# The Feedback Digest — Debrief

> The surprised pile is the only part of feedback you cannot get any other way — the debrief is where it stops being a sting and starts being data.
> Use with `the-feedback-digest-guide.md`. The sentence-and-action phase at minute 16 is the compressed version; this sheet is the full one.

## How to run this debrief

- Keep the three piles physically separated and visible. The argument of this tool is the shape of the sort, and re-mixing the piles erases it.
- Spend the debrief in the surprised pile. The "already knew" pile is for compliments; nothing changes there.
- Close on the action card, and make it name which surprised card it follows from.

## Read the piles

- Which pile is biggest? If "Already knew" dwarfs the rest, you may have only collected feedback you were already comfortable with.
- Look at "I'll dismiss." Did each dismissal have a real reason, or is that pile where the hard stuff got set aside?
- Is there anything in "Surprised me" that two different people said? Repetition across sources is the strongest signal you have.
- What did you expect to be surprised by — and isn't there? Sometimes the absence is its own data.

## The eight debrief questions

**1. What would make it true.**
"For each item in the surprised pile: what would have to be true for it to be right, even partly?"
→ Targets: reading surprise as a hypothesis instead of an attack.

**2. Content or source.**
"Is it the content that surprises you, or the source? (Same words from a different person can land differently.)"
→ Targets: separating "this isn't true" from "this shouldn't be said by this person" — the second is its own finding.

**3. The uncomfortable edge.**
"What about this is uncomfortable enough that you'd rather not look? That edge is usually where the blindspot is."
→ Targets: locating the blindspot by its temperature, not its wording.

**4. The friend test.**
"If a friend showed you this exact pile about themselves, what would you tell them?"
→ Targets: the compassion you extend to others but not to your own data.

**5. The honest sentence.**
"The sentence you wrote — is it the honest one, or the one you'd want read aloud? Rewrite it until it's true."
→ Targets: the difference between a summary and an admission; the evidence either supports it or it doesn't.

**6. The action's origin.**
"Does the action follow from the surprised pile, or from the comfortable one? A growth action from the 'already knew' pile is a way of not changing."
→ Targets: tracing the commitment back to a specific surprised card.

**7. The witness.**
"Who will you let hold you to it, and when will you check?"
→ Targets: making the commitment social and dated — feedback that nobody watches reverts.

**8. The next quarter's read.**
"What feedback would you have to collect next quarter to know whether this one shifted?"
→ Targets: pre-registering the re-run; the recurring surprise is the pattern, and its disappearance is the proof.

## Failure modes of the debrief

| Failure mode | Why it happens | The rescue |
|---|---|---|
| No surprises at all | Only safe feedback was gathered, or you're defending | Pull one item you'd have put in "dismiss" and ask what it would mean if it were true. |
| Everything is a surprise | Whole reviews are being read, not single pieces | Split each into one behavior — surprises are usually about a specific thing, not "them." |
| The dismiss pile is huge and reasonless | Judgment is being used as a wall | Force a one-sentence reason per dismissal; where you can't produce one, move it to "surprised." |