Sample frameDay 255·Product · Prioritisation

Nobody in the room is short of opinions about what to build next. What's scarce is a way to disagree in public.

Product.

RICE

5 min read·Apply by lunch

The question

When your team last settled what to build next, could anyone in that room have said out loud how sure they were — not whether the idea was good, but how much of the case rested on a figure somebody had actually gone and looked up? I'll give you the answer: nobody was asked, because nothing on the list carried a number anyone could argue with.

The idea

Sean McBride was a product manager at Intercom when the company's backlog outgrew the arguments it could hold. In 2016 he set out on the Inside Intercom blog the scoring model his team had built from first principles, and RICE has been in general use ever since. It rests on four numbers, written where anyone can see them. Reach is how many people or events the work touches in a stated period, taken from analytics rather than from the room. Impact is how much it moves things for each of them, on a fixed menu running from 3 for massive down to 0.25 for minimal. Confidence is the discount you apply for not knowing — 1.0 when data sits behind both of the first two figures, 0.8 when it sits behind one, 0.5 when both are informed guesses. Effort is total person-months across every function the work touches. Multiply the first three, divide by the fourth. The dividing is the least interesting part of it, and Day 90's cost of delay already sequences a queue the same way; reach for that one when the waiting itself has a price, and for RICE when the population is countable but your belief about it isn't. Where you point it matters as much as how you run it: the opportunity solution tree was right that a mixed feature list can't be ranked, so this is scoring inside one branch — several solutions to the opportunity you have already chosen, ordered once the choosing is done. What RICE adds is the confidence multiplier, and four written assumptions that the most junior person in the room can dispute by name.

The framework

01
Confidence is the number the model actually adds.
It's the multiplier most teams quietly drop, and it's the one that earns RICE its keep. Treat it as a discount for not knowing: 1.0 when you've got data behind both the reach and the impact, 0.8 when you've got data for one and a strong argument for the other, 0.5 when both are informed guesses. Applied honestly it does something no ranking of gut feel can — it drops a large, plausible, entirely unevidenced idea below a small, well-understood one, and it turns the remedy into research rather than advocacy. Whoever wants their idea ranked higher now has somewhere to go and something to bring back.
02
Where would the reach number actually come from?
Reach is a count of people or events over a stated period, and it should come out of your analytics rather than out of the room: accounts that hit this screen per quarter, trials that stall at this step per month, tickets on this theme per week. Naming the source is half the value, because two people who disagree about reach are usually picturing two different populations. Write the period down as well — something that touches everyone once a year and something that touches a tenth of them weekly score very differently, and only the period makes that visible.
03
Person-months of labour — everyone's, not just engineering's.
Count every function the work touches rather than the one that writes the code: design, research, data, support enablement, the migration nobody costed. Then keep the units coarse, because half a month is the finest you should ever go, and false precision in the denominator produces the most confident wrong answers the model can generate. If an item won't survive that treatment — a six-month build whose reach and impact are guesses dressed as inputs — it doesn't belong on the list at all. Split it until it can be scored honestly, or take it out and argue it as a strategic bet.
04
The score is a ranking input, never a verdict.
It doesn't know that a low-scoring item unblocks three high-scoring ones, that a regulator has set a date, or that this is the year you decided to move upmarket. Use it to order the middle of the list, where the genuine uncertainty lives, and keep an explicit override lane for strategic and dependency-driven work — each override written down as an override, and signed. What the score replaces isn't judgement. It's the version of that meeting where the reasoning stayed inside people's heads, so there was nothing precise enough to object to and the ranking settled itself.

Members only · 4 principles + template + AI mentor

364 more frameworks are waiting.

RICE is Day 1 of 365. One framework every morning for a year — across strategy, sales, negotiation, leadership, and more.

$1 /day

Billed $365/yr · cancel any time

365 frameworks

12 topics · templates · AI mentor

Reserve your place — $365/yr →

One payment a year · the whole library · no auto-renewal.

More Product frameworks

All 22 product frameworks →