Feedback people can actually hear: why the sandwich fails, and one sentence that nearly doubled how many students revised their work

Anyone who has managed people knows the feeling: you give careful, kind feedback, the other person nods and thanks you, and three months later nothing has changed. The problem is rarely that they do not want to improve.

In 1996 the psychologists Avraham Kluger and Angelo DeNisi pooled hundreds of studies on how feedback affects performance. On average, feedback helped. But in more than a third of cases it made performance worse. The biggest difference was whether feedback pointed at the task or at the person. When listeners felt that they themselves were being judged, they spent their effort on defence rather than on fixing the work.

The sandwich does not work

The praise, criticism, praise formula is taught everywhere, and it fails for a simple reason: the listener spots the formula. After the second time, every opening compliment is heard as a warning, and you have just ruined your ability to praise sincerely.

The alternative is not blunter delivery. It is separating three things we usually blend together: what happened, what it affected, and what you want next time.

Three parts, in this order

The Center for Creative Leadership calls the approach below the situation, behaviour, impact model. The name matters less than the order:

• Behaviour: what happened, specifically enough that a camera could have recorded it. Not "you lack initiative", but "the report due Tuesday arrived at 5 p.m. on Wednesday".
• Impact: what it led to. Not "that's unprofessional", but "I had to move the client meeting to next week".
• Request: what you want next time, in the future tense. It is the only part that can still change.

The order matters because the first two parts are facts. The listener cannot argue with them, so there is nothing to defend. The third is your wish, and by then the defences are low enough for it to be heard.

One sentence that got students to revise

In 2014 David Yeager's research team tried something small with American middle-school students. They wrote essays, which teachers marked with their usual comments. Then each essay got a handwritten sticky note. Half the students got a bland one: "I'm giving you these comments so that you'll have feedback on your paper." The other half got: "I'm giving you these comments because I have very high expectations and I know that you can reach them."

Students were invited, not required, to revise. Nearly twice as many revised in the second group. Same comments, same teacher. The only difference was that the listener knew the speaker believed in them.

If you have to tell someone that you are giving feedback for their own good, they probably do not feel that yet.

Ask before you tell

One simple opening changes a great deal: ask how they think it went before you comment. Very often they know. When they say it themselves you no longer need to, and something a person admits always carries more weight than something pointed out to them.

The part decided in advance

Feedback only works when the listener believes you are on their side. That belief is not built in the feedback conversation. It is built in dozens of moments before it: when you defended them to others, when you took your share of the blame instead of passing it down, when you asked how they were without attaching a request.

In other words, the quality of a feedback conversation is mostly settled before it starts. If your feedback keeps sliding off, look at the weeks before it, not only at your wording.