ICE Scoring
ICE scoring is a lightweight prioritization method that rates each idea on three factors, Impact, Confidence and Ease, usually on a 1 to 10 scale, and combines the ratings into one score. Popularized by growth marketer Sean Ellis to rank growth experiments quickly, it trades precision for speed and suits teams that need a fast, consistent first pass over many ideas.
How ICE Scoring Works
Each idea gets three ratings:
- Impact: how much the idea is expected to move the metric the team is trying to improve, such as activation or trial-to-paid conversion.
- Confidence: how sure the team is that the impact (and ease) estimates are right. Strong evidence, such as results from a previous test, justifies a high rating. A hunch justifies a low one.
- Ease: how simple the idea is to implement. It is the inverse of effort, so cheap ideas score high.
The ratings are then combined. Most teams today multiply them (Impact × Confidence × Ease), which spreads the scores out and punishes ideas that are weak on any single factor. Some teams average or add the three instead, and early descriptions of ICE used addition. The method matters less than applying one method consistently to every idea in the list.
Why ICE Scoring Matters
ICE was designed for experiment backlogs, where a growth team may have dozens of small ideas and limited data. Its value is speed: a team can score a long list in one session and move on to testing, instead of debating each idea.
The Confidence factor is what keeps it honest. A bold idea with no supporting evidence cannot dominate the list on impact alone. Product coach Itamar Gilad pushes this further by tying confidence ratings to the type of evidence available, so opinions and trends earn very little while test results earn much more.
The weakness is subjectivity. All three inputs are judgment calls on a relative scale, so the same person may rate the same idea differently a month later, and two teams' scores cannot be compared. ICE works best as a triage tool within one team, followed by an actual product experiment to replace estimates with data.
ICE Scoring Example
A growth team wants to raise trial-to-paid conversion and scores three ideas, multiplying the ratings:
- Upgrade prompt when a trial user hits a usage limit: Impact 7, Confidence 6, Ease 8. Score 336.
- Annual-plan discount email on day 10 of the trial: Impact 5, Confidence 7, Ease 9. Score 315.
- Full redesign of the pricing page: Impact 8, Confidence 4, Ease 3. Score 96.
The redesign may be the biggest idea, but low confidence and high effort push it down. The team tests the two cheap ideas first and revisits the redesign once it has more evidence.
ICE Scoring vs. RICE Scoring
| ICE | RICE | |
|---|---|---|
| Factors | Impact, Confidence, Ease | Reach, Impact, Confidence, Effort |
| Scale | Relative, usually 1 to 10 | Reach in real numbers, effort in person-months |
| Best for | Fast triage of experiments | Comparing larger product initiatives |
RICE scoring separates how many people are affected (Reach) from the effect on each person, and measures effort in real time units. In ICE, reach is folded into the Impact rating. Both are members of the wider family of prioritization frameworks.