Product OKR Quarterly Review · Hypothesis Lens

You're my PM OKR retro partner. Use hypothesis lens — not scoring.

# This quarter's OKRs + data
{paste_okrs_with_data}

# Hypothesis behind each KR (you may not have written this explicitly — think it through now)
{hypothesis_per_kr}

# Product context (launches / pivots / major user feedback events)
{quarter_context}

# Output five sections

## 1. Hypothesis scoreboard (70 words per KR)
For each KR:
- Name + score
- The hypothesis (users will do X / metric will move Y / channel Z will contribute)
- Did data confirm / challenge / overturn?
- One sentence: if the hypothesis was wrong, does the KR's "completion" even matter?

## 2. 3 mental model updates (60 words each)
Based on hypothesis tests:
- e.g. "we thought power users would drive new users; reality showed referrals came from medium-engagement users"
- This is the PM's real quarterly value.

## 3. The one thing we mis-judged (80 words)
The most important mis-bet to own — deeper than ops/marketing retros. Must dig into a product-judgment error, not an execution miss.
Include: why we bet that way at the time, what data showed, how to avoid the same class of mistake.

## 4. High-confidence signals (2-3, 40 words each)
Signals from the quarter strong enough to guide next-quarter roadmap:
- Not noise, not outliers — multi-week, multi-cohort, hypothesis-aligned.

## 5. Next quarter's 3 new hypotheses (30 words each)
Based on this quarter's learnings:
- Each: hypothesis statement + measurement + kill criteria (when do we stop betting?)

# Boundaries
- Don't do KR-by-KR scoring — that's an ops retro, not a PM one.
- Mental model updates must be genuinely new, not things you already knew.
- The mis-bet must reach the product-judgment layer, not stop at execution.