Giving and Receiving Feedback

Learning outcomes

A teammate ships a pull request that mostly works but quietly doubles the database load. You see it in the retro. Two voices fight in your head. One says be nice, keep the peace, mention it gently or not at all, because they worked hard and you do not want to seem harsh. The other says just tell them it is wrong, they are an engineer, they can take it. Most people pick one of those two voices and feel virtuous about it, and both produce bad feedback: the first is so soft the problem never lands, the second is so blunt the person stops listening. The skill is not choosing between kindness and honesty. It is doing both at once, in the same sentence, out loud, to a real person who can hear you breathe.

After studying this page, you can:

• Place feedback on the care and challenge grid, and name the failure mode when either dial is turned down.
• Replace vague praise and vague criticism with feedback that is specific about the behavior and sincere about why it matters.
• Deliver spoken feedback with situation, behavior, and impact, so it stays about an action rather than a character.
• Choose the right setting and timing for praise and for criticism, and lower defenses with a short permission ask.
• Receive hard feedback without defending, by listening, assuming good intent, asking for specifics, thanking the person, and only then deciding what to act on.
• Give your manager upward feedback framed around impact and a request, and solicit feedback actively instead of waiting for it.

Before we dive in

Written feedback, the kind you put in a review document, gives you time to choose every word and a reader who cannot interrupt. Spoken feedback gives you none of that. You are in a one-on-one, a retro, or a hallway, the words are leaving your mouth in real time, and the other person’s face is reacting before you finish the sentence. The written companion to this page, performance reviews and written feedback, can lean on careful drafting. This page is about the live version, where tone, timing, and the other person’s defenses are in play, and where most people, including most senior engineers, are surprisingly bad.

They are bad in one of two opposite ways, and it helps to name them before we go further, because the whole page is built to keep you out of both. The first failure is being too soft to be useful: you care about the person, you do not want to upset them, so you soften the message until the actual problem disappears and they walk away thinking everything is fine. The second is being too harsh to be heard: you are right about the problem, you say it plainly and coldly, and the person gets so defensive or hurt that the correct content bounces off them. Soft feedback is comfortable to give and useless to receive. Harsh feedback feels honest and lands as an attack. Good feedback is the narrow path between them, and it is a learned skill, not a personality trait.

One reassurance before the techniques. Giving good feedback is a senior skill precisely because it is hard and because almost no one is taught it. If you find this awkward, that is not a sign you are bad at it; it is the normal starting point. The moves below are concrete and learnable, and they work for the everyday channels an engineer lives in: the one-on-one with a peer or report, the retrospective, peer feedback in a review cycle, and the careful case of feedback aimed upward at your own manager.

Breaking it down

1. Care and challenge: the two dials, and the four ways feedback goes wrong

Start with the single idea that organizes everything else, because once you see it, every later technique is just a way to keep both dials up. Feedback is governed by two independent things at the same time. The first is how much you care about the person: whether you treat them as a colleague whose growth you are invested in, or as an obstacle. The second is how directly you challenge: whether you say the hard thing clearly, or dodge it. These are separate dials. You can care a lot and challenge weakly, or challenge hard while caring not at all, and the combination you land on decides whether your feedback helps.

The reason this is the master frame, the one Kim Scott named radical candor, is that the good zone is the one almost no one’s instincts point them to. Caring pulls you toward going easy. Challenging pulls you toward being blunt. The instinct is to treat them as a trade-off, to dial one down as you turn the other up. The whole insight is that they are not a trade-off: the best feedback turns both dials all the way up at once, caring personally about the person and challenging them directly in the same breath. Turn either one down and you fall into a named failure mode.

Read the grid by its corners. The corner you want is high care and high challenge: you tell the person the hard truth clearly, and they can feel that you are on their side while you do it. Drop the challenge dial while keeping care high and you get ruinous empathy, the most common failure for kind people: you are so worried about hurting them that you soften the message into nothing, and your niceness quietly lets them keep failing. Drop the care dial while keeping challenge high and you get obnoxious aggression: you say the true thing, but coldly and personally, so it lands as an attack and they stop listening. Drop both and you get manipulative insincerity: vague, political non-feedback that protects you and helps no one, the praise you do not mean and the criticism you never quite say. Three of the four corners fail. Only the top one helps, and it is the one your instincts resist.

2. Specific and sincere, not vague

The care-and-challenge frame tells you to aim for the top corner. This section is the first concrete tool for getting there, and it applies to praise and criticism equally: be specific and sincere. Specific means you name exactly what happened and why it mattered, not a vague label. Sincere means you actually mean it, and the person can tell. “Great job” fails on both counts: it names nothing, so the person cannot repeat it, and it is so generic it reads as a reflex rather than a real reaction.

Vague praise feels nice and teaches nothing. “Great job on the launch” leaves the person with a warm feeling and no idea what to do again next time. The specific version hands them something they can repeat: “The way you wrote the rollback plan before we shipped meant we recovered in five minutes when the cache broke, instead of scrambling. That preparation is exactly what I want more of.” Now they know which behavior earned the praise, why it mattered, and that you were actually paying attention, which is what makes praise feel sincere rather than automatic. Specific praise is also more useful than vague praise precisely because it is reinforcing a behavior, not just a mood.

Vague criticism is worse, because it stings without teaching. “You need to communicate better” gives the person nothing to act on and a bruise to nurse. Specific criticism replaces the label with the behavior and the consequence: “In stand-up this week, the updates jumped straight into the technical detail before the headline, and I watched two people lose the thread. Leading with the one-line status would fix that.” Same problem, but now it is a fixable behavior with a reason, not a verdict on their character. The calibration of how plainly to say this, and how much to soften it, depends on the person and the culture, which is the subject of directness across cultures; but specificity is non-negotiable in every culture, because vague feedback is unusable everywhere.

3. SBI spoken: keep it about the behavior, not the person

Specificity needs a shape to keep it on track, because under the pressure of a live conversation it is easy to slide from “what you did” to “who you are,” and that slide is what turns useful feedback into an attack. “You are careless” is a verdict on a person. It is hard to hear, easy to dispute, and impossible to act on, because no one can change a trait on command. The fix is a small spoken structure that forces you to stay on the ground of observable behavior: SBI, for situation, behavior, impact.

Name the situation so the feedback is anchored to a real moment, not a general impression. Describe the behavior you actually observed, as a fact, with no interpretation of motive. Then state the impact that behavior had, on you, the team, or the work. The written track develops this same structure in detail for review documents in performance reviews and written feedback; spoken, it is even more important, because in real time the temptation to leap to character is stronger.

The diagram is a pipeline because each stage protects you from the next mistake. Anchoring to a situation stops you generalizing into “you always.” Describing observed behavior stops you guessing at motive (“you do not care about quality”). Naming impact stops the feedback from being a bare complaint and gives the person a reason the change is worth making. Skip any stage and the feedback drifts back toward a character judgment. The structure works for praise too: “In the incident on Friday (situation), you posted a clear status update every ten minutes (behavior), which stopped three of us from duplicating each other’s work (impact)” is far stronger than “great job under pressure,” because the person now knows the exact move to repeat.

4. Logistics: where, when, and the permission ask

Even perfectly worded feedback can fail on logistics, the where and the when, so a few simple rules protect good content from a bad delivery. The first is about audience: praise can be public, criticism should be private. Praising someone in front of the team multiplies the reward and signals to everyone what good work looks like. Criticizing someone in front of the team adds public embarrassment to the message, which triggers defensiveness and makes the actual point impossible to hear over the humiliation. The same sentence that lands as helpful in a one-on-one lands as a slap in a group channel.

The second rule is about timing: feedback should be timely. Feedback given the day of the deploy, while the details are fresh and the person can still act on them, is worth far more than the same feedback saved for a review three months later, when the moment has blurred and nothing can be done about it. Timely does not mean instant; if you are angry, or the person is in front of others, wait for the right private moment soon, not the review cycle next quarter.

The third rule lowers defenses before you start. Asking permission, or framing the feedback as help, signals that you are on the person’s side and gives them a small amount of control, which makes them less likely to brace for an attack. A short opener does most of the work: “Can I share an observation?” or “Want some feedback on the rollout?” The other person says yes, and now they are receiving by invitation rather than being ambushed. How heavily you buffer all of this, the openers, the softeners, the warmth, depends on the person and their culture, which is exactly the calibration taught in directness across cultures.

Read down the rewrite column and the pattern repeats: every strong version names a concrete behavior, states why it mattered, and is delivered in a setting that lets the person actually hear it. The principle column is the checklist; a feedback line that breaks one of these rules is the line that fails in practice, however true its content.

5. Receiving feedback: the harder skill

Giving feedback well is hard. Receiving it well is harder, and it is the skill that separates people who keep growing from people who plateau, because the feedback you cannot hear is the feedback you cannot use. The difficulty is not intellectual; it is that criticism, even gentle criticism, feels like a threat to your sense of yourself as competent, and the body’s first response to a threat is to defend. The whole skill is learning to override that first response long enough to actually take in what is being said.

There is a sequence that works, and the order matters. First, listen without defending. When someone gives you hard feedback, the reflex is to explain, justify, or correct them mid-sentence (“well, the reason I did that was…”). Suppress it. Let them finish. The moment you start defending, you stop listening, and you signal that feedback to you is an argument to be won, which teaches people to stop giving it. Second, assume good intent. Read the feedback as an attempt to help, not an attack, even when it is clumsily delivered, because most feedback is a genuine, if awkward, attempt to help, and reading it as hostile guarantees you learn nothing from it. Third, ask for specifics. If the feedback is vague (“you could communicate better”), do not bristle; get curious: “Can you give me an example of where that landed badly?” This turns a stinging label into something concrete you can actually work with, and it signals that you take the feedback seriously. Fourth, thank them, sincerely, because giving you feedback cost them something, and thanking them makes them willing to do it again. Only then, fifth, decide what to act on. You do not have to accept every piece of feedback, but you make that decision after you have fully heard it and calmly, not in the defensive heat of the moment.

The deepest move here is to assume good intent and treat the conversation as a shared problem rather than a verdict to dispute, which is the heart of how Stone, Patton, and Heen teach handling a difficult conversation: shift from defending your position to getting curious about the other person’s. When the feedback itself is about a charged disagreement, not just a behavior, you are no longer in feedback territory but in a difficult conversation, and the techniques for that, the layers of what happened, feelings, and identity, live in difficult conversations.

6. Upward feedback: telling your manager

Feedback aimed at your manager is the highest-risk channel, because of the power gap, so it needs the most care, but it is too valuable to skip: managers often have the least accurate picture of their own impact, precisely because few people are willing to tell them. The two failure modes are staying silent (the safe, useless default) and blurting it out as a complaint (which reads as insubordination and gets you nothing). The path between them is to frame the feedback around impact and a request, not blame.

Framing around impact means you describe the concrete effect of the behavior rather than judging the person, exactly the SBI move from earlier, just pointed upward and with extra buffering. Framing around a request means you end with a specific, reasonable thing you would like, which gives the manager a constructive action instead of a grievance to absorb. Compare the blunt “you micromanage me” with the framed version: “When the design is reviewed line by line before I have a first draft (situation and behavior), I find it hard to build momentum and I second-guess my own judgment (impact). Could we try you reviewing at the draft stage instead, and seeing how that goes? (request)” The same message, but now it is actionable, non-accusatory, and easy for the manager to say yes to. Asking permission first (“Could I share something about how we are working together?”) matters even more upward than sideways, and how directly you deliver it should be calibrated to your manager and your shared culture, per directness across cultures.

7. Soliciting feedback: asking before they offer

The last skill closes the loop, and it exists because of a hard truth about everyone around you: people withhold feedback unless you ask for it. Giving feedback is risky and uncomfortable, so the default is silence, which means the feedback you most need, the blind spot no one mentions, is exactly the feedback that never reaches you on its own. If you wait to be told, you will mostly hear nothing, and mistake that silence for “everything is fine.” Soliciting feedback actively is how you break that silence.

The mistake most people make is asking a question so broad it invites a polite non-answer. “Any feedback for me?” almost always gets “no, you are doing great,” because it is vague and puts the other person on the spot to invent a criticism from nothing. The fix is to ask narrow, specific questions that make it easy and safe to be honest. “What is one thing I could have done better in that review?” presupposes there is something, which gives them permission to name it. “Where did I lose people in that presentation?” points at a specific moment. “If you could change one thing about how I run our one-on-ones, what would it be?” makes the criticism small and bounded. And when someone does answer honestly, the way you receive it, using the listen-assume-ask-thank sequence from earlier, determines whether they ever tell you the truth again. Soliciting and receiving are one loop: ask well, receive well, and people keep feeding you the information you need to improve.

Mental Model: feedback is a gift you are handing to a person you respect

The wrong model, the one that produces both the too-soft and the too-harsh failures, is that feedback is a judgment you deliver: a verdict you pronounce on someone’s work or, worse, on them. Under that model, giving feedback feels like an accusation, so kind people soften it into nothing and blunt people swing it like a hammer, and receiving feedback feels like standing trial, so you defend. The judgment model poisons both ends of the exchange at once, because a judgment is something done to a person, and people resist being judged.

The better model is that feedback is a gift you are handing to someone you respect, information they could not get any other way, offered because you are invested in their growth. A gift changes everything about how you give it. You do not hurl a gift; you offer it with care, which is the high-care dial. But you also do not give a useless gift to be polite; you give the real one, the honest one, which is the high-challenge dial. The whole care-and-challenge frame falls naturally out of this single picture: a gift, given by someone who is genuinely on your side, that happens to contain a hard truth.

It changes receiving just as much. When someone hands you feedback, even clumsily, even harshly, they are handing you a gift: information about how you are landing that you could not see from inside your own head, and that most people were too uncomfortable to give you at all. You do not argue with a gift. You take it, you thank the person, you assume they meant well, and later, in private, you decide what to keep and what to set aside. That separation, the feedback is about a thing you did, not a verdict on who you are, is what lets you receive a hard truth without your ego treating it as a wound. Hold the gift model on both ends, and the soft, the harsh, and the defensive failures all dissolve, because none of them survive the simple act of giving and receiving information with respect.

Common mistakes

The first mistake is vague praise: “great job,” “nice work,” “you are crushing it.” It feels generous and teaches nothing, because the person cannot tell which behavior earned it or repeat it on purpose. Name the specific thing they did and why it mattered, so the praise reinforces an action rather than a mood.

The second mistake is harsh or personal criticism: stating the problem coldly, or aiming it at the person’s character (“you are careless,” “you are bad at communicating”) instead of a behavior. It triggers defensiveness and the true content bounces off. Use situation, behavior, impact to keep it about a fixable action, delivered as someone on their side.

The third mistake is ruinous empathy: caring so much about not hurting the person that you soften the message into nothing and let them keep failing. Kindness that withholds the truth is not kind. Keep the care, but turn the challenge dial back up and say the hard thing clearly.

The fourth mistake is public criticism: correcting someone in a meeting or a team channel, where the embarrassment drowns out the point. Praise in public; criticize in private, in a one-on-one, where the person can actually hear you.

The fifth mistake is defensiveness when receiving: explaining, justifying, or arguing the moment feedback arrives, which stops you listening and teaches people to stop telling you the truth. Listen fully, assume good intent, ask for specifics, thank them, and decide what to act on only afterward.

The sixth mistake is never soliciting feedback: waiting to be told and mistaking the resulting silence for “everything is fine.” People withhold feedback unless asked. Ask narrow, specific questions that make honesty easy, and receive the answers well enough that people keep giving them.

Mastery Questions