AID vs SBI feedback models: which one to reach for

Put them side by side and they look like the same model with the letters shuffled. SBI is Situation, Behaviour, Impact. AID is Action, Impact, Desired outcome. Both keep you off the person's character, both take about fifteen seconds to say, and both turn up in the same management training. But they start in different places and, more importantly, they stop in different places — and that ending carries an assumption most people never notice they're making.

What AID actually is

A — Action. What the person did. Not a summary of how they generally operate: the thing itself. "You reordered the roadmap priorities in the shared doc before the review."

I — Impact. What it led to. Same move as SBI's third letter, and the one people most often skip. "I ended up contradicting my own slide in front of the client."

D — Desired outcome. What you'd like instead, next time. "Going forward, drop a line in the channel whenever that doc changes after Friday." This is the letter that makes AID a different tool rather than a shorter SBI.

AID has no single canonical origin the way SBI has the Center for Creative Leadership — it's a coaching-training staple that turns up in leadership programmes under slightly different expansions ("desired behaviour", "development"). That's worth knowing mostly because it means nobody can tell you you're doing it wrong.

The two real differences

AID drops the anchor to a moment. SBI's Situation exists for one reason: it forces the conversation to be about a specific Tuesday rather than about how someone generally is. AID begins at the Action, and in practice you either smuggle the moment into that first sentence or you lose it — and losing it is how "you reordered the priorities before Tuesday's review" quietly becomes "you change things without telling people", which is a claim about character wearing an action's clothes. If you use AID, put the moment back in yourself. Nothing in the acronym reminds you.

AID ends forward instead of stopping. SBI's silence at the end is deliberate — it hands the floor over. AID uses that space to state what you want, which is genuinely the more efficient move when the fix is obvious and undisputed, and a problem when it isn't.

The same moment, both ways

SBI: "In Tuesday's roadmap review, the priorities in the shared doc had changed since Friday and I hadn't seen the edits, so I contradicted my own slide in front of the client."

AID: "You reordered the roadmap priorities before the review without flagging it, and I contradicted my own slide in front of the client. Going forward, drop a line in the channel whenever that doc changes after Friday."

Read them back to back and the trade is clear. SBI leaves a silence the other person will fill — possibly with the reason ("the client emailed Sunday and I didn't want to leave it stale"), which is information you'd want before prescribing anything. AID fills that silence yourself. If the fix really is "tell me when you change it", the silence was wasted time. If there was a reason, you've just issued an instruction that solves the wrong problem, and the other person now has to choose between following it and reopening the point.

The assumption inside the D

Every model assumes something. SBI assumes the observation is worth making. AID assumes you already know what should happen instead — and that assumption is safe far less often than it feels.

It's safe when the desired outcome is procedural: flag the change, copy me on the client thread, run the reconciliation before it ships. It's shaky the moment the fix involves judgement the other person has more of than you do. The engineer knows why the migration was staged that way; the person who ran the workshop knows why they cut the exercise. Prescribing an outcome over their reasoning is how managers accumulate a reputation for having opinions about work they didn't do — and it's a fast route to the pattern where the same problem keeps coming back, because the fix addressed a cause that was never the cause.

The tell is simple: if you can't say why your desired outcome is better in one sentence, you don't have a D yet. Stop at Impact, ask, and let the fix come from the conversation.

Which to reach for

Reach for SBI when:

• The facts might be disputed, or the person hasn't accepted the observation yet
• You're going sideways to a peer or upward to your manager, where prescribing an outcome isn't yours to do
• You don't yet know what should change — or you suspect they know better
• The person tends to go defensive, and you want the first thing they hear to be checkable rather than directive

Reach for AID when:

• The observation genuinely isn't in question and the useful part is the direction
• You're coaching a report who's asked what to do differently — the D is the answer to their question
• It's small, procedural and recurring, where a conversation would be heavier than the thing deserves
• You're following up on something already discussed, and repeating the diagnosis would be condescending

That last case is the one AID earns its keep on. Gallup's workplace research has been making the same point for years — what separates managers whose feedback works isn't the quality of the annual conversation, it's how ordinary and frequent the small ones are. A model compact enough to use in thirty seconds gets used; a model that requires you to book time doesn't.

The direction of the feedback matters more than most model comparisons admit. Downward, AID reads as coaching. Sideways it reads as presumptuous, and upward it reads as instructing your own manager — which is why DESC exists: it carries an explicit ask rather than a stated outcome, and an ask is something a peer or a boss can decline without a confrontation.

Where it sits in the family

Five models, five different endings — that's the whole map:

• SBI ends on the impact. An observation, handed over.
• AID ends on your desired outcome. An observation plus direction.
• COIN ends on next steps you agree together. Direction, but co-authored — the fuller comparison is here.
• DESC ends on a request and what improves if you get it. Built for sideways and upward.
• STAR ends on the result of a full story. Built for recognition and decision-point lessons.

Pick by the ending you need, and the letters sort themselves out — all five in one table, with the direction each one survives. And whichever you pick, the wording is the easy half — which of the two things is actually broken when feedback doesn't land is usually the relationship, not the acronym. It's also worth noting that none of these five ask for suggestions about the next attempt rather than an account of the last one; that's feedforward, and it's a different act entirely.

Where Mirorly fits

AID's whole efficiency comes from being sure enough to skip the conversation — you name the action, the impact, and what you want instead. Notice what happens when the feedback points the other way. Nobody runs AID on you: your team doesn't state your action, its impact and their desired outcome, because the third step means telling their manager how to behave, and the arithmetic of that is brutal. So the person with the most say over what changes gets the least direction about what should.

Mirorly is the structural workaround. You answer specific behavioural questions about how you work, the people around you answer the same set, responses are aggregated with an anonymity threshold so honesty costs nothing, and the reading is the gap between your answers and theirs. No one has to invent a D to your face. The core leadership behaviours template is the usual starting point, and if you want the giving side sharpened too, a dozen worked rewrites show the same swap running on real moments.

Common questions

The one-line summary

AID is SBI with the ending swapped — it drops the situation that anchors feedback to a specific moment and adds a desired outcome in its place, which makes it the faster tool when the observation isn't in dispute and you genuinely know what should change, and the wrong one when the fix needs judgement the other person has more of; downward it reads as coaching, sideways and upward it reads as instruction, and the honest test before using it is whether you can say in one sentence why your desired outcome is better than whatever they were going to do.