RICE vs ICE: which prioritization framework to use
RICE adds a Reach term that ICE leaves out. That single difference makes RICE better for teams with quantifiable user data — and ICE better for early-stage decisions where Reach is unknowable.
Both score ideas with multiplication. RICE has 4 inputs, ICE has 3. The missing term — Reach — is the entire difference, and it matters more than it looks.
Need the RICE formula with worked examples? See RICE score calculator: the formula with 3 worked examples.
At a glance
| RICE | ICE | |
|---|---|---|
| Formula | (Reach × Impact × Confidence) / Effort | Impact × Confidence × Ease |
| Inputs | 4 | 3 |
| Best for | Mid-to-late stage products with usage data | Early stage / experiments where Reach is unknowable |
| Time per scoring | 5–10 min per item | 2–5 min per item |
| Origin | Intercom, 2016 | Sean Ellis / growth-hacking community |
The Reach difference
RICE forces you to estimate how many customers each idea affects in a given period. That sounds obvious, but it has two effects:
- It penalizes ideas with narrow audiences. A feature for 50 enterprise customers scores lower than a feature for 5,000 mid-market customers, all else equal. ICE doesn't capture this.
- It requires the team to have the data. If you can't estimate Reach, you can't score RICE. For a 6-month-old startup with 200 users, Reach is approximately "everyone or nobody"; the term degrades to noise.
The Effort vs Ease difference is cosmetic — both are denominators that reward small bets. ICE phrases it positively ("Ease, 1–10") because the growth-hacker culture preferred upbeat framing.
When to use RICE
- You have an established product with real usage data by feature, by user segment, by geography
- The team is scoring a backlog rather than testing hypotheses
- You need a defensible quarterly prioritization for stakeholders
Used at Intercom, Shopify, and most mid-to-late stage SaaS companies for the same reason: the Reach denominator surfaces non-obvious priorities.
When to use ICE
- Early stage — limited usage data, every idea is partly speculative
- Growth experiments — ICE was designed for testing hypotheses, not shipping features
- Decision under high uncertainty — when Reach is unknowable, asking for it produces fake numbers worse than not asking
Sean Ellis's original framing: ICE was specifically built for evaluating growth hacks, where the question is "is this worth trying?" not "which of these should we ship?"
When neither is enough
- For specific high-stakes decisions (which market to enter), the right tool is SWOT, not a scoring framework
- For strategic positioning, scoring frameworks don't help at all — use Five Forces
- For team alignment on a single bet, a premortem reveals more than scoring would
A diagnostic
If you're scoring 20+ items and you can't credibly estimate Reach for most of them, that's a signal you should be running ICE — or, more likely, that your team needs more user research before any scoring will produce reliable rankings.
Run them
Full RICE Academy guide →. For ICE, the catalog entry has the worksheet template.