How to Give Feedback as a Leader
There is a specific problem with giving feedback once you are senior, and it is not the one most advice addresses. Your feedback arrives amplified. A mild observation from the founder is heard as a serious concern; a serious concern is heard as a warning. Most leaders discover this by accident — they make a passing remark and find a week later that someone has restructured their entire project — and the lesson they draw is to soften everything, which produces feedback so hedged that nothing changes. This page is about the structure that avoids both failures.
What’s inside
Your feedback lands harder than you intend
The two-part structure
Frequency beats quality
Four things that make feedback fail
Feedback to people more expert than you
Praise, which is undervalued
Where Blomma fits
Your feedback lands harder than you intend
The amplification is structural rather than a matter of your tone. Most people have little contact with you, so each interaction carries a lot of inferential weight. They also know you decide their scope, their compensation, and whether they stay, which means anything you say is heard with that in the background whether or not you intend it.
So the recalibration required is genuine. What feels to you like a mild observation — “I wondered about the second section” — may be received as a directive. What feels to you like a direct piece of criticism may be received as an existential threat.
Two practical consequences. First, be more precise about weight than feels natural: say explicitly how much this matters. “This is a small thing and I’m mentioning it once” and “this is important and I want to see it change” are different sentences, and your audience cannot tell them apart from tone alone.
Second, be careful with speculation. “I wonder if we should” gets heard as “we are going to.” If you are thinking out loud, say that you are thinking out loud.
The two-part structure
Most feedback frameworks are too elaborate to use under pressure. Two parts is enough for nearly everything.
An observation, stated plainly and specifically.What you saw, not what it means about the person. “This report didn’t answer the client’s second question.” “You committed to Thursday and it landed Monday without a flag.” Factual, singular, checkable.
A question that hands it back.“What happened, and how do we make sure the next one does?” “What would have made it possible to flag that earlier?”
That is the whole structure, and its power is in what it omits. It contains no theory about their character, no accumulated history, and no solution. The observation makes the gap undeniable; the question makes them the one who closes it.
What this replaces: the sandwich, which everyone recognises and which teaches people to discount your praise; the general characterisation, which cannot be discussed and can only be disputed; and the pre-solved version, where you identify the gap and specify the fix, which removes the learning.
One addition for the harder cases: after the question, name the weight. “This is the second time, and I need it to be the last.” That sentence is uncomfortable to say and it is what makes serious feedback land as serious.
Frequency beats quality
The single highest-leverage change most leaders can make to their feedback is to give more of it, sooner, smaller.
Feedback delivered in the week it happens is small, specific, and low-stakes. The same feedback three months later has accumulated significance, arrives with a history attached, and lands as a verdict. The content is identical; the frequency determines whether it is a correction or an event.
Frequency also does something structural: it makes feedback ordinary. In a team where small feedback happens weekly, a serious conversation is a variation on a normal interaction. In a team where feedback happens quarterly, any feedback is a signal that something is wrong, and everyone braces.
The practical version: when you notice something, say it within the week. Not a scheduled session — two sentences at the end of a conversation you were already having. Most leaders’ feedback problem is not that they deliver it badly, it is that they store it up until it is too heavy to deliver at all.
And notice what you are storing. If you have three things you have not said to someone, you have a backlog, and the backlog will eventually get delivered all at once in a way that is unfair to them.
Four things that make feedback fail
Character claims.“You’re not detail-oriented” cannot be acted on and will be defended. “This missed two of the three questions” can be. Stay on the work.
Stacking.Five pieces of evidence reads as a case you have been building, and the recipient responds to the case rather than the content. One or two instances, then stop.
Softening past the point of clarity.So much hedging that the person leaves unsure whether anything was wrong. This is the most common leader failure, and it is worse than being too blunt because it produces no change and still creates unease.
Delay.Covered above, and worth repeating because it causes more failures than delivery technique ever does.
A fifth, specific to leaders: giving feedback in front of others. At your level this is not a small error — a critical remark from the founder in a group setting is a status event, and it will be remembered for years.
Feedback to people more expert than you
A situation founders face constantly and rarely handle well: giving feedback to someone who knows their domain far better than you do.
The move is to stay on what you can legitimately observe. You may not be able to assess their architecture, but you can observe that the project has slipped twice, that two of their people have left, that they escalate decisions you expected them to own, or that their explanations arrive after the fact rather than before. Those are your observations and they are valid.
What does not work is contesting them on their expertise. You will lose, and you should — they know more than you. Founders who attempt this either get talked out of a legitimate concern or override someone on a technical matter where they are wrong, and both outcomes are worse than staying on the observable.
And be explicit about the boundary: “I’m not going to tell you how to solve this — that’s your call. What I’m telling you is that the pattern I’m seeing is X.” That sentence keeps you on solid ground and makes the feedback harder to deflect.
Praise, which is undervalued
Most leaders under-invest here, treating recognition as pleasant rather than functional. It is functional.
Specific praise, delivered close to the event, does the same thing critical feedback does — it tells people what the standard is. Vague praise does not: “great work” carries no information, so it neither reinforces the behaviour nor tells anyone else what good looks like.
So apply the same discipline. Name the specific thing and why it mattered. “You flagged the risk on Wednesday rather than waiting — that gave us a week we wouldn’t have had.” That sentence teaches the behaviour to everyone who hears it.
And do it publicly when the behaviour is one you want to spread, particularly for the invisible things: someone who documented their work, prevented a problem, or brought you unwelcome news early. Those behaviours are exactly the ones that go unrewarded by default, and what you praise is the clearest statement you make about what the company values.
Where Blomma fits
Feedback is a behavioural skill, which means it improves with reps rather than understanding — and the reps are unforgiving. Every counterparty is a real relationship, the amplification means your mistakes cost more than they would have earlier in your career, and there is nowhere to try the sentence out first.
Blomma is an always-on AI career coach with no stake in your team. Use it to construct the observation-then-question version of something you have been sitting on, which is usually the reason it has not been said — you have the concern and not the sentence. Use it to check whether what you have drafted contains a character claim or five stacked instances. Use it to work out how to give feedback to someone more expert than you without straying onto their ground. And use it to audit your backlog: the three things you have not said to someone, which will otherwise arrive all at once.
The second application: your managers have the same problem at their level and mostly no training in it, and feedback quality across a company is set by the management layer rather than by you. It is one of the most reliably coachable skills there is..
Feedback as a leader is not a matter of finding the right words for a hard conversation once a quarter. It is a high-frequency, low-weight habit, calibrated for the fact that everything you say arrives louder than you sent it.