The awkward part of prioritization is that most ideas sound reasonable on their own. Simplify onboarding? Sensible. Send reminders? Also sensible. Build a better recommendation engine? Hard to argue against.
RICE gives those ideas a common set of questions: how many people will this reach, how much will it help, how sure are we, and what will it take? I would use it to expose the assumptions behind a roadmap, not to end the conversation.
The formula is:
RICE score = (Reach × Impact × Confidence) / Effort
Use confidence as a fraction in the calculation: 80% becomes 0.8. A higher score suggests more estimated impact per unit of work, provided the inputs are comparable. The method comes from Intercom's published prioritization framework. 1
Define the inputs before you score
Reach is the number of people or events affected within a defined period. Use the same period and unit for every idea. Do not compare monthly users with quarterly transactions.
Impact estimates the effect on each affected person against a shared goal. Intercom's scale uses 3 for massive impact, 2 for high, 1 for medium, 0.5 for low, and 0.25 for minimal.
Confidence expresses how much evidence supports the estimates. Intercom uses 100%, 80%, and 50% as useful bands. A precise-looking score does not make weak evidence stronger.
Effort estimates the combined work across the team. Intercom uses person-months; include relevant product, design, and engineering work, not only implementation time. 1
A RICE prioritization example
Suppose a fictional learning app wants to improve lesson completion. It is comparing three ideas for the next quarter.
All numbers below are illustrative. They are not Aniket's employer data or reported product results.
| Idea | Reach: users/quarter | Impact | Confidence | Effort: person-months | RICE score |
|---|---|---|---|---|---|
| Simplify lesson setup | 1,000 | 1 | 80% | 2 | 400 |
| Add progress reminders | 2,000 | 0.5 | 50% | 1 | 500 |
| Build a new recommendation flow | 800 | 2 | 50% | 4 | 200 |
For lesson setup:
(1,000 × 1 × 0.8) / 2 = 400
The initial order is reminders, lesson setup, then recommendations. That is the start of the discussion, not the final roadmap.
The reminders score depends on reaching 2,000 people. Does that estimate account for notification permissions? Does the expected benefit include the risk of unwanted messages? Could the setup improvement be required before a recommendation feature helps anyone?
If the answer changes when somebody asks about notification permissions, the spreadsheet has done its job. It has shown us where the decision is fragile.
Test how sensitive the order is
Change the reminders' confidence from 50% to 30%. Its score falls from 500 to 300, and lesson setup moves ahead of it.
That does not mean 30% is the correct estimate. It means the original ranking was sensitive to an uncertain input. The next useful action may be a small research task or experiment rather than an immediate commitment to build.
Keep a short evidence note next to each input: analytics for reach, research for impact, and a team estimate for effort. Revisit those notes when new information arrives.
When not to follow the highest score
RICE is not an incident-response policy. A live security problem, a required compliance change, or a dependency can require action even if its ordinary feature score is low.
Intercom explicitly warns against using RICE as a hard-and-fast rule and notes that dependencies and necessary features can change the order. 1
There is another limit: ranking individual items does not automatically produce the best combination under a fixed capacity. You still need to check what fits, what depends on what, and what the team can realistically deliver.
Practise with Sprint Rush
Sprint Rush is the portfolio's time-limited backlog game. Cards show reach, impact, confidence, and effort; the player chooses what to build within a limited capacity.
Use it to practise spotting expensive ideas and preserving capacity. Treat its simplified scoring and short timer as game rules, not a replacement for research, estimation, or a real planning conversation.
After a round, ask yourself: did I choose based on value, or did I simply accept the first reasonable idea?
I would take the highest-scoring idea into a planning conversation with its evidence attached. If the team chooses something else, record the reason. A dependency, a commitment, or a weak assumption is a better explanation than quietly adjusting the numbers until the favourite idea wins.