“Any feedback for me?” is the most common question in professional life and one of the least productive. It arrives at the end of a meeting, when the other person is already packing up, in a form that offers them two options: say something nice, or start an argument. They pick nice. You leave with “no, it was great, really good work” and no idea what to change.
The failure is not that people are dishonest. It is that honest feedback is expensive to give and you have not paid for it. It costs the other person social risk, cognitive effort, and time. Vague praise costs nothing. If your ask does not lower that price, you will get the cheap answer every single time.
This playbook lowers the price. It is seven steps, takes about forty minutes of preparation the first time and ten minutes thereafter, and it works on managers, clients, colleagues, editors and customers.
Step 1: Narrow to one behaviour, not your whole self
“How am I doing?” asks somebody to evaluate a person. That is an enormous request and the polite answer is always “fine.” “In Thursday’s pricing call, what did I do that made the client push back on scope?” asks somebody to recall a scene. That is a small request with a specific answer.
Before you ask anyone anything, write one sentence in this shape:
The frame: “I am trying to get better at [specific behaviour] in [specific situation]. I think the thing holding me back is [your current hypothesis].”
The hypothesis matters more than it looks. It gives the other person something to disagree with, and disagreeing with a stated theory is socially far cheaper than volunteering a criticism from a standing start. You have converted “tell me my flaws” into “correct my analysis,” which most competent people find almost irresistible.
Bad: “How did the presentation go?”
Good: “I’m trying to get tighter at handling pushback in exec reviews. My theory is I over-explain when someone challenges a number, and it reads as defensive. Does that match what you saw?”
Step 2: Pick three people, and pick them deliberately
One source of feedback is an anecdote. Three is a signal. But not any three — you want three different vantage points on the same behaviour:
- The person above the work. Your manager, the client, whoever judges the outcome. They see whether it landed.
- The person beside the work. A peer who was in the room. They see how you operated, and they are the only one with nothing riding on being kind about it.
- The person downstream of the work. Whoever had to use, fix, sell or live with what you produced. They see what it cost after you left.
If you can only get one, take the peer. Managers grade against expectations they may never have told you. Downstream users describe symptoms. Peers describe behaviour, which is the only thing you can actually change on Monday.
Step 3: Use the ask that defeats politeness
Three scripts. Use them close to verbatim; the specificity is doing the work.
The one-thing ask (for a manager or client, in person):
“I want to get better at running these reviews, and I’d rather hear the awkward version than the polite one. If you had to name one thing I should do differently next time — not a list, just the single highest-value one — what would it be?”
The replay ask (for a peer who was present):
“Can I walk you through the moment it went sideways? At the point where Dana asked about the timeline, I did X. What did that look like from where you were sitting? What would you have done instead?”
The advance ask (before the work, which is where the best feedback lives):
“Here’s the draft. I’m specifically unsure about the second section — I think it’s too long but I can’t see what to cut. Don’t worry about typos, I’ll catch those.”
Notice what all three do: they name the thing you want examined, and they explicitly waive the thing you do not. Telling someone what not to review is the single most underused move in this entire playbook. It frees them to spend their limited attention where you need it.
Step 4: Ask the follow-up. Everyone stops one question too early
The first answer is almost always a summary. The second answer is the feedback. When they say something — even something vague — you get exactly one job: make it concrete before you respond to it.
- “Can you give me the moment you’re thinking of?”
- “What would the better version have sounded like?”
- “Is that a one-off, or do you see it a lot?”
- “What did you think I was thinking when I did that?”
- “What’s the cost of not fixing it?”
That last question is the sorting mechanism. If they cannot name a cost, the note is a preference, not a problem, and you are free to ignore it. Half of all feedback fails this test. Discarding it is not arrogance; it is the only way to have room for the half that matters.
And one rule for the whole conversation: do not explain yourself. Not once. The instant you say “well, the reason I did that was…” you have taught the other person that criticism triggers a defence, and they will stop. If you must say something, say “that’s useful, keep going.” You can evaluate later, alone, when nobody is watching your face.
Step 5: Triangulate before you act
Three people, one behaviour. Now apply the rule:
- All three name it: it is real and it is visible. Fix it first.
- Two of three name it: it is real but situational. Work out which situation, then fix it there.
- One names it: it is a data point about that relationship, not about you. Log it. Do not restructure your working life around it.
The single-source note is where most people lose months. One senior person says something offhand, it stings, and it becomes the organising principle of the next quarter. Triangulation is the cheapest insurance available against optimising for one person’s taste.
Step 6: Convert the note into a behaviour and a proof
Feedback that stays as an adjective does nothing. “Be more concise” changes no meetings. Every note you decide to keep gets translated into this template:
Behaviour: In [situation], instead of [what I do now], I will [specific replacement action].
Proof: I will know it worked when [observable thing someone else could verify].
Check date: [30 days out].
Worked example, from the real note “you over-explain when challenged”:
Behaviour: When someone challenges a number in a review, instead of walking through the derivation, I will answer in one sentence and stop talking. If they want the derivation they will ask.
Proof: In the next four reviews, no follow-up question from me on a challenged figure until the challenger has spoken again. Ravi is watching for it.
Check date: 16 September.
That is a change you can succeed or fail at. “Be less defensive” is not.
Step 7: Close the loop — the move almost nobody makes
Thirty days later, go back to the person who gave you the note and tell them what you did with it. Two sentences:
“You told me in July that I over-explain under challenge. I’ve been answering in one sentence and stopping since then. Has it landed differently, or am I fooling myself?”
This does three things at once. It verifies the fix with the only person qualified to verify it. It tells them their honesty was not wasted, which is the single strongest predictor of whether they will be honest with you again. And it makes you, permanently, the person on the team who is safe to tell the truth to — which is a compounding professional asset that costs one email a month.
The four ways this fails
- You ask everyone. Ten sources produce noise and contradiction, and contradiction is a licence to change nothing. Three, chosen for vantage point, beats ten chosen for availability.
- You ask at the wrong moment. Immediately after a bad outcome you are asking for consolation and they know it. Wait 48 hours — long enough that neither of you is managing the other’s emotions, short enough that the detail survives.
- You collect instead of change. Feedback-gathering feels like progress and is not. A note you have not converted into a behaviour and a check date is a note you have not received. Cap yourself: two active behaviours at a time, no more.
- You only ask when things go badly. The most valuable feedback of your career is about the thing that worked, because that is the thing you will try to repeat without knowing which part carried it. Run this playbook after a win at least as often as after a loss.
The ten-minute version
- Write the one-sentence frame: behaviour, situation, your hypothesis.
- Name three people — above, beside, downstream.
- Ask each of them one scripted question. Waive what you do not need reviewed.
- Follow up until you have a moment, a better version, and a cost.
- Say nothing in your own defence.
- Triangulate: three votes act now, two votes act narrowly, one vote is a log entry.
- Convert to behaviour + proof + check date.
- Diarise the 30-day loop-close.
The people who improve fastest are almost never the ones with the most talent or the harshest bosses. They are the ones who have made it cheap for other people to tell them the truth, and then made it obvious that the truth got used. Both halves are a system, not a personality trait. This is the system.