RICE Calculator: Score Features With Real Reach, Not Guesses

RICE ranks a list of things you could build. Four inputs, one number, and the arithmetic is not the hard part:

RICE = (Reach × Impact × Confidence) ÷ Effort

• Reach — how many people this affects in a fixed window. Pick the window and keep it the same for every row: users per quarter, accounts per month. A number, not an adjective.
• Impact — how much it moves the thing you care about for one person who hits it. The common scale is 3 = massive, 2 = high, 1 = medium, 0.5 = low, 0.25 = minimal.
• Confidence — how much you trust the two numbers above, as a percentage. 100% means you have data. 80% means you have some. 50% means you are guessing and should say so.
• Effort — total person-months across everyone who touches it: design, engineering, QA, docs. Round up. Effort is the only input the team already knows.

This page is a methodology guide for scoring with RICE — not a live interactive calculator. Use the formula above in a spreadsheet or planning doc, and treat Reach as a count you can defend.

The scoring is easy. Reach is where it breaks.

Every team that abandons RICE abandons it for the same reason: Reach turns into a number someone made up in a planning meeting. Impact has a scale. Effort has a sprint history. Reach has whoever spoke loudest.

Two rows with invented Reach do not rank against each other — they rank the confidence of the two people who invented them. That is not prioritization, that is a debate with decimals.

Where an honest Reach number comes from

You need a count of people who actually asked, in a window you can defend:

1. Count requests, not opinions. One person asking three times is one. Duplicates merged into one request keep their combined votes, so the count survives your inbox.
2. Keep the window fixed. Votes in the last 90 days for every row, or the last 30 for every row. Never mix.
3. Say which customers. "40 votes" and "40 votes, 9 of them paying" are different rows. If you weight by segment or ARR, write the weighting down next to the score.
4. Write the date on the score. A RICE table with no date is a fossil in six weeks.

If you cannot produce that count for a row, the honest move is to lower Confidence, not to raise Reach.

Try Feedjolt

Feedjolt is a public feedback board: your users post requests, other users vote, and duplicates get merged so one request carries the whole count. That vote count is a Reach number you can point at in a planning meeting.

The embeddable widget is on the free plan — one script tag, so people can file from inside your product instead of hunting for a board URL.

Get Reach from real votes — start a Feedjolt workspace and collect the vote counts your RICE table needs.

Also on this site: ICE calculator for a faster pass, and bug or feature request? for sorting the list before you score it.