Skip to content
Documentation

RICE prioritization, customer-interview synthesis, PRD template picking

FieldValue
TypeSkill Resource
Source~/.copilot/skills/product/references/rice-and-discovery.md
DescriptionNot specified

Source Content

RICE prioritization, customer-interview synthesis, PRD template picking

Absorbed from the former product-manager-toolkit skill. The working PM’s bench — discovery synthesis, prioritization math, and PRD scaffolding in one place. Helps separate signal from noise and ship outcomes, not features.

Use for

  • Scoring feature requests with RICE (with portfolio analysis).
  • Synthesizing customer interviews into pain points, JTBD, and themes.
  • Picking a PRD template by context (full / one-pager / brief / epic) — hand off to references/prd.md to actually draft it.
  • Sequencing a quarterly roadmap against capacity.
  • Catching feature-factory patterns before they ship.

Examples

  • “RICE these 12 feature requests” → score each on Reach × Impact × Confidence / Effort, flag low-confidence honestly (not inflated), and rank — using scripts/rice_prioritizer.py.
  • “Here are 8 customer interviews — what are the patterns?” → tag pain/goal/behavior/quote, cluster mentions, and surface anything that appears ≥3 times with JTBD framing — using scripts/customer_interview_analyzer.py.
  • “Build me a Q3 roadmap from this backlog” → sequence quick wins against strategic bets, hold ~20% capacity for unplanned, and call out the trade-offs by name.
  • “Should we be using RICE or WSJF here?” → pick the framework that matches the decision being made (RICE for feature picks, WSJF for SAFe shops, MoSCoW for scoping a release).

How I work

  1. Pick the workflow. Prioritization, discovery, or PRD scaffolding — name it before opening a tool.
  2. Prioritize with RICE honestly. Reach × Impact × Confidence / Effort. Low confidence is signal, not a number to inflate.
  3. Synthesize interviews into patterns. 3+ mentions = pattern; pain + severity; JTBD framing; map to opportunities.
  4. Choose the PRD template. Full / lean / one-pager / epic — driven by audience and scope, not habit.
  5. Sequence the roadmap. Mix quick wins with strategic bets; reserve ~20% capacity for unplanned.
  6. Validate before building. Test hypotheses with low-fi prototypes; watch behavior, not stated preference.
  7. Close the loop. Compare actual outcomes to RICE inputs; tune the scoring model.

Who I learn from

Modern product: Marty Cagan (Inspired, Empowered — discovery and delivery are different muscles), Teresa Torres (Opportunity Solution Trees), Melissa Perri (Escaping the Build Trap).

Decision frameworks: Daniel Kahneman (Thinking, Fast and Slow), Annie Duke (Thinking in Bets — score decisions by process, not outcome), Roger Martin (Playing to Win).

Customer learning: Indi Young (Practical Empathy), Erika Hall (Just Enough Research), Steve Blank (customer development is a contact sport).

Self-rubric

  • Outcome, not output — recommendations point at a user outcome, not a feature count.
  • Numbers traced to sources — RICE inputs cite reach data and effort estimates, not vibes.
  • Confidence is honest — low confidence flagged, not inflated to win the prioritization game.
  • Patterns, not anecdotes — interview synthesis names the pattern threshold (≥3 mentions).
  • Trade-offs explicit — what’s deprioritized, and why.
  • A senior PM would accept this. Otherwise revise.

Templates and tools

  • references/rice-and-discovery-frameworks.md — RICE, Value vs Effort Matrix, MoSCoW, ICE, Kano, Customer Interview Guide, Hypothesis Template, Opportunity Solution Tree, JTBD, North Star, HEART, Funnel Analysis, Feature Success Metrics, Product Vision Template, Competitive Analysis, Go-to-Market Checklist.
  • references/prd-templates.md — the template library shared with references/prd.md (full / epic / one-pager / feature brief).
  • scripts/rice_prioritizer.py [input.csv] [--capacity N] [--output text|json|csv] — RICE scoring, portfolio analysis (quick wins / big bets), and a capacity-sequenced roadmap. Run with sample as the input arg to generate a synthetic CSV fixture.
  • scripts/customer_interview_analyzer.py <interview.txt> [json] — extracts pain points, delights, feature requests, JTBD, sentiment, themes, quotes, metrics, and competitor mentions from a transcript.

External references