RICE Scoring Model Prioritization: A Step-by-Step Guide

The RICE scoring model ranks ideas by multiplying Reach, Impact and Confidence, then dividing by Effort. The result is one number per idea, which lets you compare a pricing page redesign against a new integration without arguing from opinion. Use it when you have more good ideas than capacity and need a defensible order.

What RICE Is and Where It Came From

RICE was popularized by the product team at Intercom as a way to stop prioritization debates from being won by the loudest person in the room. The formula is simple:

RICE score = (Reach × Impact × Confidence) ÷ Effort

Each letter forces a question that teams usually skip:

• Reach: How many people will this affect in a set time period?
• Impact: How much will it move the goal for each person reached?
• Confidence: How sure are you about the Reach and Impact numbers?
• Effort: How much total work will it take from the whole team?

The division by effort is what makes the model useful. A modest idea that takes one week often beats a large idea that takes three months, and RICE makes that visible.

RICE is not a forecast. It is a structured way to compare ideas using the same yardstick. The value comes from the conversation about each input, not from the decimal places in the output.

When RICE Works and When It Doesn't

RICE fits best when you have 5 to 30 candidate ideas, a shared goal, and at least rough data on how many people each idea touches. Product roadmaps, marketing experiments, internal process fixes and website changes all qualify.

It fits poorly in three situations:

• Compliance or legal work. If something must happen, scoring it is theater. Put it in the plan first and score everything else around it.
• Tiny lists. With three options, just talk it through.
• Pure exploration. If you have no idea who an initiative affects, your Confidence will be so low that the score says nothing.

If you only need to sort work into must-have and nice-to-have, the MoSCoW method is faster. If your question is which features customers will actually value, the Kano model answers that better. RICE is the tool for ranking when capacity is the constraint.

Step-by-Step: How to Score Ideas With RICE

Step 1: Fix the goal and the time window

Before you score anything, write down the one goal this planning cycle serves and the period you are scoring for. "Increase trial-to-paid conversion this quarter" works. "Grow the business" does not.

The time window matters because Reach must be counted over the same period for every idea. Pick a quarter and stick to it. If you are unsure which goal to anchor on, a North Star metric gives you a single number to measure Impact against.

Step 2: List the ideas in one sentence each

Write each candidate as a single sentence a new hire could understand. Vague items like "improve onboarding" produce vague scores. "Add a five-step setup checklist inside the app" can be scored.

Step 3: Score Reach

Reach is a count of people or events within your time window. Use real numbers whenever you can:

• Monthly signups × 3 for a quarter
• Customers on a particular plan or using a particular tool
• Visitors to a specific page, from your analytics
• Support tickets of a certain type

Do not use percentages or vague labels. "Most customers" is not a Reach. "1,800 new trials this quarter" is.

Step 4: Score Impact on a fixed scale

Impact is the hardest input to measure, so use a fixed scale and do not invent your own mid-meeting:

Tie each level to your goal. If the goal is trial-to-paid conversion, a 2 might mean you expect roughly a 10% relative lift for the people reached, a 1 about 5%, and a 0.5 about 2 to 3%. Write those anchors down once so everyone uses them the same way.

Step 5: Score Confidence honestly

Confidence is a percentage that discounts your Reach and Impact estimates:

• 100%: You have hard data, such as a prior test or clear usage numbers.
• 80%: You have solid supporting evidence, such as multiple customer requests plus a comparable result elsewhere.
• 50%: You have an informed guess and some anecdotes.
• Below 50%: It is a hunch. Treat the score as a prompt to run a cheap test first.

Confidence is where bias creeps in. The cognitive biases that hurt business decisions show up here constantly, especially optimism and the tendency to defend your own idea. Have the person who proposed the idea state the evidence, then let someone else set the percentage.

Step 6: Score Effort in person-months

Effort is the total work from everyone involved: design, engineering, marketing, QA, support training. Count it in person-months (or person-weeks if your ideas are small). Round up. A task that "takes two weeks of dev" usually also takes a week of design review and a week of rollout work.

Get the estimate from the people who will do the work, not from the person who proposed the idea.

Step 7: Calculate, sort, and sanity-check

Multiply Reach, Impact and Confidence, then divide by Effort. Sort from highest to lowest. Then read the top five and bottom five out loud. If something feels wrong, do not override the score. Find the input that is off and fix it, or admit that a gut feeling is evidence you left out of the model.

A Worked Example

Say you run a 12-person software company selling scheduling software to fitness studios. You have 3,000 paying studios and about 1,800 new trial signups per quarter. The goal this quarter is to increase trial-to-paid conversion and reduce churn. Your delivery capacity is two engineers and one designer, which you estimate at 9 person-months for the quarter after meetings and support.

Here are six competing ideas and their scores:

Here is how the top score comes out: 1,800 × 2 × 0.8 = 2,880, divided by 3 person-months, gives 960.

What the numbers say

The onboarding checklist wins because it reaches every new trial and has support behind it: your support team already sees the same setup questions weekly, and a competitor's trial data pointed the same way. That justifies 80% Confidence.

The mobile app is the interesting case. It has the same Impact as the checklist and a bigger Reach, but it eats 8 of your 9 person-months and rests on a 50% Confidence guess. It drops to third. That is the model doing its job: the idea may be good, but it is too expensive for the evidence behind it.

The custom reporting dashboard scores last even though it has the highest Impact. The three largest customers asked for it loudly, but only 150 studios would use it. This is the classic case where a big account's request feels urgent and the math says otherwise.

The decision

With 9 person-months of capacity, you commit to:

• Onboarding checklist (3)
• Pricing page redesign (1.5)
• QuickBooks integration (2)
• Failed-payment reminders (1)

That totals 7.5 person-months and leaves 1.5 as a buffer for bugs and surprises. The mobile app and dashboard wait. For the mobile app, you plan a two-week customer interview round to raise or lower Confidence before next quarter, which costs a fraction of the build.

Notice one more thing. The failed-payment reminders score only 240, but they cost one person-month and directly protect existing revenue. If a studio pays $120 a month and 5% of 3,000 studios hit a failed payment each quarter, recovering even a third of those is real money. When an item is cheap and the logic is clear, you can include it even if it is not near the top. RICE ranks. You still decide.

Common Mistakes (and How to Avoid Them)

Mistake 1: Inflating Confidence to rescue a favorite idea

This is the most common failure. Someone wants their idea to win, so every input drifts upward: Reach gets a little bigger, Impact becomes a 3, and Confidence lands at 100% with no data behind it.

Fix: Require one sentence of evidence next to every Confidence score above 50%. If the evidence is "I just know," the score is 50% or lower. Also have the proposer present the idea while a neutral person assigns Confidence.

Mistake 2: Using different units or time windows

If one idea counts Reach per month, another per year, and a third as a share of customers, the scores are meaningless. Effort has the same problem when one estimate is in dev days and another in team weeks.

Fix: Fix the time window and units in Step 1 and put them at the top of your scoring sheet.

Mistake 3: Treating the score as the decision

A RICE score of 960 versus 900 is not a meaningful gap. The inputs are estimates, and small differences are noise.

Fix: Group ideas into tiers instead of fighting over rank. Items within roughly 20% of each other are a tie, and you can break the tie with strategy, risk or dependencies.

Mistake 4: Scoring too many things

Scoring 80 ideas burns a day and produces fatigue-driven numbers.

Fix: Pre-filter. Cut anything that doesn't serve this quarter's goal, then score the 10 to 20 that remain. If you need a faster first pass, use the Eisenhower Matrix to clear out obvious noise.

Mistake 5: Never revisiting scores

Once a score is on a spreadsheet, it tends to stay there. But Reach changes as you grow, and Confidence changes the moment you run a test.

Fix: Re-score at the start of every planning cycle, and any time you learn something material.

Template: A Copyable RICE Sheet

Set up a spreadsheet with these columns:

• Idea (one sentence)
• Goal it serves
• Reach (count per [time window])
• Impact (0.25, 0.5, 1, 2 or 3)
• Confidence (percentage)
• Evidence (one sentence, required above 50%)
• Effort (person-months)
• RICE score = (Reach × Impact × Confidence) ÷ Effort
• Decision (Commit, Test first, Defer, Drop)

Add a header row stating the goal, the time window and the impact anchors. Add a capacity line at the bottom: total person-months available, and a running total of committed effort. The "Test first" column is underrated. It turns low-Confidence, high-potential ideas into small experiments instead of either bets or rejections.

Running the Session With Your Team

Keep the scoring meeting to 60 to 90 minutes. A format that works:

1. The facilitator reads the goal, window and Impact anchors.
2. Reach is filled in beforehand from data, so nobody debates it live.
3. The group scores Impact and Effort together, with the people doing the work giving Effort.
4. A neutral person sets Confidence after hearing the evidence.
5. Sort, review the tiers, and map the top items against capacity.

If your team struggles to agree on inputs, decide in advance how you will break ties. The consensus vs consent approach is useful here: you do not need everyone to love a score, only to confirm they can live with it for this quarter.

Key takeaways

• RICE ranks ideas with one formula: (Reach × Impact × Confidence) ÷ Effort. Use it when capacity, not ideas, is the bottleneck.
• Fix the goal, time window and Impact scale before scoring. Mixed units are the fastest way to produce a meaningless ranking.
• Confidence is where bias hides. Require evidence for any score above 50%, and let someone other than the proposer set it.
• Treat scores within about 20% of each other as ties, and use strategy, dependencies and risk to break them.
• Low-Confidence, high-potential ideas deserve a cheap test, not a build. Use the score to decide what to learn next.
• Re-score every planning cycle. Reach grows, Effort estimates improve, and new evidence changes Confidence.