RICE scoring explained, with a worked example

RICE is the most widely used product prioritization formula because it does one thing well: it lets you compare features that have nothing in common — a bug fix, a new integration, an onboarding redesign — on a single scale. The formula is (Reach × Impact × Confidence) ÷ Effort. The discipline is in how you set each factor honestly.

The four factors

Reach — how many people or events this affects in a set time period. A real number, not a rating: "users per quarter," "checkouts per month." If a feature touches 2,000 users a quarter, Reach is 2,000.

Impact — how much it moves the needle for each person it reaches, on a fixed scale: 3 = massive, 2 = high, 1 = medium, 0.5 = low, 0.25 = minimal. The fixed scale is what keeps "Impact" from becoming wishful thinking.

Confidence — how sure you are of your reach and impact estimates: 100% (high, you have data), 80% (medium), 50% (low, it's a hunch). This is the honesty valve. When you're guessing, lower it.

Effort — total work in person-months, across design, engineering, and QA. It's the denominator, so high-effort items get pushed down unless their numerator is large.

A worked example

Say you're a PM for a mobile app weighing three items:

Item	Reach	Impact	Confidence	Effort	RICE
Faster onboarding	5,000	2	0.8	2	4,000
Add Apple Pay	3,000	1	1.0	1.5	2,000
Dark mode	8,000	0.5	1.0	3	1,333

Onboarding wins despite a smaller reach than dark mode, because its impact is higher and its effort is lower. Dark mode reaches everyone but barely moves satisfaction per person and costs a lot — exactly the kind of popular-but-low-leverage work RICE is designed to catch.

Where RICE misleads

RICE can be precisely wrong. The risks: inventing reach numbers and leaving Confidence at 100%; rating everything Impact 2 or 3 because you like it; and forgetting that the score is a starting point for a conversation, not a verdict. A 4,000 and a 3,800 are a tie — don't let two decimal places of false precision override judgment. Use the score to surface the obvious wins and obvious skips, and reason about the middle.