What needs to be finished first
A list of 5-30 initiatives, features or experiments of comparable size is available, each with short description of expected outcome.
Current user numbers, conversion rates or other reach-relevant metrics are known to the team so Reach can be estimated realistically.
What needs to be ready before start
Table with columns Initiative, Reach, Impact, Confidence, Effort, RICE score; predefined scales (Impact: 0.25 / 0.5 / 1 / 2 / 3; Confidence: 50%/80%/100%); link to initiative description.
One PM or PO as owner; one Engineering Lead for effort estimates; one researcher or analyst for reach data; optionally designer for impact assessment.
Initiative list with descriptions; known reach metrics (active users per month, funnel numbers); historical impact of similar features if available; rough effort range from past projects.
30-60 min
Create table. Make definitions per factor visible (Reach = number of users per quarter; Impact = quantitative measure on outcome; Confidence = percentage; Effort = person-months). Align team on Impact and Confidence scales before estimating.
The one question this method answers
Which initiative delivers the greatest expected outcome per invested effort, and where is estimation uncertainty so high that the order says little?
Marker: Sektion
| Step | Duration | Action | Hint |
|---|---|---|---|
1Section 1: Calibrate scales | 10 min | Define scale per factor: Reach in absolute users or events per quarter; Impact as multiplier (0.25 minimal / 3 massive); Confidence in %; Effort in person-months. Score one example initiative together for calibration. | If scales remain unclear, participants score in different units. Calibration is mandatory, not optional. |
2Section 2: Estimate Reach | 10-15 min | Per initiative: how many users (or events) does the feature touch per quarter? Use numbers from baseline metrics. If uncertain, use range with mean. | Reach is often overestimated. A feature in Settings does not reach 100% of users. Check funnel data, do not guess. |
3Section 3: Impact and Confidence | 10-15 min | Per initiative: Impact level (0.25/0.5/1/2/3) on outcome goal. Confidence level (50% speculative / 80% tested / 100% proven). If Confidence <50%, park initiative or validate first. | Impact = 3 only with documented evidence (past data, A/B tests). Inflation devalues the scale. |
4Section 4: Effort and score | 10-15 min | Estimate effort per initiative in person-months (including design, QA, rollout). Score = (Reach x Impact x Confidence) / Effort. Sort in table. | Effort is often underestimated. If team historically needs 1.5x longer than estimated, apply multiplier. Otherwise expensive initiatives move upward wrongly. |
5Section 5: Sanity check and sorting | 10 min | Review top-5 ranked initiatives. Gut check: does order intuitively make sense? If not, revisit estimates. Document assumptions per top initiative. | RICE is tool, not oracle. If top-1 is obviously nonsensical, do not enforce ranking; inspect estimates. |
What comes out at the end
Table with initiatives, RICE factors, score and rank; plus assumptions log with source per Reach number and rationale per Impact level; plus list of top-N initiatives for next quarter with owner.
One version per prioritization round with date. For later estimate correction, edit log on entry. After 3-6 months, retrospective: actual outcome vs estimated Impact, derive calibration learning.
Compact working template for RICE with context, input, output artifacts, and next step.
| Item | Description | Rating | Evidence | Owner | Next step |
|---|---|---|---|---|---|
| 1 | |||||
| 2 | |||||
| 3 |
What consequence follows from the matrix?
Concrete filled scenario, fictional example
Concrete filled scenario, fictional example
| Initiative | Reach (users/Q) | Impact | Confidence | Effort (PM) | Score |
|---|---|---|---|---|---|
| Inline validation checkout | 18,000 | 2 | 80% | 1.5 | 19,200 |
| Onboarding tutorial slot 1 | 12,000 | 1 | 100% | 2 | 6,000 |
| Apple Pay integration | 18,000 | 0.5 | 80% | 2.5 | 2,880 |
| Empty-state explainer video | 4,000 | 0.25 | 50% | 1 | 500 |
| AI suggestions in dashboard | 8,000 | 2 | 50% | 6 | 1,333 |
Assumptions log:
Top 3 for Q3: inline validation, onboarding tutorial, Apple Pay. AI suggestions only after validating risky assumptions (Confidence too low).
Recognize symptoms and steer against them
70%+ of initiatives are rated Impact 2 or 3, everything looks important.
Recalibrate scale: 3 only for documented, dramatic effect. 1 is the normal case. Anyone saying 3 must provide evidence.
Reach numbers are round estimates without relation to actual funnel position.
Pull funnel data before workshop. Derive Reach only from baseline. If data is missing, mark initiative as Reach unclear and plan data spike.
Engineering estimates 1 PM, reality becomes 3 PM, initiative rises in score and disappoints.
Have experienced engineer estimate effort, with historical multiplier (for example x1.3). Definition of Effort includes design, QA, rollout, support.
Initiatives with 50% Confidence still go into roadmap without validation step.
Low-confidence initiatives need discovery pre-step (interview, spike, prototype) before they enter score as normal. Otherwise optimistic bias.
Top-1 is implemented immediately without sanity check or strategic context.
RICE is discussion aid, not command. Strategic bets, compliance items or platform investments need separate track next to RICE score.
Done signals checkable in under a minute
Go to the profile for purpose, similar methods, and sources or continue to the next method in the catalog.