Feedback without evidence is just doubt

I told someone their approach worried me. I had not done the work to say why. The concern may well have been right, and that turned out not to matter at all.

21 September 2026 · 5 min read · Growing engineers

Early on as a manager I watched someone take on a project that was more ambiguous than their level really called for. They wanted it. They were determined to own it and to deliver it, and they were not wrong to want it.

I thought they were being set up to fail.

So I said something. I raised concerns about how they were going about it and suggested they change their way of working. All of which sounds like a manager doing the job: you see a risk to someone you are responsible for, and you do not sit on it.

The problem was that I had done no work at all. My concern was a feeling. I had not gone into the detail, I had not collected anything, and I could not point at a single thing and say “this is what worries me and here is why”. I brought them a conclusion and kept the reasoning, because there was no reasoning to keep.

What that actually sounds like from the other chair

Strip the evidence out of feedback and look at what is left.

“Your approach concerns me” with nothing behind it is not a statement about the approach. It is a statement about them. There is nothing in the room to examine except the person, so that is what the conversation becomes about, and they spend it defending themselves rather than thinking about the work.

It is worse when the person is ambitious and has just put their hand up for something hard. They have taken a risk in public. An unevidenced concern from their manager, at that moment, reads as a verdict on whether they were right to. I meant it protectively. That is not how it arrives.

Evidence changes what is on the table. If I had walked in with the actual shape of the thing, the dependencies nobody had mapped, the places the scope quietly widened, then there is a document between us and we are both looking at it. The conversation stops being about whether I rate them and starts being about what is true.

This is the same failure as handing over the task and keeping the context, pointed at a person instead of a piece of work. In both cases I transferred the output of my thinking without any of the thinking, and then wondered why it did not take.

But you cannot wait for a dossier

The obvious objection is that this turns every piece of feedback into a research project, and by the time you have your evidence the moment has gone. That is a real tension and it deserves a straight answer rather than a slogan.

Two things resolve most of it.

Say the worry as a worry, and label it as one. “Something about this is bothering me and I have not done the work to say what. I am not asking you to change anything yet.” That is honest, it takes ten seconds, and it is a completely different object from a suggestion to change your way of working. It puts the uncertainty where it belongs, which is on me.

Then commit to a date. “Give me until Thursday and I will have gone through it properly.” Doing the homework is not the same as delaying. What I actually did was skip the homework and keep the authority, which is the worst of both.

The version of this I would defend is: your instinct is allowed to start the process. It is not allowed to be the whole of it.

What the evidence is really for

Over the years I have come to treat data as the foundation of these conversations, and the reasons are more practical than they sound.

It makes the thing transparent. Both of you can see the same object, so there is no guessing at what the manager privately thinks.

It takes my bias out, or at least exposes it. Plenty of my gut feelings have turned out to be about how I would have approached something, not about whether their approach works.

It makes disagreement possible, which is the one people miss. Someone can argue with a number, a timeline, a list of dependencies. Nobody can argue with a feeling, so a feeling does not invite a response, it invites compliance or resentment.

And it protects them from me. If I am wrong, doing the work is what surfaces that before I have spent someone’s confidence on a hunch.

The part I have not resolved

I may well have been right about that project. That is the uncomfortable bit, and it is why this one stayed with me.

Being right is not the same as being useful. A correct concern, delivered with nothing to look at, helped nobody. They could not act on it, I could not prove it, and the only thing that actually got transmitted was that their manager had doubts about them.

If the instinct is worth acting on, it is worth an afternoon. And if it is not worth an afternoon, it was probably not worth saying in the form I said it.