The formula Intercom, 2016
RICE was developed at Intercom to compare wildly different projects on one scale. Each factor is estimated per time period (usually a quarter):
Reach
How many people this touches per quarter. Use real data: "800 support tickets/quarter mention SSO" beats "lots of people want it."
Unit: people or events / quarterImpact
How much it moves the needle per person reached. A fixed scale keeps estimates honest: 3 = massive, 2 = high, 1 = medium, 0.5 = low, 0.25 = minimal.
Unit: multiplier (0.25–3)Confidence
How sure you are about your Reach and Impact numbers. 100% = backed by data, 80% = strong signal, 50% = educated guess. Below 50%? It's a moonshot — go get data first.
Unit: percentage (0.5–1.0)Effort
Total work across product, design and engineering, in person-months. Effort divides — a cheap win can beat a big bet.
Unit: person-monthsLive prioritization lab
The sliders control the highlighted request, "Slack notifications". Every change recomputes its RICE score, re-sorts the backlog table, and re-orders the 3D bars — tallest (highest score) on the left. Drag the 3D scene to rotate.
| # | Request | R | I | C | E | RICE |
|---|
Worked example: CSV export
Support tags 350 requests/quarter asking for CSV export (Reach = 350). It removes a real workflow blocker but only for reporting tasks — Impact = 1 (medium). The count comes straight from Linear's customer-request label, so Confidence = 100%. One engineer estimates 2 weeks ≈ 0.5 person-months.
Compare that to a mobile app: Reach 1,200, Impact 2, but Confidence 50% and Effort 12 person-months → (1200 × 2 × 0.5) ÷ 12 = 100. The unglamorous CSV export outranks the shiny app 7-to-1 — that's the whole point of dividing by effort.
A triage workflow that works in Linear
Capture everything in one inbox
Pipe requests from Slack, support and sales into a Linear Triage queue. One intake point, zero "I DM'd the PM about it" losses.
Deduplicate and count Reach
Merge duplicates into a canonical issue and attach each customer as a customer request. The attachment count becomes your Reach evidence.
Score weekly, not per-request
Batch-score new issues in a 30-minute weekly triage. Estimating Impact and Confidence side-by-side keeps the scale consistent.
Sort by RICE, cut at capacity
Sort the backlog view by the score field. If the team has 6 person-months this quarter, draw the line where cumulative Effort hits 6.
Re-score when facts change
RICE is a snapshot, not a contract. New data moves Confidence; churned customers move Reach. Revisit each cycle.