The formula
RICE score = (Reach × Impact × Confidence) ÷ Effort
Originated at Intercom as a way to compare product bets on a shared scale instead of by gut feel or whoever argues loudest. Each factor forces a specific, separate judgment: how many people does this actually reach, how much does it move the needle for each of them, how sure are you about both of those numbers, and how much work is it.
What each input means
Reach is a count, not a feeling - people or customers affected per a fixed time period (commonly a quarter). Impact is a categorical multiplier (massive, high, medium, low, minimal) because precise impact estimates this early are usually false confidence. Confidence discounts the score for guesswork - a high-reach, high-impact idea backed by no data should score lower than one backed by real signal. Effort is person-months, and it's a divisor, so underestimating it inflates the score.
What RICE doesn't do
RICE ranks relative priority within one list, on one team's own judgment calls - it doesn't make those judgment calls objective. Two people scoring the same idea will produce different reach and impact numbers, and that's fine; the value is in a consistent framework for the conversation, not a formula that removes the need for one.