The Feedback Script (Situation-Behavior-Impact)

The hiring scorecard helped you decide who joins the team. Once they are on it, your most frequent influence on how they work is the feedback you give — the small, steady stream of “do more of this” and “let’s change that” that shapes a person over months. Most managers give this feedback badly, or avoid it entirely, because they have no script and so the words come out vague, harsh, or both.

This page gives you a script. It is called SBI — Situation, Behavior, Impact — and it does one thing well: it turns a vague reaction (“you did great,” “you need to communicate better”) into a specific, factual, low-drama message the other person can actually act on. You will get the template, a short guide to using it, filled-in examples for both praise and correction, and a way to practice this week.

Why feedback needs a script at all

Feedback goes wrong in a predictable way. Under the mild stress of a hard conversation, the brain reaches for shortcuts — and the handiest shortcut is a character label. “You’re careless.” “You’re a great team player.” “You’re not detail-oriented.” Labels feel true and they feel efficient, but they fail on two counts.

First, a label is not actionable. If I tell you that you’re “careless,” what exactly do you do differently tomorrow? You can’t act on an adjective. Second, a label attacks identity, and identity is defended. Tell someone they are careless and their energy goes into arguing that they aren’t — not into changing anything. The conversation becomes a debate about their worth as a person.

SBI fixes both problems by refusing to label. Instead of naming what someone is, you describe what they did and what happened as a result. Behavior can be changed. Facts are hard to argue with. And because you never touch the person’s character, their defenses stay down and they can actually hear you.

The evidence — describing behavior lowers defensiveness

Work on feedback and performance conversations (including research summarized by the Center for Creative Leadership, which developed the SBI model) generally finds the same thing: feedback framed as specific, observable behavior is received far better than feedback framed as judgment or personality. The reason is not politeness — it is that a factual description gives the recipient something to verify and something to change, while a character judgment gives them only something to defend. Keep the figures loose; the direction is robust: behavior beats labels.

The template

The script has three parts, delivered in order. Keep each part short and concrete.

SBI FEEDBACK SCRIPT

1. SITUATION  — anchor it to a specific time and place
   "In [the meeting / the shift / last Tuesday's handover]..."
   → Pin the feedback to one real moment, not "generally" or "always."

2. BEHAVIOR   — describe what you observed, like a camera would
   "...you [did / said / this specific, observable thing]..."
   → Only what a camera or microphone could have recorded.
     No adjectives about their character. No assumed intentions.

3. IMPACT     — state the effect it had (on the work, on you, on others)
   "...and the effect was [what happened / how it landed / what I felt]."
   → The consequence that makes the behavior worth mentioning.

THEN — pause, and invite their view:
   "How did you see it?"  /  "Does that match your read?"

The three parts answer three questions the other person is silently asking: When are you talking about? What exactly did I do? Why does it matter? Skip any one of them and the feedback wobbles — a Situation with no Behavior is a mystery, a Behavior with no Impact sounds like nitpicking, an Impact with no Situation sounds like a general complaint.

How to use it

Anchor to one situation. “You’re always late” invites an argument about the word always. “In yesterday’s 9 a.m. standup” cannot be argued with — you were both there. Specificity is not just fairer; it is disarming.

Describe behavior like a camera. The test for good Behavior language: could a video camera have recorded it? “You interrupted Maria twice before she finished” passes — a camera sees it. “You were dismissive” fails — that is your interpretation of what the camera saw. State the observable thing and let the interpretation stay yours to raise, not yours to assert as fact.

State impact, don’t guess at intent. Impact is what happened as a result: the report went out with an error, the new nurse looked rattled, you felt reassured the shift was covered. What you must not do is narrate their motives — “you did that because you don’t care” — because you cannot observe intent, and guessing at it wrongly is how trust dies. Describe the effect; ask about the cause.

Then hand them the microphone. SBI is the opening of a conversation, not a verdict. After the Impact, stop talking and ask: “How did you see it?” Often there is context you are missing — the interruption was because the meeting was overrunning and a decision was due in five minutes. Inviting their perspective turns a one-way pronouncement into a shared problem-solving session, and it is what separates feedback from a scolding.

Timing and privacy. Give feedback close to the event, while the details are fresh and correctable — not saved up for a quarterly review, where it lands as an ambush. Praise in public is often fine; correction is always private. Correcting someone in front of others adds humiliation to the message, and humiliation is the surest way to make sure they defend rather than change. One more rule: correct in a calm moment, not in the heat of the mistake.

In practice — the camera test in a warehouse

A shift lead wants to tell a picker they are “sloppy.” She catches herself and runs the camera test: what did the camera actually see? It saw three orders last Thursday packed without the fragile-item padding, two of which came back damaged. So she says exactly that — Situation (Thursday’s evening orders), Behavior (packed without padding), Impact (two returns, a frustrated customer). The picker, who braced for “you’re sloppy,” instead hears a fixable, factual thing and says, “Oh — I didn’t know those SKUs counted as fragile.” A label would have started a fight. The behavior started a fix.

Filled-in examples

The script works identically for feedback you want to reinforce (do more of this) and feedback you want to correct (let’s change this). Praise is not an exception to structure — vague praise is nearly as useless as vague criticism.

Reinforcing feedback (a cafe):

SITUATION:  "During this morning's rush, when we had the line out the door..."
BEHAVIOR:   "...you started pre-calling the regulars' usual orders and kept
             the till moving while I made drinks..."
IMPACT:     "...and we cleared the queue in about half the usual time — plus
             three customers told me how smooth it felt. That made a real
             difference. Do that again."

Notice the last line. “Great job today!” would have felt nice and taught nothing. This version tells the person exactly which behavior to repeat, which is the entire point of praise — to reinforce a specific action, not to hand out a warm glow.

Corrective feedback (a software team):

SITUATION:  "In the release call on Tuesday..."
BEHAVIOR:   "...you pushed the deploy live before the QA sign-off had come
             back in the channel..."
IMPACT:     "...and we shipped a bug that reached customers for about an hour
             before we caught it. It also left QA feeling their check didn't
             matter."
THEN:       "How did you see it? Was there pressure I'm not aware of?"

The Impact names two costs — the customer-facing bug and the effect on a teammate — and the closing question opens the door to context. Maybe there was pressure: someone said the fix was urgent. Now it is a conversation about the release process, not an accusation about one person’s judgment.

The common trap — the sandwich, and the vagueness twins

The classic bad habit is the feedback sandwich: bury the criticism between two compliments (“You’re doing great! One thing — the report was late. But really, great work!”). It backfires two ways. People learn to hear the praise as a warning siren for the criticism to come, so your genuine praise gets devalued. And the real message gets so muffled that the person walks away unsure whether anything was actually wrong. Deliver praise and correction as separate, clean messages — each one specific, each one on its own.

Its cousins are the vagueness twins: vague praise (“nice work!”) and vague criticism (“you need to be more proactive”). Both leave the person nothing to grab. Run the camera test on every piece of feedback — if a camera couldn’t have recorded the behavior you’re describing, you haven’t finished writing the message.

Case study — the andon cord as instant, specific feedback

Toyota’s production system, refined through the 1980s and studied worldwide since, gives every line worker an andon cord they can pull the moment they see a defect — stopping the line so the problem is fixed at its source. Strip away the manufacturing setting and it is a feedback principle: feedback works best when it is immediate, tied to a specific observed event, and aimed at the process rather than blaming the person. A defect caught and named on the spot is cheap to fix; the same defect discovered weeks later, attached to a vague complaint about “quality problems,” is expensive and demoralizing. Give feedback like an andon cord — close to the event, about the behavior, in service of the work.

How it connects

SBI is the tool that makes the one-on-one safe to be honest in — a standing private slot is the natural home for corrective feedback delivered calmly. It also feeds the retro: a good retro is a team running SBI on its own process, describing what happened without blaming who. And it is the everyday counterpart to the feedback you gather when hiring with the scorecard — the same discipline of judging observable evidence, not gut impressions.

Try this

Pick one piece of feedback you have been sitting on — praise or correction, whichever you have been avoiding. Write it out in three lines: Situation (one specific moment), Behavior (only what a camera saw — run the test), Impact (the effect it had). Then add the closing question you will ask. Deliver it this week, in private if it is corrective, and notice how the person responds when you describe a behavior instead of hanging a label on them.

Reflect

1. Think of the last feedback you gave. Was it a behavior you described, or a character label you applied? How did it land?
2. Which do you avoid more — giving praise or giving correction — and what does that avoidance cost the person and the team?
3. Have you used the sandwich method? What did the recipient likely walk away actually believing?
4. When you last received vague feedback, what did you wish the person had said instead? Can you give what you wished for?
5. Where in your week is the safe, private place to deliver corrective feedback — and if you don’t have one, what would it take to build it?

Show reflections

1. Most honest answers land on “a bit of both.” The useful move is to rewrite that feedback using the camera test — if you can’t say what a camera recorded, you were labeling, and the person had nothing concrete to act on.
2. Many managers avoid correction (fear of conflict); some avoid praise (assume it’s obvious). Both cost you: unspoken correction lets a fixable habit harden, and unspoken praise lets a good behavior fade for lack of reinforcement.
3. The sandwich usually teaches people to distrust your praise (it signals bad news coming) and to under-weight your criticism (it got muffled). If your answer is “they weren’t sure what I meant,” that is the sandwich working exactly as it fails.
4. This connects the receiving end to the giving end. What you wished for is almost always specificity — a real moment, a concrete behavior, a clear effect. Notice that and you know what to give others.
5. The one-on-one is the obvious answer: a recurring private slot means corrective feedback has a calm home and never has to be delivered as an ambush or in public.