He landed in a role that fit and thrived. The unkindness had been the year of silence, not the ending.

Try this

Pick one person on your team you have a nagging worry about — or, if you’re not managing yet, one whose performance you’ve quietly judged. Do just the diagnosis this week, no action: write the gap in one specific, evidence-based sentence (not a feeling). Then, honestly, run the three-root check — skill, will, or situation — and for the situation root especially, ask what you might not have made clear or provided. If you can’t write the specific sentence, that’s your finding: you have a feeling, not a case. Notice how different the situation looks once it’s named precisely instead of felt vaguely.

Reflect

1. Think of an underperformer you’ve worked with or managed. Looking back honestly, was the real root skill, will, or situation — and did the response at the time match the actual cause?
2. When you feel someone “isn’t trying,” how often have you checked whether the setup — clear expectations, resources, fit — was actually in place first?
3. Recall a hard-performance conversation you’ve had or witnessed. Was it vague, accusatory, or specific-and-generous? What did the shape of it produce?
4. Is there a situation you’re currently dragging out with rolling extensions because ending it feels unkind? Whose interests is the delay actually serving?
5. Does part of you treat letting someone go as a personal failure? How would your willingness to act early change if you saw an early, humane exit as a kindness instead?

Show reflections

1. Most people find, on honest reflection, that at least one past case was misdiagnosed — usually a situation or skill problem read as a will problem. The lesson isn’t guilt; it’s that the diagnosis step is the whole game, and skipping it sends the response in the wrong direction for months.
2. The honest answer is often “rarely, and only after I’d already decided.” The value is in noticing your default: if you jump to “they don’t care,” the fundamental attribution error is likely at work, and the situational check is the discipline that protects you and them.
3. Vague conversations produce nothing to act on; accusatory ones produce defense and hidden causes; specific-and-generous ones produce the truth you need to diagnose. If you can recall which shape yielded a real answer, you’ve located the technique worth repeating.
4. Naming whose interests the delay serves is the sharp part. Usually the honest answer is “mine — it spares me a hard conversation,” dressed up as compassion for the person. That reframe is often what makes the overdue call finally possible.
5. If letting someone go reads as personal failure, you’ll avoid it, and the avoidance harms the person, the team, and you. Seeing an early, dignified exit as freeing someone from a slow grind — rather than as your defeat — is what lets you act while it’s still kind to act.