Skip to content
OnePageCanvas

How to fill in the RICE Prioritization

Reach × impact × confidence ÷ effort — rank the backlog with arguments, not volume. This guide walks every block in the recommended order — what belongs there, the questions that unlock it, and patterns from real canvases.

RICE scores each candidate idea on four factors — how many people it Reaches, how much Impact it has on each, how Confident you are in those estimates, and how much Effort it costs — then ranks by (reach × impact × confidence) ÷ effort. The output is a backlog ordered by expected value per unit of work.

The score is not the point; the argument is. RICE forces every 'we should obviously build X' claim to name its numbers, and most prioritisation fights end the moment both sides write down reach and effort. Treat the final ranking as a strong default you may consciously override, not as arithmetic truth.

Developed by Sean McBride at Intercom (circa 2016) for scoring product roadmap ideas, and since adopted well beyond product management for any ranked-backlog decision.

1

Reach

How many people does each idea touch, per period?

For each idea, estimate how many users or customers it affects in a defined period — e.g. customers per quarter. Use real data wherever it exists: traffic to that screen, users of that feature, tickets about that problem. One note per idea, in the form 'Idea — number and source'. The shared time period is what makes scores comparable.

Ask yourself

  • How many users hit this per quarter, from analytics rather than instinct?
  • Is this everyone, a segment, or a handful of loud voices?
  • What period are we counting over — and is it the same for every idea?
  • Where did this number come from?

Patterns that work

  • Same period for every idea, or the comparison is meaningless
  • Pull numbers from analytics and tickets, not memory
  • Loud requests often score surprisingly low here — that is the finding
2

Impact

How much difference does it make to each person reached?

Impact on the goal, per person reached, on Intercom's scale: 3 = massive, 2 = high, 1 = medium, 0.5 = low, 0.25 = minimal. Name the goal first (conversion, retention, satisfaction) — impact without a stated goal is vibes. One note per idea: 'Idea — score, because…'.

Ask yourself

  • Impact on which goal, exactly?
  • For one affected user, how big is the change: massive, high, medium, low?
  • Are we scoring the average user reached, or the best case?
  • What evidence suggests this effect size?

Patterns that work

  • Use the fixed scale: 3 / 2 / 1 / 0.5 / 0.25
  • Score the average reached user, not the ideal one
  • State the goal the impact is measured against
3

Confidence

How much of this is data, and how much is hope?

Discount for uncertainty: 100% = solid data on both reach and impact, 80% = data on one and judgement on the other, 50% = educated guess. Below 50% means you are not prioritising, you are gambling — send that idea to research instead of the backlog. Confidence is what stops exciting fictions outranking boring certainties.

Ask yourself

  • Which of our reach and impact numbers rest on real data?
  • What would raise this to 80% — and how cheap is finding out?
  • Are we above 50%? If not, what research replaces the guess?
  • Whose estimate is this, and what is their track record?

Patterns that work

  • Use 100% / 80% / 50% — finer precision is false precision
  • Below 50%: route to research, not the roadmap
  • This factor exists to punish enthusiasm unsupported by evidence
4

Effort

What does each idea cost, in person-months?

Total work across everyone involved — product, design, engineering, QA, launch — in person-months (or person-weeks, applied consistently). Effort is the divisor, so underestimating it inflates a score faster than any other error. Include the second 80%: review, edge cases, rollout, and support.

Ask yourself

  • Person-months across every discipline, not just engineering?
  • What did the last similarly-sized thing actually take?
  • What is the smallest version that keeps most of the impact?
  • Who signed off this estimate?

Patterns that work

  • One unit (person-months) applied to every idea
  • Estimate from comparable past work, not optimism
  • A smaller version with the same impact beats a bigger score
5

Ranked Backlog

What does the arithmetic say — and where do you overrule it?

Compute each idea's score — (reach × impact × confidence) ÷ effort — and list them highest first, one note per idea with the working shown. Then apply judgement: dependencies, strategy, and commitments may reorder items, but every override gets written down with its reason. A silent override is just the loudest voice winning again, with extra steps.

Ask yourself

  • Does the top of the list surprise anyone — and is the surprise a data error or a bias?
  • Which items are we overriding, and what is the stated reason?
  • Do the top three fit in the quarter's capacity?
  • What gets explicitly dropped, and who tells the requester?

Patterns that work

  • Show the working in each note so the ranking can be argued with
  • Overrides are allowed; silent overrides are not
  • Re-score quarterly — reach and confidence move with the data

Ready to fill yours in?

The editor carries this whole guide with it — every block has these prompts and starter notes built in. Free, no signup, autosaves in your browser.

Open the RICE editor