Most feedback is delivered to be said, not to be acted on. The giver gets it off their chest, the receiver nods, and three weeks later nothing has changed.
That's not feedback. That's a conversation that produced no change.
Feedback that actually changes behaviour is a craft. It's not about being honest, or being direct, or having the courage to say the hard thing. Those are starting conditions. The craft is in the parts most people skip — being specific, being well-timed, and being interested in whether the receiver actually heard it.
Why most feedback misses
Three reliable failure modes:
It's about the giver, not the receiver. The feedback gets framed around what the giver thinks, what they expected, how they felt. The receiver hears it as a complaint and defends. No behaviour changes because no behaviour was clearly named.
It's vague. "Be more strategic." "Communicate better." "Take more ownership." Each is true in the abstract and useless in practice. The receiver can't act on it because they can't see what specifically they're supposed to do differently.
It's late. The behaviour happened a month ago. The conversation about it is now disconnected from the moment when changing course would have mattered. Even if the feedback is clear, it's harder to use than feedback in the moment.
A piece of feedback that fails any of these usually fails entirely. The receiver leaves the room polite, defensive, or confused. None of those produce change.
What useful feedback looks like
It targets a specific behaviour. Not a trait, not a pattern, not a personality. A specific thing that happened in a specific moment. "In the planning meeting on Tuesday, when X came up, you committed the team to the deadline before checking with engineering" is feedback. "You don't think before committing" is an attack on the person.
It includes the impact. Why does the behaviour matter? Who got affected? What got harder because of it? Without the impact, the receiver can argue the behaviour wasn't a problem. With the impact, the conversation has somewhere useful to go.
It happens close to the event. Not necessarily immediately — you can pick the moment. But days, not weeks. The further from the event, the more the conversation feels archived rather than live, and the less change it produces.
It leaves room for them to talk. The point isn't to deliver a verdict. It's to start a conversation about behaviour. The receiver might have context the giver didn't have. They might disagree, and the disagreement might be reasonable. They might agree and need help figuring out what to do differently. None of that happens if the feedback is delivered as a one-way broadcast.
How to actually do this
A practical structure that works:
Set the frame. "I want to talk about something I noticed — it's not a big deal, but I think it's worth flagging." That sentence does a lot. It signals the size of the thing and gives the receiver permission to receive it without bracing.
Name the specific behaviour. Concrete, recent, factual. No interpretation yet. Just what happened.
Name the impact. What changed because of the behaviour. What was harder, slower, or worse than it could have been.
Ask, then listen. "What's your read on it?" or "How did it look from your side?". Genuinely listen to the answer. The conversation is now a two-way thing instead of a delivery.
Agree on what would be different next time. Specific again. Not "be more careful" — "next time, before committing the team to a date, check with engineering first". The specificity is what makes the feedback actionable.
This whole sequence takes ten minutes and produces more change than an hour of vague critique.
Where feedback usually fails
A few patterns to watch:
The compliment sandwich. Positive, negative, positive. It's been mocked for a reason — most receivers see the structure coming and discount the middle as the only real content. Don't dress up the feedback. Just deliver it well.
The accumulated quarterly review. Saving up six months of small observations for a formal review is the worst possible delivery mechanism for behaviour change. Each observation lands stale and out of context. The receiver feels ambushed. Real-time feedback compounds. Stored feedback explodes.
The feedback that's actually a complaint. Some "feedback" is venting dressed as critique. The giver wanted to express their frustration. They didn't want change. The receiver can usually tell, and the conversation produces resentment instead of action. Be honest with yourself about which one you're delivering.
The harder side: receiving it well
Feedback is a two-sided craft. Giving it well is half the work. Receiving it well is the other half.
Three habits that make you better at the receiving side:
Assume good faith on the first pass. Even if the feedback is delivered badly. Even if the giver got something wrong. Take the part that's true seriously before defending the part that isn't.
Separate the message from the messenger. Sometimes the person delivering the feedback isn't your favourite person, or doesn't have full context, or has an axe to grind. The feedback can still be partially right. Filter for what's true.
Ask for it more often than feels comfortable. The cost of not getting feedback is invisible — you don't know what you don't know. Asking specific questions ("how did the way I ran that meeting come across?") gets you better information than generic "any feedback for me?". Both beat silence.
The shift
Feedback as a transaction is broken. Feedback as a conversation about specific behaviour, with stakes both sides can engage with, is one of the most leveraged tools you have for moving teams.
Make it specific. Make it timely. Make it a two-way thing. Then do it more often.
If you're building a high-performing team, feedback is the connective tissue that makes it possible. And taking feedback without taking it personally is the receiving-side skill that closes the loop.