Prioritization · Evolved from RICE / WSJF
Risk-Adjusted RICE
Classic reach × impact × confidence ÷ effort scoring, extended with explicit risk terms for regulatory exposure, model risk, data sensitivity, and reversibility.
Executive Guidance
How to hold this framework as a leader
RICE is a lightweight scoring model. Its role is to make prioritization defensible, not to be right in every case. Executives should treat the RICE score as a starting position for a conversation, and expect leaders to disagree with the score where their judgment is stronger than the inputs.
In AI-era portfolios, unmodified RICE understates the risk of low-reversibility, high-blast-radius work. A risk multiplier that penalizes autonomous or safety-critical AI work is the minimum viable adaptation.
When to Apply
- Prioritizing across a mixed portfolio of AI and non-AI initiatives
- Regulated environments where reversibility materially changes cost
- Roadmap trade-offs requiring executive-visible justification
AI-Era Notes
Add a risk multiplier that penalizes low-reversibility, high-blast-radius, or model-dependent work. Shipping an autonomous capability is not equivalent to shipping a UI change.
Key Trade-off
Provides false precision if confidence scores are unexamined; risk-adjustment for AI work is essential.
Operating Sequence
The order in which to install the framework
- 01For each candidate, estimate reach, impact, confidence, and effort with named evidence.
- 02Apply a risk multiplier for AI-affecting work based on reversibility and blast radius.
- 03Publish the resulting ranking with the underlying inputs, not just the score.
- 04Discuss the top and bottom of the ranked list in a leadership session; do not accept the score without discussion.
- 05Record the final decision and its deviation from the score as an ADR.
Key Artifacts
The documents this framework produces
RICE Ledger
Inputs, evidence, and risk multipliers for every candidate.
Owner: Product
Prioritization Decision Log
Final decisions and their deviation from the ranked score.
Owner: Product Leader
Operating Checklist
What "good" looks like when installed
Inputs
- Every reach, impact, and confidence value has cited evidence.
- Effort is estimated by the team that would build the work.
- AI-affecting candidates carry a risk multiplier reflecting reversibility.
Governance
- The ranked list is visible to affected stakeholders.
- Score deviations are recorded with rationale as decision records.
- The ranking is reviewed at least once per planning cycle.
Common Antipatterns
- False precision in confidence scores
- Prioritizing agent capabilities purely on reach without risk weighting
Boardroom Questions
- Where has RICE told us to do something our judgment says we shouldn't?
- How are we accounting for the blast radius of autonomous work?
- What is the confidence basis for our top-ranked bets?