Someone on your team was reliably good. For the last few months they have not been. The work arrives late, or arrives thin. They have gone quiet in meetings where they used to have a view. Nothing is dramatic enough to act on and nothing is fine.

Most leaders sit with this for longer than they later think was wise, and then move quickly once they do act. Both parts of that pattern come from the same place: the situation is uncomfortable, so it is avoided, and then handled in a way that resolves the discomfort rather than the problem.

There is a step worth taking first, which is to be precise about what you are looking at.

Four situations that look identical from the outside

Capability. The role changed underneath them. Scope grew, the team grew, the technical demands moved, and what made them strong in the previous version of the job is not what the current version needs. This is common after a promotion or a reorganisation and it is often invisible because the job title did not change.

Motivation. They can still do it. They have stopped wanting to. Something happened: a promotion that went elsewhere, a project killed, a manager change, a promise that quietly expired. The performance drop is downstream of a conclusion they reached and did not tell you about.

Circumstance. Something outside work is consuming them. Illness, a family situation, a personal crisis. They have decided not to disclose it, often for reasons that have nothing to do with trust in you.

System. Nothing changed in the person. The conditions changed. Unclear priorities, a peer who blocks them, a process that now takes three times as long, decisions that no longer get made. They are performing exactly as the system now permits.

These four require entirely different responses, and only two of them are about the individual. If you begin the conversation having already privately decided it is motivation, you will run a conversation about commitment while the real problem is a broken handover with another team, and you will end up with a demotivated person who is now also under suspicion.

The part leaders skip

Before the conversation, it is worth asking what you have already concluded, and what evidence you actually have for it.

Most of us arrive at these conversations with a diagnosis already formed. That is not a flaw, it is how experienced judgement works. The problem is that a formed diagnosis makes the conversation a confirmation exercise. You ask the questions that test your theory. They sense the direction, become defensive, and the defensiveness confirms your theory.

The second thing worth checking is your own contribution. Not as a ritual of self-criticism, but as a factual question. Did the role change without a conversation? Did you stop giving them the interesting work? Did something you said six months ago land differently from how you intended? Leaders are often surprised here, and it is much better to be surprised in your own head than in the meeting.

Running the conversation

The conversations that work differ from the rest in three ways.

Name the observation, not the interpretation. “The last three reports came in after the deadline and with less analysis than usual” is a fact you can both look at. “You seem disengaged” is a verdict they now have to defend themselves against.

Say plainly that the change is what concerns you. Not the standard in isolation, but the gap between now and what you know they can do. That framing tells them you have not written them off, which is the thing they are most likely to be listening for.

Then stop talking. The most important information in these conversations usually arrives after a silence that feels too long. Leaders tend to fill it, and lose the one thing they came for.

Do not solve it in the room. The instinct to leave with a plan is strong, and it usually produces a plan aimed at the wrong one of the four situations. Ending with “I want to think about what you have told me, let us pick this up on Thursday” is a stronger position than a fast resolution whose consequences you will have to carry.

When it is genuinely performance

Sometimes it is. The capability is not there for the role as it now exists, the person knows it, and the kindest thing available is clarity: what needs to change, by when, what happens if it does not, and what support exists in between. Delay is not compassion here. Ambiguity is harder on people than a clear standard, and a person left in an unspoken probation for six months is being treated worse than a person told directly where they stand.

It is not about being kind or being firm. It is about whether the person hears it while there is still time to do something, or only once it has been decided.

Worth sitting with

  • Which of the four is it? What evidence do you have, and what are you assuming?
  • What did you conclude before you gathered any of it?
  • What has changed in the conditions around this person in the last six months?
  • What are you avoiding by waiting, and what is the wait costing the rest of the team?
  • If you knew for certain it was circumstance, would you run the conversation differently? If yes, why are you not running that version first?