ICE Scoring · The Three-Factor Growth Prioritization Framework
Impact, Confidence, Ease — the lightweight scoring framework Sean Ellis built to prioritize a high-volume growth experiment backlog. The math, the use cases, and how it compares to RICE.
Attribution. ICE scoring was developed by Sean Ellis (the marketer who coined the term "growth hacker" and led growth at Dropbox, LogMeIn, and Eventbrite). Ellis published the framework via his GrowthHackers community in the 2010s. It is closely related to but distinct from Intercom's RICE.
The formula
ICE Score = Impact × Confidence × Ease
Each factor is scored 1–10 (or 1–5 in some variants), giving each item a single comparable number.
Impact. How much will this move the metric we care about, if it works?
Confidence. How confident are we that it will work?
Ease. How easy is this to build and run?
Higher scores get prioritized.
Why it works for growth experimentation
ICE is designed for high-volume A/B test backlogs where you might be evaluating 50–100 candidate experiments and need to decide which 5 to run this sprint. The simplicity is the feature — it takes 30 seconds per item, not 30 minutes.
The 1–10 scales are intentionally informal. The point is relative comparison, not precise estimation. "Is this more impactful than that?" is easier to answer than "how many MRR dollars will this produce?"
Use ICE for sprint-level experiment selection. Use RICE for quarter-level product roadmap decisions. They're complementary, not competing.
The biggest risk with ICE is confidence inflation. When everything scores 8/8/8, you have a list, not priorities. Force the team to spread scores — what's the 2/10 Impact thing on the list? What's a 9/10 Ease thing that someone could ship in a day?