You just watched someone give a presentation. It was fine in parts and rough in others. Now they're looking at you, waiting. You do what every article about feedback tells you to do: start with a positive, deliver the constructive criticism in the middle, end with encouragement. The feedback sandwich.
They nod. They say thanks. They give the same presentation next time with the same problems. Your feedback changed nothing.
This isn't because you delivered it wrong. It's because the entire model of feedback — observe a problem, describe it to the person, hope they fix it — has a structural flaw. Describing a problem doesn't create the conditions for solving it.
For 50 years, improv teachers have been giving feedback to performers multiple times per class, every class, every week. The feedback loop in improv is tighter and faster than in almost any other domain. And the best improv teachers discovered something that most feedback advice ignores: telling someone what's wrong doesn't help if they can't feel the alternative.
The Problem with Notes
In improv, post-performance feedback is called "notes." A director or teacher watches a scene and then tells the performers what worked and what didn't. Notes are the most common form of feedback in improv, and they're also the least effective.
Will Hines, one of the most experienced improv teachers in the country, puts it bluntly: "An exercise is better than a note."
Here's why. When you tell a performer, "You're not listening to your scene partner," you've accurately described the symptom. But the performer already knows they should be listening. They aren't not-listening because they lack the information that listening is important. They're not-listening because their cognitive bandwidth is consumed by something else — fear of failure, internal planning, status anxiety. The note adds a new item to their monitoring list ("remember to listen") which, perversely, consumes more of the bandwidth that listening requires.
The note describes the output failure without addressing the input problem.
This translates directly to every feedback context:
- "You need to be more concise in meetings." (The person knows. They're not being verbose by choice — they're overexplaining because they're anxious about being understood.)
- "Your code reviews should be more thorough." (They know. They're rushing because they're overwhelmed with other tasks, not because they forgot that thorough reviews matter.)
- "You should speak up more in group discussions." (They know. They're quiet because the environment doesn't feel safe enough to risk being wrong, not because they need the information that speaking is valued.)
In every case, the feedback accurately identifies the gap between current behavior and desired behavior. And in every case, the person receiving it cannot close the gap with the information alone — because the gap isn't caused by missing information.
The Exercise Approach: Design the Experience
The improv alternative to notes is exercises — structured experiences where the correct behavior emerges naturally from the constraints of the activity, without the performer having to consciously execute the instruction.
Instead of telling someone "listen to your scene partner," an improv teacher runs a mirroring exercise. Two people face each other; one moves, the other mirrors. The exercise makes not-listening physically impossible — you can't mirror someone without paying exquisite attention to them. After five minutes of mirroring, the performer's body knows what listening feels like. No note could produce that knowledge.
Instead of telling someone "make simpler choices," a teacher runs a one-word scene. Two people build a story one word at a time. With only one word per turn, complexity is structurally impossible. The performer experiences simplicity working. The note is unnecessary — the exercise made the note redundant.
This is the core insight: the best feedback doesn't describe correct behavior. It creates conditions where correct behavior is the only available option. The person experiences the change rather than being told about it. And experienced change sticks in a way that described change doesn't.
Side-Coaching: Feedback in Real Time
There's a second tool in the improv teacher's kit that most feedback models miss entirely: side-coaching. Instead of watching a scene, waiting for it to end, and then delivering notes after the fact, a teacher calls instructions during the scene while it's happening.
"Slow down." "Look at your partner." "Say less." "Let that land."
Side-coaching works because it intervenes at the moment the behavior is occurring, not after it's been encoded into memory as a completed event. The performer can adjust in real time and immediately feel the difference. "Slow down" during a scene is infinitely more useful than "you were rushing" after it — because during the scene, the performer can experience what slowing down feels like in context.
The analogy in non-improv contexts:
- A manager who sits in on a meeting and passes a note card — "Let Maria finish her thought before jumping in" — is side-coaching. More effective than the post-meeting debrief.
- A writing mentor who reads a draft in progress and says "stay with this paragraph, you're about to run away from the interesting part" is side-coaching. More effective than margin notes on a finished draft.
- A sports coach calling adjustments during practice reps is side-coaching. There's a reason coaches don't wait until after practice to give all their feedback at once.
Side-coaching requires trust. The person receiving it must believe that the coach is on their side and that the interruption is support, not judgment. This is why safety in the room is a prerequisite, not a nice-to-have. Without safety, side-coaching feels like surveillance. With safety, it feels like guidance.
The "Less Is More" Principle
The best improv teachers give fewer notes, not more. This is counterintuitive — wouldn't more feedback produce more improvement? No. And the reason is cognitive load.
A performer who receives eight notes after a scene can't work on all eight simultaneously. They'll either try to (and become so self-monitoring that they're worse than before) or pick one or two and forget the rest (in which case the other six notes were noise). Either way, the volume of feedback degraded rather than enhanced the performer's ability to improve.
The improv principle: one note. The most important one. And then an exercise that trains it.
This maps directly to effective feedback in any domain:
- Don't list every problem. Identify the one problem that, if solved, would improve the most other things. The "keystone note."
- Deliver it clearly, without softening it to the point of ambiguity.
- Then create an opportunity to practice the correction — ideally immediately, while the context is fresh.
Marcus Buckingham's research at Gallup (First, Break All the Rules, 1999) found that the best managers focused attention on strengths rather than trying to fix every weakness — and when they did address a weakness, they focused on one specific, actionable change rather than a comprehensive critique. The improv approach arrives at the same conclusion from a different direction: fewer notes work better because human cognitive bandwidth can only process one correction at a time.
The Safety Prerequisite
All of this depends on one condition that most feedback advice treats as optional: the person receiving feedback must feel safe enough to be changed by it.
This sounds obvious. It isn't. Consider what feedback actually asks of someone: admit that your current behavior is wrong, accept that someone else has a better understanding of your performance than you do, and change in a direction they specify. That's a high-vulnerability request. In a low-trust environment, the rational response to feedback isn't openness — it's defense.
Amy Edmondson's research on psychological safety (1999) showed that teams with higher psychological safety didn't make fewer mistakes — they reported more, learned faster, and improved more consistently. The safety didn't reduce errors. It made errors available for processing rather than concealment.
The same mechanism applies to feedback: safety doesn't make people more receptive because they're nicer. It makes them more receptive because the cost of accepting feedback — the status hit of admitting imperfection — is low enough that the benefit of improvement outweighs it.
Improv teachers know this intuitively. The first exercises in any improv class aren't about improv skills — they're about establishing safety. Can I look ridiculous here and survive? Can I fail and be supported? Can I be bad at something without being judged? Until those questions are answered "yes" through experience (not through the teacher saying "this is a safe space"), feedback is decorative.
Practical Framework: Giving Feedback That Lands
1. Diagnose the Input, Not the Output
Before giving feedback, ask: what is causing this behavior? If someone is being too verbose in meetings, is it because they don't know conciseness matters (unlikely) or because they're anxious, compensating for status uncertainty, or unclear on expectations (much more likely)?
Address the cause, not the symptom. "I notice you tend to elaborate a lot — are you feeling uncertain about whether your point is landing?" is a different intervention than "try to be more concise." The first one addresses the input. The second describes the desired output and hopes the person can reverse-engineer the path.
2. Design an Experience Before Delivering a Note
Can you create a situation where the person will naturally exhibit the behavior you want, and then name it? This is the exercise-over-note principle.
If someone needs to listen better: put them in a role where listening is the only task. "In the next meeting, your job is to summarize each person's point before the group moves on." They'll practice listening because the role requires it. After the meeting: "Did you notice anything different about your attention?" The experience teaches. Your naming reinforces.
If someone needs to make faster decisions: give them a timed constraint. "You have 60 seconds to choose a direction on this. Whatever you pick, we'll run with it and adjust later." After: "How did that feel compared to your usual process?" The constraint creates the behavior. The debrief creates the awareness.
3. Side-Coach When Possible
If you have the relationship and the context, give feedback in the moment rather than after the fact. A gentle redirection during the behavior is more useful than a thorough analysis after it.
"Hey, I want to make sure I understand your core point — can you give me the one-sentence version?" in the middle of a meeting is a side-coaching intervention that teaches conciseness in real time without ever using the word "concise."
4. One Note, Then Practice
When you must give post-hoc feedback, resist the temptation to be comprehensive. Pick the single most impactful change. Deliver it clearly and specifically. Then create an opportunity to practice it.
"The main thing I'd change: when you present data, lead with the conclusion, then show the evidence. Your current structure builds to the conclusion, which means people are uncertain about your point for the first three minutes. Want to try restructuring the first slide right now and see how it feels?"
One note. Specific. Actionable. Followed by immediate practice.
The Honest Caveat
This framework is designed for developmental feedback — helping someone improve at something they're genuinely trying to improve at. It assumes good faith on both sides: the giver wants to help, the receiver wants to grow.
It doesn't cover:
Performance management conversations where the purpose isn't development but documentation. If you're building a case for termination or delivering a formal warning, the dynamics are different. The note-to-exercise ratio flips — documentation requires specificity and completeness, not minimalism.
Feedback across large power differentials where the receiver has reason to believe that admitting imperfection is dangerous. In these contexts, safety must be built before feedback is possible, and sometimes the power structure itself makes genuine safety impossible. A senior VP who says "I want honest feedback" but has fired people who gave it has made feedback structurally impossible regardless of technique.
Situations where the person doesn't want to improve. Not everyone receiving feedback is a willing learner. Some people are performing competence, not pursuing it. The exercise approach assumes the person will experience the correct behavior and want to replicate it. If they don't want to change, no feedback model — exercise-based or otherwise — will work.
The Underlying Principle
The best feedback doesn't add information. It creates experience. When someone experiences the correct behavior — feels what listening is like, feels what conciseness produces, feels what commitment unlocks — they don't need to be told. They know. The note becomes unnecessary because the body already learned.
The best feedback is the note you didn't have to give.
This article draws on the improv knowledge graph at The Physics of Connection. For the full model of teaching, feedback, and creating conditions for growth, explore the Teaching Improv path, or start with Giving Notes and Side-Coaching.
Sources cited: Edmondson (1999), Administrative Science Quarterly. Buckingham & Coffman (1999), First, Break All the Rules. Johnstone (1979), Impro. Hines, Improv Nonsense Substack. Halpern, Close & Johnson (1994), Truth in Comedy.