# The Listening Ladder — Debrief Sheet

> Six questions, four failure modes of the debrief itself, one rule: the validator cards are the evidence.
> Use with `the-listening-ladder-guide.md`. The debrief runs the last three minutes of the fifteen.

---

## How to run this debrief

- Keep the cards in hand, not on the table. Every question points at a card someone held.
- Ask **to the trio first**, then to the room. The trio has the freshest evidence; the room has the pattern.
- Where a round produced nothing — no correction, no *not yet*, no silence — say so plainly. A clean round is a round that was skipped, not a round that went well.

---

## What the cards already tell you

1. **Round A misses.** The words came back, the facts did not. That is the cost of one-way listening — the same cost as a handover email nobody checked.
2. **Round B over-reaches.** Naming a feeling is a guess, not a reading. A listener who guesses *and checks* beats one who guesses and states.
3. **Round C goes quiet.** The pair who sat in the unsaid gap without rushing to fix it are the ones who actually heard. Most teams skip this rung every working day.

---

## The six questions

**1. The skipped rung.**
"Which rung did your pair climb easiest — and which one did you skip?"
→ Targets: where the pair's listening stops by default. The skipped rung is the one they skip at work; ask what skipping it costs.

**2. The "not yet".**
"Where did the validator say *not yet* — and what changed before it became *landed*?"
→ Targets: the repair move, isolated. This is the strongest single indicator in the debrief: if the card flipped and nothing changed, the validator went polite.

**3. The unsaid question.**
"What is the one round-C question you will ask in a real conversation this week?"
→ Targets: commitment, singular. One question survives the week; a list of five does not.

**4. The correction.**
"Speaker: when the listener named your feeling badly, what did it cost you to say 'no, not quite'?"
→ Targets: whether the room can correct a hearing without escalating it. If nobody corrected anything across three rounds, the exercise was theatre — say so.

**5. The held silence.**
"Listener: what was it like to ask your round-C question and then not fill the quiet?"
→ Targets: tolerance for the gap. Watch for the ones who laugh it off; that is usually the person who solves too early at work.

**6. The cost of rung one.**
"Think of the last thing someone explained to you at work that you heard as facts only. What did you miss?"
→ Targets: transfer off the table. Ask for a named case, not a principle — a principle means nobody is volunteering.

---

## Failure modes of the debrief itself

| Failure mode | Why it happens | The rescue |
|---|---|---|
| The debrief turns into advice-giving | Someone heard the gap and jumps to fix it | Rule: report what you heard, not what they should do. Advice is round-A listening in a suit. |
| The validator says "good job" | Politeness hides the miss | Only two faces on the card — landed, or not yet. Ask which *not yet* they are still not saying. |
| Round C stays silent for the wrong reason | Nobody has a question to ask | Model one aloud: "What didn't you say that I should know?" Then wait the full five seconds. |
| Feelings get treated as diagnoses | An over-confident paraphrase lands as truth | Always close round B with "have I got that right?" and let the correction be boring, not brave. |

---

## Scoring cues (assessment-type sessions)

If you ran the pre/post measure (`the-listening-ladder-assessment.json`), these observables during the rounds are what justify the movement in the report:

| Observable | Points to |
|---|---|
| Listener restates before replying in round B | Paraphrase becoming a habit, not a rule |
| Speaker corrects a feeling and the pair stays warm | The room can survive being misheard |
| Round C's question is followed by silence, not solution | Tolerance for the unsaid |
| Validator says *not yet* at least once | The check is real rather than ceremonial |

---

## Reading the pre/post

Score at pair level only (see `the-listening-ladder-assessment.json`). Report `pair_delta` and `construct_deltas`, honour `report_note`, and never rank individuals — the exercise measures the quality of the listening loop, not the sharpness of anyone's ear.