Most bad feedback isn't dishonest - it's too vague to be useful. "You've seemed disengaged lately" is honestly meant and genuinely unhelpful, because it gives the other person nothing concrete to change. Structured feedback models exist specifically to fix this, by forcing feedback into a shape specific enough to act on.
The core structure
Situation: name the specific context - when and where. "In yesterday's client call" is a situation. "Lately" is not.
Behavior: describe exactly what was observed, in terms specific enough that a video camera could have recorded it. "You interrupted the client twice before they finished their question" is behavior. "You came across as impatient" is already an interpretation, not an observation - useful information, but a different kind, and worth keeping separate from the behavior itself.
Impact: name the actual effect - on the work, the team, the relationship, or the person giving the feedback. "The client had to repeat the question, and we lost a few minutes of the call to it" is a concrete impact. This is the step that answers the unspoken question underneath most feedback: why does this matter?
Why separating these three actually helps
Bundled together, feedback tends to collapse into a verdict: "you were impatient with the client and it was unprofessional." Separated, each piece is individually harder to argue with - the situation is a fact, the behavior is a fact, and the impact is at most a fair inference from a fact, rather than a character judgment dressed up as an observation. This is also what makes structured feedback easier to receive: there's a fact to engage with, not just a label to defend against.
Common variants worth knowing
Impact on you vs. impact on the work. Some versions of this model distinguish your own reaction ("when you did X, I felt Y") from the effect on the actual work or team ("when you did X, the deliverable was late"). Both are legitimate, and conflating them can weaken feedback that's really about work quality by making it sound like a personal preference.
Adding a fourth step: the ask. Several variants extend the structure with an explicit fourth element - what you'd like to see differently, stated as concretely as the behavior itself was described. Without it, structured feedback can accurately describe a problem while leaving the other person to guess at the fix.
Where this breaks down in practice
Treating it as a script instead of a structure. Feedback that mechanically announces "the situation was X, the behavior was Y, the impact was Z" in that exact cadence reads as stilted and can feel more like being handled than being talked to. The three components matter; saying them in a memorized formula generally doesn't help and can actively undercut the feedback's sincerity.
Skipping straight to impact. Impact without a specific behavior attached collapses back into vague feedback with extra steps - "this really affected the team" still doesn't tell anyone what to do differently. The behavior step is the one doing the real work; situation and impact exist to make it land, not to replace it.
Confusing behavior with motive. "You didn't seem to care about the deadline" describes an inferred internal state, not an observed behavior - a common and easy mistake, since motive-language often feels more natural than pure behavioral description. "The deliverable came in two days after the agreed date" is the behavior; whether the person cared is a separate, unverifiable question that structured feedback deliberately avoids answering.
For a fuller script built on this same behavior-and-impact core, specifically for the harder case of holding someone accountable rather than just describing what happened, see accountability without humiliation.