Impact × Confidence × Ease, scored 1–10 and averaged. A faster, lighter call than RICE — built for growth experiments and quick backlog triage.
▸ Try the interactive toolICE is a simplified prioritisation model — faster and lighter than RICE — designed for growth experiments, quick backlog decisions, and early-stage teams where speed of iteration matters more than precision. Each item is scored on three factors on a 1–10 scale, then averaged. Highest average wins.
ICE trades RICE's rigour for speed. There's no separate Reach number and no person-month estimate — just three quick gut-calibrated scores. That makes it ideal for ranking a long list of small experiments fast, and poor for high-stakes decisions where the imprecision matters.
(Impact + Confidence + Ease) ÷ 3 = ICE score (each scored 1–10)
Impact: how much it could move the metric · Confidence: how sure you are · Ease: how cheap and fast to try
When you have thirty experiment ideas and an afternoon, RICE is too heavy. ICE gets you a defensible ranking fast — as long as you respect its limits.
| The shortcut | What it costs | What it gives you instead |
|---|---|---|
| Analysis paralysis | Over-scoring small experiments wastes the time the experiment would have taken. | Three quick scores rank a long list in minutes. |
| Pet experiments | Without a score, the team runs whoever's idea is loudest. | Even a rough average exposes low-ease, low-confidence darlings. |
| Ignoring ease | High-impact ideas chosen regardless of how hard they are to try. | Ease rewards fast, cheap tests — the lifeblood of iteration. |
| False precision elsewhere | Heavy frameworks imply accuracy the inputs don't have. | ICE is honest about being a quick gut-calibrated triage, not a measurement. |
Get all the ideas down first — don't pre-judge. ICE's speed lets you score a long list, so capture everything.
If this worked, how much would it move the metric you care about? 10 = transformational, 1 = negligible.
How sure are you it'll work, given what you know? 10 = strong evidence, 1 = pure hunch. This taxes wishful thinking.
How fast and cheap is it to try? 10 = a day, 1 = a quarter. Ease is what keeps the experiment loop fast.
Average the three, sort descending, run from the top. Re-score freely as you learn — ICE is meant to be cheap to redo.
A growth team had two dozen test ideas and kept arguing about which to run. Scoring each on ICE in a single session, a “rewrite the signup headline” idea — modest impact, high confidence, trivial ease — floated near the top, while a “rebuild the referral system” idea sank on ease.
They'd been about to start the referral rebuild because it sounded ambitious. ICE showed that several near-free experiments could run — and teach them something — in the time the rebuild's planning alone would take.
A simple three-column score per idea, averaged and sorted — deliberately lightweight.
| Factor | Asks | 1–10 means |
|---|---|---|
| Impact | If it works, how big? | 10 = transformational, 1 = trivial |
| Confidence | How sure are you? | 10 = strong evidence, 1 = hunch |
| Ease | How fast / cheap to try? | 10 = a day, 1 = a quarter |
ICE's honesty is its strength: it doesn't pretend to be precise. It's a fast, shared gut-check that ranks a long list well enough to start — and re-rank as you learn.
The danger is using ICE where RICE belongs. Because the scores are quick and subjective, ICE is excellent for low-stakes, high-volume decisions (experiments) and dangerous for high-stakes ones (a quarter-defining bet). Match the tool to the stakes: ICE for triage, RICE when the cost of being wrong is high.
Its subjectivity is fine for experiments, risky for major bets. Escalate to RICE when the cost of error is high.
If everything is an 8, the ranking is noise. Score comparatively.
Skipping the confidence honesty makes ICE a hype meter. Tax your guesses.
ICE is cheap by design — update scores as experiments teach you. A stale ranking wastes its main advantage.
RICE adds Reach and Effort for precision; ICE drops them for speed. Same family, different stakes.
ICE ranks which experiments to run; the lean-experiment and A/B-testing tools (Module 3 & 5) run them.
The Confidence score is only as good as the discovery behind it — interviews and data raise it honestly.
ICE is the default triage tool for AARRR-stage growth experiments.
Take five things you could try this month. Score each on Impact, Confidence, Ease (1–10), average, and rank — fast, no overthinking.
Notice which ideas rose on Ease alone. Those are your cheapest learning opportunities.
If scoring five ideas took longer than five minutes, you're treating ICE like RICE — the whole point is that it's quick enough to redo weekly.