Nine ways to rank a backlog, and a guide to which one fits.
A reference your assistant can draw on when you need to prioritize. It holds the formulas and when-to-use notes for nine frameworks, recommends Opportunity Score for customer problems and ICE or RICE for ideas, and starts from one rule: rank problems, not features.
Original instructions by Pawel Huryn (phuryn/pm-skills), preserved byte for byte.
The nine frameworks in SKILL.md. Underlined in red: the three it marks as recommended.
What it is
A reference guide, not a fixed report template. It gives your assistant the formulas, scales and caveats for each framework so it can pick one for your situation and apply it correctly.
It teaches three related Opportunity Score formulas (current value, opportunity, customer value created), the ICE and RICE scoring rules, and a comparison table for all nine. It also links four ready-made templates for Opportunity Score, ICE and RICE.
Who it is for
- Product managersranking a backlog or a set of customer problems.
- Founders and small teamswho need a quick, defensible order.
- Teams comparing methodssuch as RICE versus ICE.
- People learning prioritizationwho want formulas with context.
Works well for
- Choosing a framework for a given kind of decision
- Scoring problems with Importance and Satisfaction data
- Ranking ideas with ICE or RICE
- Explaining why two methods disagree
Not designed for
- Producing real scores without your survey or effort data
- Roadmap dates, resourcing or delivery planning
- A mandated report format; output shape depends on your request
- Collecting customer survey responses for you
What you bring, what you get
- The list to rank: problems, ideas, tasks or requirements
- Importance and Satisfaction scores, if you have them
- Reach, confidence and effort estimates, if you have them
- The decision the ranking should support
- A recommended framework, with reasons
- Correct formulas and scales for it
- Scored rankings, with assumptions marked
- Where methods disagree, and pointers to templates
Where it runs
Any AI assistant that can read a Skill file, including Claude Code and Codex. Download the free plaintext package from Finch and install it with your assistant’s Skill method, or point the assistant at SKILL.md.
No extra API key, script or service. You still need access to an assistant model, which may carry its own subscription or usage cost. The template links open third-party documents; the Skill does not fetch them.
Limits and licence
Scores are only as good as the inputs. Without survey or effort data, any ranking is illustrative and should be rerun with real numbers.
MIT licence, Copyright (c) 2026 Pawel Huryn. Source: phuryn/pm-skills at commit 2a12295a7717, path pm-execution/skills/prioritization-frameworks. The package adds a README, NOTICE, provenance note and a promotional cover; the instructions are unchanged.