RICE Scoring: The Complete Prioritization Guide

RICE scoring is a prioritization framework, created at Intercom, that ranks initiatives with one number: RICE Score = (Reach × Impact × Confidence) / Effort. Reach counts people affected per time period, Impact rates the per-person effect, Confidence discounts for uncertainty, and Effort is the person-months required.

Quick Answer: RICE turns "which thing should we build?" into a comparable number. Multiply how many people a bet reaches, how much it moves them, and how confident you are, then divide by the effort it takes. Higher score wins. It exists to make prioritization transparent instead of political.

Why founders need a scoring model instead of gut feel

Founders need a scoring model because gut feel is invisible, unarguable, and biased toward whoever spoke last or loudest. A shared formula makes the reasoning behind a decision explicit, so the team debates the inputs rather than the conclusion.

When you prioritize by instinct, three failure modes show up fast. The freshest idea always feels the most urgent. The person with the most authority quietly wins ties. And low-effort, high-visibility work crowds out the harder bets that would actually move the business.

RICE fixes none of those by magic. What it does is force each competing item onto the same axes, so a "small tweak with a huge audience" and a "deep feature for a niche" become directly comparable. The number is not the decision. It is a structured starting point for the conversation.

That distinction matters most for early founders, where every week spent building the wrong thing is a week of runway gone. If you are still deciding whether an idea is worth building at all, prioritization sits downstream of validation, covered in the complete guide to startup idea validation. RICE assumes you already have a backlog of plausible bets and need to order them.

There is also a second, quieter benefit: RICE creates a written record of why you chose what you chose. Six months later, when a bet did not pay off, you can look back at the exact Reach, Impact, Confidence, and Effort you assigned and learn where your estimates were wrong. Gut decisions leave no such trail — they are unfalsifiable, which means you never improve your judgment. A scored backlog is a feedback loop for your own decision-making.

The framework earns its keep most in a team setting. When two people disagree about what to build next, RICE relocates the argument from personalities to numbers: "I think Reach is closer to 500 than 5,000, and here is the funnel data" is a conversation that resolves, whereas "I just think mine is more important" is not. That is the real product of the framework — not the score, but the structured disagreement it makes possible.

The four RICE factors at a glance

RICE stands for Reach, Impact, Confidence, and Effort — three factors you multiply and one you divide by. Each answers a specific, answerable question about a single initiative, and each uses its own unit so the final score stays meaningful.

The table below summarizes what each factor measures before we go deep on each one.

The takeaway: three of the four factors are estimates you can pressure-test with evidence, and only Impact uses a fixed tier scale. Reach, Confidence, and Effort all scale continuously, which is why the final scores spread out rather than clustering. Get the units right and the formula does the rest.

Reach: how many people or events an initiative affects per period

Reach is the number of people or events an initiative will affect within a defined time period, measured in real units like signups per month or checkouts per quarter. It is the factor most often invented rather than estimated, so anchor it to actual data.

Pick a consistent time window for every item you score — usually a month or a quarter — and never mix windows within one comparison. "Per month" for one feature and "per quarter" for another silently distorts the ranking by a factor of three.

Count only the people the initiative genuinely touches. If a change affects users who reach the checkout page, your Reach is the number who hit that page in your window, not your entire user base. Use analytics, funnel counts, or cohort data wherever you have them; where you do not, that uncertainty belongs in Confidence, not in a padded Reach number.

A common trap is treating Reach as a marketing aspiration ("this could reach millions"). Reach in RICE is a measured or evidenced count over a real period, not a total addressable market. Keeping it grounded is what stops the whole score from inflating.

When you genuinely lack data — a brand-new segment, a feature no analog exists for — estimate Reach from the closest proxy you have. The number of users who currently perform a related action, the size of a waitlist, or the traffic to an adjacent page are all defensible anchors. Write down the proxy you used. The habit of naming your source turns a guess into an assumption you can later test, and it makes your Confidence percentage easier to set honestly.

One more nuance: Reach can count events rather than people when that fits the goal better. If an initiative improves a checkout flow, the meaningful unit might be checkouts per month rather than distinct users, because the same person checking out repeatedly is genuinely reached each time. Choose whichever unit the goal actually cares about, and apply it the same way across every item in the comparison.

Impact: how much an initiative moves each affected person

Impact estimates how much an initiative moves each affected person toward your goal, expressed on Intercom's fixed multiplier scale: 3 for massive, 2 for high, 1 for medium, 0.5 for low, and 0.25 for minimal. It is deliberately coarse because per-person impact is hard to measure precisely.

The tiered scale is a feature, not a limitation. Trying to distinguish an impact of 1.7 from 1.8 is false precision — you rarely know per-person effect that finely. Snapping every estimate to one of five rungs keeps the conversation honest and comparable across your whole backlog.

Impact is per person, not in aggregate. A feature that massively helps a small group and a feature that mildly helps everyone are separated cleanly, because the aggregate scale of the audience already lives in Reach. Double-counting audience size in both Reach and Impact is one of the most common ways teams accidentally inflate a score.

Tie your tier to the goal you actually care about — activation, retention, revenue, whichever metric this quarter hinges on. "Massive" should mean massive for that objective, not massive in some vague general sense.

A useful discipline when you are torn between two tiers is to describe, in one sentence, what the affected person would actually do differently. If you can articulate a concrete behavior change — "they finish setup instead of abandoning it" — you probably have a high or massive impact. If the best you can manage is "it would be a bit nicer," you are looking at low or minimal. The sentence test keeps Impact anchored to observable behavior rather than to how excited you feel about the idea.

Because the scale tops out at 3, Impact can never rescue a low-reach bet on its own. That ceiling is intentional. It prevents a single passionately-championed feature from dominating the ranking purely on claimed impact, and it forces genuinely important work to justify itself on reach and effort as well.

Confidence: how sure you are about your estimates

Confidence is a percentage that discounts your score for uncertainty in the Reach and Impact estimates, using 100% for high confidence, 80% for medium, and 50% for low. It is the framework's built-in humility mechanism — the check on wishful thinking.

Anchor each level to the strength of your evidence. Use 100% when you have hard data or a completed test. Use 80% when you have some evidence but real gaps. Use 50% when the numbers are largely educated guesses. If you find yourself wanting a number below 50%, that is a signal the bet is not yet ready to prioritize — it needs a validation experiment first.

Confidence is where RICE connects directly to evidence quality. A bet backed by customer interviews and usage data earns a higher multiplier than one backed by a hunch, and that gap flows straight into the ranking. The mechanics of turning evidence into a defensible Confidence percentage are worked through in how to calculate a RICE score step by step.

Resist the pull toward 100% on your favorite idea. The whole point of Confidence is to tax optimism. A 50% multiplier literally halves a score, which is exactly what a shaky-but-exciting idea deserves relative to a well-evidenced one.

Effort: the total person-months an initiative requires

Effort is the total amount of team time an initiative requires, estimated in person-months across every role involved — design, engineering, product, QA, and anyone else. It is the only factor in the denominator, so it directly penalizes expensive bets.

Count all roles, not just engineering. A feature needing two weeks of design, one month of engineering, and a week of QA is roughly 1.5 person-months, not one. Undercounting Effort quietly promotes work that is heavier than it looks.

Keep the estimate coarse. Half-month and whole-month granularity is enough; RICE is a ranking tool, not a project plan, and false precision on Effort adds no ranking accuracy. If an item is genuinely tiny, a minimum of 0.5 person-months keeps it from dividing your score into an unrealistically large number.

Because Effort divides, it is where the framework rewards leverage. Two bets with identical Reach, Impact, and Confidence but different Effort will rank strictly by which is cheaper to ship — the essence of "biggest bang for the buck."

Effort is also the factor most vulnerable to optimism bias in the opposite direction from Confidence. Teams routinely underestimate how long work takes, and a lowballed Effort inflates the score of the very bets most likely to blow their timeline. If your team has a track record of estimates running long, it is reasonable to pad Effort deliberately, or to score with a rougher upper-bound estimate rather than a best case. The goal is a ranking that survives contact with reality.

Where two initiatives share components — a shared data model, a reusable UI element — be careful not to double-count the shared Effort in both. Score the marginal person-months each item adds on top of what you would build anyway. Otherwise you penalize the second item for work the first one already paid for, and the ranking drifts.

How to compute and rank RICE scores

Compute a RICE score by multiplying Reach, Impact, and Confidence, then dividing by Effort: (Reach × Impact × Confidence) / Effort. Do this for every backlog item, sort descending, and the highest scores are your prioritized shortlist.

Work through the factors in order for each item. First estimate Reach as a count per your chosen period. Second, pick an Impact tier. Third, set Confidence as a decimal (80% becomes 0.8). Fourth, estimate Effort in person-months. Then run the arithmetic.

Here is a small hypothetical, illustrative example — these numbers are invented purely to demonstrate the formula, not real market data — comparing two imagined features scored over one month:

For the first row: (2,000 × 2 × 0.8) / 2 = 1,600. For the second: (300 × 1 × 1.0) / 1 = 300. The takeaway is that streamlined onboarding ranks far higher despite lower Confidence, because its reach and per-person impact dwarf the export feature — exactly the kind of counterintuitive result gut feel would miss.

The absolute number is meaningless on its own; a RICE score only means something relative to your other scores. A 1,600 is not "good" in any universal sense — it is simply higher than 300. Use the ranking, not the raw value.

Score items in a batch, not one at a time. Consistency across the set is what makes the ranking trustworthy, and you calibrate your tiers far better when you assign Impact to ten items in one sitting than when you score them piecemeal over weeks. Sit down with the whole backlog, run every item through the same four questions in the same order, and only then sort. Scoring in isolation invites drift, where an "8-out-of-10 impact" assigned in January quietly means something different from one assigned in March.

Recalculate periodically. Reach shifts as your user base grows, Confidence rises as experiments resolve, and Effort estimates sharpen as you learn. A RICE backlog is a living document, not a one-time exercise. Many teams re-score at the start of each planning cycle, treating the previous scores as a baseline to revise rather than gospel to defend.

Common RICE scoring mistakes founders make

The most common RICE mistake is treating the output number as an objective verdict rather than a structured input to a human decision. The score organizes the debate; it does not end it. Below are the errors that most often distort the ranking itself.

• Padding Reach with total addressable market. Reach is a measured count over a real period, not an aspirational ceiling. Inflating it here inflates the entire score.
• Double-counting audience size in Impact. Impact is strictly per person. The size of the audience already lives in Reach, so folding it into Impact multiplies the same factor twice.
• Defaulting Confidence to 100%. Optimism on a favorite idea defeats the purpose of the factor. If you have no evidence, 50% or a validation experiment is the honest move.
• Counting only engineering in Effort. Design, QA, and product time are real person-months. Omitting them promotes work that is secretly expensive.
• Mixing time windows. Scoring one item per month and another per quarter distorts the comparison invisibly. Pick one window and hold it constant.
• Never re-scoring. Estimates decay as you learn. A backlog scored once and never revisited slowly drifts out of touch with reality.

RICE is a lightweight tool, and its simplicity is deliberate — but simplicity has trade-offs. If you want a variant without the Effort denominator or the full four-factor structure, the differences are broken down in RICE vs ICE scoring compared. Choosing the right variant matters less than applying whichever one you pick consistently.

If you are using RICE inside a validation-first workflow, Edmired can hold your Reach, Impact, Confidence, and Effort estimates against the evidence behind each bet, so your Confidence percentages reflect real signal instead of enthusiasm. The framework is only as good as the honesty of its inputs.

Key Takeaways

• RICE Score = (Reach × Impact × Confidence) / Effort, a framework created at Intercom to rank initiatives by expected value per unit of work.
• Reach is a measured count over a fixed time period, never a total addressable market — grounding it is what keeps the whole score realistic.
• Impact uses a fixed per-person tier scale (3 massive, 2 high, 1 medium, 0.5 low, 0.25 minimal), deliberately coarse to avoid false precision.
• Confidence discounts for uncertainty at 100%, 80%, or 50%, tying each score directly to the strength of the evidence behind it.
• Effort is total person-months across all roles, and as the sole denominator it rewards cheaper, higher-leverage bets.
• The score is a ranking input, not a verdict — it only has meaning relative to your other scores, and it organizes debate rather than ending it.
• Re-score regularly, because Reach, Confidence, and Effort all shift as your product and evidence evolve.

Frequently Asked Questions

What does RICE stand for in prioritization?

RICE stands for Reach, Impact, Confidence, and Effort. You multiply Reach, Impact, and Confidence, then divide by Effort to get a single comparable score. It was created at Intercom to prioritize product initiatives by expected value relative to the work each requires.

How is a RICE score calculated?

Calculate a RICE score with the formula (Reach × Impact × Confidence) / Effort. Estimate Reach as people affected per time period, pick an Impact tier, set Confidence as a percentage, and estimate Effort in person-months. Higher scores rank first. The value only matters relative to your other items.

What is a good RICE score?

There is no universally "good" RICE score — the number is only meaningful relative to the other scores in your backlog. A score of 1,000 means nothing alone; it simply ranks above a 400 and below a 2,000. Use RICE to order competing bets, not to judge any single one in isolation.

What is the difference between RICE and ICE scoring?

ICE scoring drops Reach and Effort, scoring only Impact, Confidence, and Ease. RICE adds Reach to capture audience size and replaces Ease with an Effort denominator, making it more rigorous but slower to estimate. ICE is faster for quick gut-check ranking; RICE is better when estimates need to be defensible.

Can RICE scoring be used outside of product features?

Yes. Although RICE was designed for product feature prioritization, the same four factors apply to marketing campaigns, validation experiments, growth initiatives, or any backlog of competing bets. As long as you can estimate reach, per-unit impact, confidence, and effort consistently, the framework ranks them.