An engineering manager who works for me came to me about a hire. He had a candidate late in the process and a worry about how they would sit in the team — not competence, balance. He could name the concern, he could describe what it would do to the group he had built, and he had a view about what he wanted to do.
He was not there to find out what I made of the candidate. He was there for permission to say no.
Turning down somebody who is not obviously weak is a specific kind of hard. When a candidate is clearly not good enough the decision makes itself. When they are perfectly decent and you still do not want them, the call is yours and your name is on it. Anybody who worries about being fair feels that, and those are the people you want running your hiring.
So I gave him no opinion on the candidate. I asked him to talk me through the concern, and then I asked what he wanted for the team. By the end of the second answer it was obvious he had decided some time ago. Then I told him I agreed.
I contributed nothing to that conversation about the candidate. What I contributed was that his read was good enough to act on.
What managers stop doing
He is not a special case. What he needed was feedback at a level of detail most managers stop applying to anyone who appears to have it handled.
That is the failure I see most, and it does not feel like a failure while you are committing it. The strong performer is the person you do not have to think about, so you stop thinking about them, and everything you say to them coarsens without your ever deciding it should. Development points become general encouragement to stretch. Recognition becomes “you’re smashing it.”
Silence reads as endorsement. It is absence, and “you’re doing great” is the same absence with a sentence wrapped round it.
There is a second route to the same place, and in my experience it is much the more common. This manager does not go vague at all. They keep paying close attention, they stay specific, and every specific thing they say is about a gap. None of it is wrong or lazy. It is that across the whole time somebody works for them, they never once say out loud what that person is good at.
I have worked with and for a lot of managers. One of them did the recognition part properly. One.
Which invites the obvious question: if almost nobody does this, how did any of us get anywhere? Mostly on our own. High performers tend to be self-driven whatever the circumstances, and plenty arrive with no help at all. Some of that is survivorship. For others the recognition came from outside the reporting line, from a coach or a mentor or a peer who happened to say the thing at the right moment.
I am not claiming these people would have failed without a manager paying attention. The claim is smaller. They got there anyway, slower and with far less certainty than they needed to. Supporting them is the job whether or not they would have managed without it.
The years before that conversation
That conversation was years in the making, and it did not start anywhere near there.
It started with feedback reaching me from around the organisation that he needed to step up: be more visible, be more decisive, hold a position when it was uncomfortable. About as actionable as tagging “make no mistakes” onto the end of a prompt. (And we all know how that worked out!) So I got a great deal more constructive with him.
He needed that. He did not need somebody telling him he was excellent, he needed tactics. How to run the conversation he had been working around. How to say the thing plainly instead of at an angle. What decisive looks like when you are the person everyone is waiting on. That is ordinary developmental feedback, delivered kindly and repeatedly, and there was no route past it.
I never had to work out what sat underneath, because he had named it himself. He is a people pleaser and he minds what people think of him, including the people who work for him.
What I did not expect was how little difference the tactics made on their own. They gave him the moves, but the thing actually holding him up was permission to use them. He would get to the right answer and then come and check it, because knowing was not the same as being willing to be the one who said it.
Every time he checked and I agreed, he got something he could not get anywhere else. Not “you’re doing great”, but confirmation that the specific judgement he had just made, on that specific problem, was the judgement I would have made. Do that enough times and someone stops needing to ask.
The obvious objection is that I could have saved years by telling him plainly that the call was his. Expectations conversations have their place, and an experienced hire finding the edges of a new job can take that steer and run with it. Somebody growing into a role is different. Knowing you are allowed to do a thing is not the same as being ready to be the person who did it, and only time and repetition close the distance between the two. Treating that gap as one conversation about expectations is harsh and unsupportive.
The opposite diagnosis
There is a competing account of what goes wrong with high performers, and it points the other way. On that account they are short of challenge, not praise. Their manager, unwilling to unsettle the one person who never causes trouble, softens every difficult conversation until nothing survives it. The person stays excellent and stops getting better, and everyone tells them they are brilliant while it happens.
That account is correct, and it is not a rival explanation. It has the same cause.
Specific criticism and specific praise draw on the same thing: a manager still looking closely at the work. Stop examining what somebody does in detail and you lose the ability to say what was excellent about it and what was not. What remains is a general impression, and a general impression only ever produces general sentences. “You’re crushing it” and “I think you could be pushing yourself harder” come from exactly the same place. Neither requires you to have read the RFC.
So forget the balance of warmth and challenge in your one-to-ones. Ask whether anything you said to that person last month contained information they could act on. Then ask whether a year of it would add up to a picture they could recognise themselves in.
The gaps-only failure is the harder of the two to see. It looks like a manager doing the job well, and for a stretch that is what it is. It was the right thing with the manager I have been describing, and skipping it would have left him confident and still short of the tactics. The failure is never coming out of that mode.
Stay in it long enough and somebody is assembling a picture of themselves entirely out of shortfalls, because shortfalls are the only material they have been handed. Some people run on that happily for years. For others it removes the only reference point they had. A list of gaps tells you where you fell short of a standard nobody has described to you. It cannot tell you what good looks like. It cannot tell you whether you are hitting it, or whether you are any better than you were a year ago. Getting better requires knowing where the bar sits and how far off it you are. Without that, somebody can push very hard for a very long time with no idea whether any of it is moving them.
Confidence has to be supplied
Most managers treat confidence as a by-product. Someone does good work, it accumulates, and confidence arrives on its own as a reasonable inference from the record. On that model there is not much for a manager to do, because reality will handle it.
I think that has it backwards. Confidence in the sense that matters is not a general feeling about yourself. It is knowing what you are good at, and why, in enough detail to rely on it. For anyone ambitious, that is the same information as knowing where the bar sits. Tell somebody precisely what was excellent about a piece of work and you have told them two things at once: that this is a strength of theirs, and that it cleared the standard you hold them to. For anybody who takes their bar from outside themselves, the recognition is the calibration.
People cannot generate it alone. Their own mechanism is invisible to them — the thing they did that a less capable person would not have done — and it is invisible precisely because it came easily. The standard is even further out of reach. They know what they produced. They do not know what you were holding it against when you judged it. Both of those things sit with you, and they stay there until you say them out loud.
So feedback is an input to performance rather than a reward for it. You are not thanking someone for last quarter; you are deciding whether they have what they need for the next one. Withholding it is not a neutral act. It costs nothing to supply, which is not the same as having supplied it.
Specificity is what does the work here, rather than warmth. “You’re so reliable” is a compliment. It is pleasant to receive, it recognises nothing in particular and it calibrates nothing at all. “The way you framed the tradeoffs in that RFC meant the team could actually decide instead of deferring again” is information about how they operate, and something they can repeat on purpose.
What high resolution sounds like
None of this needs a new ritual or a different one-to-one template. It needs more precise sentences.
Lead with the celebration and let it stand on its own. Not every conversation requires a second half. “You fucking killed it, and this is what you did” is a complete conversation. The reflex to append a development point converts specific feedback back into the generic kind, because the person stops hearing the first half while they wait for the turn.
Be specific about what they did, not how it turned out. “Great job on the launch” tells somebody the launch went well, which they knew. “The way you handled comms when the deploy went sideways meant nobody panicked and the customer never clocked it, that is judgement, keep doing it” tells them which of their own behaviours produced the result.
Tell them you appreciate them, and not only their output. This reads as the soft item and it is not. There is a real difference between being valued for what you produce and being valued as a person on the team, and everybody can tell which one they are getting. People who suspect they are worth exactly their last delivery do not stop delivering, and they start looking the moment the delivery dips.
Tell them you have their back, when it is true and never when it is not. People decide how much risk to take based on what they believe will be defended when it goes wrong. Somebody who thinks they are exposed makes conservative choices, and conservative choices from your strongest engineers are an expensive way to run a team. Saying it out loud changes that calculation; thinking it changes nothing.
Stop looking for the development point in every conversation. The reflex to find a growth edge every time comes from a reasonable place, because it feels like the part of the job that is real work. But if ninety-five per cent of what somebody did was excellent, opening with the other five is not rigour. It is what you reach for when you have not looked closely enough to say anything precise about the ninety-five.
Most managers, reading that, will say they already do it. A good number are right about the frequency and wrong about the contents. Praise handed out every week that never names anything in particular does not become high-resolution through repetition.
The review cycle will not rescue it either. By the time you write up in March something somebody did in October, the specifics have gone, yours as much as theirs. What survives compression into a ratings form is precisely the general impression the exercise was meant to improve on. Anything worth saying has to be said the week you noticed it.
Which input this person is short of
None of this amounts to praising everybody more. That would be the same mistake in the other direction.
Plenty of high performers do not want recognition in any form you would think to give it. They want autonomy, or scope, or a harder problem, or the room to get on with it. Telling that person repeatedly how brilliant they are is noise, and it can land worse than that — like a manager fishing for credit for having noticed.
Working out which input this particular person is short of, and supplying that one, is the whole skill. It is also why the answer cannot be a policy. You can mandate that every manager gives feedback fortnightly. You cannot mandate that they know what kind.
So learn their tells. People signal uncertainty in ways that do not look like uncertainty: some go quiet, some overcommunicate, some ask permission for things they have had the authority to do for a year. His was the checking. It still is, occasionally, and I get to say “come on, forgiveness, remember” and he goes and does the thing. But the direction is unmistakable across staff meetings, one-to-ones and Slack. He acts without asking, he volunteers, he pushes. None of it came from a single conversation. It came from a few years of small ones, most of which were nothing more than me confirming that what he already thought was right.
This is also where the argument runs out of road. Reading somebody’s tells means seeing them work often enough to know what normal looks like for them. With your own reports that is achievable. With a skip-level, or somebody who reaches you through a matrix and appears in your week for a single review meeting, it is not. Pretending otherwise produces confident, specific-sounding feedback assembled from a sample of two interactions. If you are not close enough to the work, get the detail from the people who are, and say that is where it came from.
The part only you can see
The easier somebody makes hard things look, the more specific you have to be with them. You may be the only person in a position to see the gap between what they did and what somebody less capable would have done in the same situation.
Say it. Out loud, in detail, close to when it happened.