How to give feedback that lands (without sounding like HR)

A manager opens a 1:1 with "So, I wanted to give you some feedback," and watches the other person's face close like a door. Shoulders rise an inch. The listening stops; the bracing starts. Whatever gets said next — however carefully sandwiched, however many "great jobs" cushion it — lands on someone who is no longer taking in information. They're managing a threat. By the time the manager finishes, the report has heard one sentence and is already rehearsing their defense of it.

This is the central problem with how most of us were taught to give feedback: the methods optimize for the giver's comfort delivering a verdict, not the receiver's ability to use it. We learned to be "honest," "direct," "candid" — words that describe how it feels to say the thing, not whether anything changes afterward. And somewhere along the way, "feedback" stopped meaning "useful information about your work" and started meaning "the uncomfortable part of the meeting where I tell you what's wrong." No wonder people flinch. We trained them to.

This guide is the overview of an entire approach to giving feedback that doesn't sound like an HR script and doesn't trigger the door-closing reflex. It's the anchor for a series of deeper pieces on specific moments — the defensive reaction, the repeated mistake, the power imbalance, the feedback you keep avoiding. Start here for the principles; the rest extend them.

Why "feedback" became a word people dread

The corporate version of feedback carries three pieces of baggage, and naming them is the first step to dropping them.

It's bundled with judgment. In most organizations, feedback arrives attached to ratings, reviews, and calibration meetings — moments where what's said about you affects your money and your standing. So even a genuinely developmental comment, offered with care, gets heard through that filter. The receiver can't tell whether you're helping them grow or building a case. The format taught them not to trust the difference.

It's delivered in formulas that everyone can see through. The feedback sandwich — praise, criticism, praise — is so transparent that the moment you open with a compliment, the other person is just waiting for the "but." The praise gets discounted as setup, the criticism lands anyway, and the closing praise feels like an apology. Formulas signal that the giver is following a procedure rather than telling the truth, which is exactly the HR-script quality that makes feedback feel hollow.

It assumes the giver is the expert on the receiver. The deepest flaw, and the hardest to see. Marcus Buckingham and Ashley Goodall made the case directly in their HBR essay The Feedback Fallacy: the belief that other people can reliably rate you, and that telling you your shortcomings drives improvement, runs against most of what the research actually shows. People learn far more from attention to what they did well and how to do more of it than from a catalogue of where they fell short. The standard model has the mechanism backwards.

None of this means feedback is useless. It means the default form of it — verdict-shaped, formula-delivered, expert-from-above — is the form that fails. The fix isn't to give less feedback. It's to change what you think you're doing when you give it.

The one distinction that fixes most of it

If you change one thing, change this: stop trying to judge the past and start trying to improve the future.

Almost every piece of bad feedback is a judgment about something already finished — "that presentation was unclear," "you dropped the ball on the launch," "your writing is too dense." All true, perhaps. All useless. The event is over; the person can't go back and fix it; the only thing your judgment produces is the defensive reaction, because you've handed them a verdict they can't change and can only dispute.

Future-focused feedback takes the same observation and points it forward: "Next time you present to that group, open with the recommendation before the analysis — they decide fast and want the answer first." Same underlying issue. Completely different conversation. One asks the person to defend who they were; the other invites them to be better next time. The first is a courtroom. The second is coaching.

Douglas Stone and Sheila Heen, whose book Thanks for the Feedback reframed the whole topic around the receiver's experience, draw a related line between three things we sloppily call "feedback": appreciation (you matter), coaching (here's how to improve), and evaluation (here's where you stand). Most feedback conversations fail because the giver is delivering evaluation while believing they're offering coaching — and the receiver, hearing evaluation, responds to the threat instead of the lesson. Decide which of the three you're actually doing before you open your mouth. If it's coaching, keep judgment out of it entirely. If it's evaluation, don't disguise it as helpful tips.

This single shift — future over past, coaching over verdict — quietly solves most of what makes feedback hard. The defensiveness drops because there's nothing to defend. The relationship survives because you're visibly on their side. And you stop sounding like HR, because HR delivers verdicts and you're doing something else.

Anchor to a behavior and a moment, not to character

The second principle is mechanical, and it's where the popular models (SBI, COIN, STAR, and the rest) all converge: describe a specific behavior in a specific situation, not a trait.

"You're not detail-oriented" is a character verdict. It's unfalsifiable, it's an identity attack, and there's nothing actionable in it. "In yesterday's spec, the API section skipped the error cases, and the team built without them" is an observation. It names a moment, a behavior, and an impact — and it leaves the person's character entirely out of it. They can fix the second one. They can only resent the first.

The various feedback frameworks are mostly different packaging for this one idea, and which acronym you use matters far less than whether you hit the three notes: what happened (the specific situation), what you observed (the behavior, not your interpretation of it), and what it led to (the concrete effect). Skip the impact and it sounds like nitpicking. Skip the situation and it sounds like a general complaint. Lead with your interpretation — "you didn't care about the deadline" — and you've turned an observation into an accusation about their inner state, which they will correctly dispute because you can't actually see inside their head.

We compare the specific models — SBI, COIN, and STAR — in a dedicated piece, because each fits a slightly different moment. But the meta-skill underneath all of them is the same: behavior and moment, never character. (It's also why no formula rescues a bad approach — the same reason the feedback sandwich quietly backfires.)

What it looks like rewritten

The principles are easier to trust when you see the same feedback delivered both ways. Two common moments, the verdict version and the version that lands:

A report whose work keeps needing rework. The verdict: "You're careless — I keep finding things I have to send back." It names a trait, offers no path forward, and guarantees a defense. The rewrite: "The last two PRs each came back for missing test coverage on the edge cases. Going forward, can we make 'edge cases tested' the last thing you check before you mark it ready? I'd rather you take the extra twenty minutes than have it bounce." Same problem, named as a behavior, pointed forward, with the impact and the fix both visible.

A peer who dominates meetings. The verdict: "You talk too much in meetings and people tune out." True, maybe — and impossible to act on without resentment. The rewrite: "In the planning sync, I noticed the two newer folks didn't get a word in before we ran out of time. I think we'd get better calls if we heard from them earlier — want to try going round-robin on the big decisions?" No character claim, a shared goal ("better calls"), and a concrete mechanism instead of a complaint.

The pattern holds across every example: strip the trait, name the moment, show the impact, and hand over something to do — not something to be.

Match the shape to the relationship

Feedback isn't one skill — it's three, depending on which direction it travels, and the direction changes everything about how it has to be shaped.

Giving feedback to a direct report is the most-discussed case and the one most fraught with power. They can't easily push back, which means your feedback carries more weight than you intend and more threat than you notice. The discipline here is restraint: fewer, sharper observations; future-focused framing; and an explicit invitation to disagree, because the power gap will otherwise turn your tentative suggestion into a command they resent.

Giving feedback to a peer runs on a different engine — you have no authority, so it runs entirely on the relationship. The risk isn't that they'll feel crushed; it's that they'll feel you've overstepped. Peer feedback lands when it's clearly in service of a shared goal ("I want this project to go well, so —") and curdles when it reads as you grading them without standing to do so.

Giving feedback to your boss is the highest-degree-of-difficulty version, and it's the mirror image of a problem we cover from the other side: most managers can't get honest upward feedback because the power gap makes honesty a bad trade for the person below. When you're the one giving it upward, you're on the costly side of that trade — which means framing it as a question, anchoring hard to specifics, and tying it to outcomes your boss already cares about.

Each of these deserves its own deep treatment — the peer and direct-report versions are linked above, and giving feedback upward to your boss is the hardest of the three. The boundary worth drawing now: this whole pillar is about you giving feedback to someone else. The mirror skill — you asking for and receiving feedback well — is a separate craft entirely, and confusing the two is why so many managers are fluent at dishing it out and hopeless at taking it in.

The delivery moment: surviving the first defensive reaction

Even perfect feedback meets resistance, because the receiving brain treats "here's how you could improve" as a status threat first and a gift second. The defensive reaction — the justification, the counter-example, the slightly-too-quick "yeah, I know" — is not a sign your feedback failed. It's the normal first thirty seconds of it working.

The mistake most managers make is to argue with the defense. The person says "well, the timeline was impossible," and the manager doubles down on the original point, and now it's a debate about who's right instead of a conversation about what to do next. The move that works is the opposite: absorb the first reaction without contesting it. "That's fair, the timeline was brutal — and I still want to figure out how the error cases get caught next time." You've conceded the defensible part, which lowers the threat, and you've redirected to the future, which is where the useful conversation lives. You don't have to win the first thirty seconds. You have to survive them.

This is also where giving feedback connects to the mistakes that derail first-time managers — the instinct to either avoid the hard conversation entirely or to over-deliver it once cornered. The calibrated middle — say it early, say it small, say it forward-looking, survive the reaction — is a learned skill, not a personality trait.

The follow-up nobody does

Here's the part missing from almost every feedback guide: what happens after. Most feedback is delivered once and never mentioned again, which teaches the receiver one of two wrong lessons — either it didn't matter (you never followed up, so why change?) or it was an ambush (you raised it, made it awkward, and vanished).

Useful feedback has a second beat. A week or two later, you name the change you saw: "The spec for the new service had every error case — that's exactly the thing." Or, if nothing changed, you ask about it without heat: "I noticed the error cases came up again — what's getting in the way?" The follow-up does two things the original delivery can't. It proves the feedback was about improvement, not judgment, because you cared enough to notice the improvement. And it makes the next piece of feedback cheaper to give and easier to hear, because you've established that feedback here is a loop, not a verdict dropped from a height.

The follow-up is also where the highest-cost feedback failure hides: the feedback you never give at all. The observation you swallow because the moment's awkward, the pattern you let slide because raising it feels disproportionate. That silence has a price — the person keeps doing the thing, the team works around them, and six months later it's a performance problem that a thirty-second comment could have prevented. The manager who gives small feedback early almost never has to give large feedback late.

The mirror: when did your team last give you a real one?

You've just read two thousand words on how to give feedback to other people. Here's the uncomfortable question: when did anyone last give you a straight one?

For most managers the honest answer is "I'm not sure," and that's not an accident. The better you get at giving feedback, the more your team experiences you as the person who evaluates — which is exactly the dynamic that makes them hold back what they see in you. The skill you're building here quietly raises the cost of the feedback flowing the other way. Giving well and receiving well are mirror skills, and almost nobody is good at both by default.

That mirror is what Mirorly is built for. It's the 360 feedback round you run on yourself first — you answer a set of calibrated questions about how you work, send the same questions to the people around you, and read your self-view side-by-side with theirs. Not a verdict from above. Not an HR cycle. A structured way to find out whether the version of you in your head matches the one your team actually works with. The core leadership behaviors template is the place most managers start: twelve questions, your answers first, one link out, and a clear picture of the gap between how you think you're landing and how you actually are.

Spend as much energy learning what your team sees in you as you spend telling them what you see in them, and you stop being the manager who's great at giving feedback and bad at hearing it. That's the whole point.

The principles, in one place

• Stop judging the past; start improving the future. Coaching, not verdicts.
• Know whether you're offering appreciation, coaching, or evaluation — and don't disguise one as another.
• Anchor to a specific behavior in a specific moment, with its concrete impact. Never to character.
• Match the shape to the direction: down (restraint), sideways (shared goal), up (framed as a question).
• The defensive reaction is the feedback working. Absorb it, redirect forward, don't debate it.
• Follow up — name the change, or ask about its absence. Feedback is a loop, not a one-time verdict.
• The feedback you never give is the most expensive kind.
• And learn what your team sees in you as seriously as you tell them what you see in them.