SBI vs COIN vs STAR: which feedback model actually fits

If you've taken any management training in the last decade, you've collected a small alphabet of feedback frameworks — SBI, COIN, STAR, and a few cousins — each presented as the structure that finally makes giving feedback simple. The trouble is that nobody tells you when to use which, so they blur together into a vague sense that feedback should have steps, and you end up using none of them well. The frameworks aren't competitors fighting to be the one true method. They're tools shaped for different moments, and the actual skill is knowing which one the moment in front of you calls for.

Here's the useful reframe: a feedback model's only real job is to stop you from doing the thing that wrecks most feedback — leaping straight to a character judgment ("you're careless," "you're not strategic enough") instead of describing what actually happened. Every one of these frameworks is a forcing function for specificity. Where they differ is in what they're optimized to produce at the end of the conversation. Pick by the ending you need.

What each model actually is

SBI — Situation, Behavior, Impact. Developed by the Center for Creative Leadership, it's the leanest of the three. You name the situation (when and where), describe the behavior you actually observed, and state its impact. "In yesterday's client call (situation), you answered the pricing question before the client finished asking it (behavior), and they went quiet for the rest of the meeting (impact)." That's it. No interpretation, no character verdict — just an observable sequence the other person can recognize and respond to.

COIN — Context, Observation, Impact, Next steps. COIN is SBI with a fourth move bolted on. Context and Observation map roughly to Situation and Behavior; Impact is the same. The addition is Next steps — you don't end on the impact, you end on an explicit agreement about what happens next. It turns a feedback statement into a feedback conversation that closes with a commitment.

STAR — Situation, Task, Action, Result. STAR comes out of behavioral interviewing and captures a fuller arc: the situation someone was in, the task they were responsible for, the action they took, and the result it produced. Because it records the whole story — including what they were trying to do — it's built for capturing performance in full, not for quick course-correction. Its corrective variant, STAR/AR, adds an Alternative Action and the Alternative Result it would have produced — "here's what you did and what happened; here's what you could have done and what that would have changed."

SBI: the default for in-the-moment correction

If you only learn one, learn SBI, because it fits the most common feedback moment: something just happened, it wasn't great, and you want to address it cleanly without turning it into a referendum on the person. SBI's brevity is the point — it's fast enough to use in the hallway or at the end of a meeting, while the situation is still fresh and small. It keeps you anchored to one observed behavior, which is exactly the discipline that stops corrective feedback from sliding into the character attacks that make people defensive.

Reach for SBI when the feedback is timely, specific, and doesn't need a formal plan attached — most day-to-day course-corrections. Its limitation is the flip side of its virtue: it ends on impact and then just stops, with no built-in forward motion. For a one-off nudge that's fine. For a recurring problem, you'll want the fourth step COIN adds.

COIN: when you need a next step or an agreement

Use COIN when the conversation has to go somewhere — a development discussion, a recurring issue, anything where "and here's what we'll do about it" matters as much as the observation itself. By forcing an explicit Next-steps move, COIN prevents the most common failure of well-delivered feedback: the conversation that lands the point cleanly and then evaporates, because nobody named what changes. The next-steps beat also hands the receiver something to do, which lowers the threat — there's a path out of the discomfort, not just a verdict to sit with.

In practice it sounds like: "In the last two sprint reviews (context), the status updates ran long and we didn't get to the blockers (observation), so the team left without decisions on the things that were actually stuck (impact) — let's try a two-minute cap per update and a standing blockers slot, and check next month whether decisions are landing (next steps)." The fourth move is what separates a comment that vanishes from one that changes how next month runs.

The risk with COIN is over-formality. Bolting a "next steps" negotiation onto a tiny piece of feedback ("you mispronounced the client's name — what's our plan?") is absurd and makes you sound like you're running a performance-improvement process over nothing. Save it for feedback that genuinely warrants a commitment.

STAR: when you're reinforcing strong work

STAR is the one most managers underuse, because they think of feedback as exclusively corrective. But the full-arc structure shines for recognition — and specific recognition is one of the highest-return, lowest-cost things a manager can do. "You were handed the migration with two days' notice and half the docs missing (situation/task), you flagged the risks early and shipped a rollback plan nobody asked for (action), and we had zero downtime (result)" is praise the person can actually internalize, because it names exactly what they did right and can repeat. Compare that to "great job on the migration," which feels nice and teaches nothing.

For correction, STAR/AR works when the lesson is genuinely about a decision point — the person did something reasonable, but a different action would have produced a better result, and you want them to see the fork. It's heavier than SBI, so reserve it for moments worth the weight.

Match the model to the moment

The quick version, if you only remember one paragraph: SBI for fast, in-the-moment correction. COIN when the feedback needs to end in an agreement or you're tackling a recurring pattern. STAR to make recognition specific enough to repeat, and STAR/AR when the lesson hinges on a decision point. Notice none of them is about the size of the criticism — they're about the shape of the ending you need: a clean observation (SBI), a commitment (COIN), or a full story for learning or recognition (STAR). Pick by the ending, and the right tool is usually obvious.

Where all three fall short

Here's the part the frameworks don't advertise: none of them solves the actual problem. A model makes you specific and keeps you off the person's character — genuinely valuable, and the reason any of them beats winging it. But specificity is the easy half of feedback. The hard half is whether the receiver trusts that you're on their side, and no acronym supplies that. Kim Scott's Radical Candor frames it as caring personally and challenging directly at the same time — and the frameworks only ever help with the challenging-directly half. Run a flawless SBI on someone who suspects you don't have their interests at heart and it lands as a well-structured attack.

This is also why no model rescues a bad approach — it's the same reason the feedback sandwich fails: structure isn't the lever. The lever is trust plus specificity, and the framework only ever delivers the second. Use SBI, COIN, or STAR to make sure you're describing a behavior instead of judging a person — and then spend your real effort on the thing the acronym can't touch: being someone whose feedback the other person believes is meant to help. (And when the feedback travels up to a direct report or sideways to a peer, the power dynamics shift what "trust" requires — the model stays the same; the relationship work changes.)

The mirror: which model is anyone using on you?

You can get fluent in all three and still be flying blind on the one feedback that matters most — the feedback about you. The better-structured your feedback to others, the more you're the one evaluating, and the less straight input comes back the other way.

That's the gap Mirorly is built to close. It's the 360 feedback round you run on yourself first — you answer a set of calibrated, behavioral questions about how you actually work, send the same ones to the people around you, and read your self-view side by side with theirs. The questions are written the way good feedback is: anchored to specific behaviors, not character verdicts — SBI's discipline, built into the instrument. The core leadership behaviors template is where most managers start.

The one-line summary

SBI, COIN, and STAR aren't rival methods — they're tools for different endings: SBI for a clean in-the-moment correction, COIN when you need a commitment, STAR to make recognition specific or to teach at a decision point; match the model to the ending you need, and remember that every framework only delivers specificity, never the trust that decides whether the feedback actually lands.