ICE Prioritization
Score every experiment idea on Impact, Confidence and Ease to run a high-velocity testing program fairly
- Difficulty
- Easy
- Time to result
- ~days to results
- Steps
- 3
- Confidence
- 90%
To run a lot of experiments, you need ideas sourced from across the company — and a transparent way to compare them so contributors understand why an idea was or wasn't picked. ICE scores each idea on Impact (best-case upside), Confidence (belief it'll work) and Ease (effort), keeping prioritization deliberately simple.
Origin
Sean Ellis created ICE specifically to sustain a high-velocity testing program: without a systematic comparison, people whose ideas are rejected get upset and waste time. Intercom's RICE adds Reach, which Sean considers already baked into Impact.
Core principles
- 01Sourcing ideas company-wide requires a transparent scoring system or contributors disengage
- 02Impact = best-case scenario of how much this could move the metric
- 03Reach is already a component of impact — extra factors risk over-complicating
- 04Better tests beat more tests, but if forced to optimize one thing, more testing wins
How to run it
- 1
Collect ideas from across the company
Open experiment idea submission beyond the growth team.
- 2
Score each idea I, C, E
Rate Impact (best-case upside), Confidence (likelihood it works), and Ease (low effort = high score).
Pro tip Judge Impact as the best-case scenario — reach is folded in, so you don't need a separate reach term.
- 3
Rank and communicate why
Prioritize by combined score and use the transparency to explain to contributors why their idea did or didn't get chosen.
Pro tip A systematic comparison makes people more likely to accept rejections and to submit better ideas next time.
In the wild
Asked about Intercom's RICE (adding Reach), Sean argues Reach is already part of the Impact term, so the addition is unnecessary — while acknowledging that spending ~30 minutes estimating effort per idea (as in 'detailed RICE') can be worth it to avoid unlikely-to-work tests.
→ Keeps ICE simple; concedes deeper estimation has value in some cases.
Common mistakes
Over-complicating prioritization
Adding factors to feel rigorous rarely improves decisions; Impact/Confidence/Ease captures what matters.
Rejecting ideas without a transparent reason
Contributors disengage and stop submitting good ideas if they can't see why theirs lost.
Is it for you?
Best for
Growth or product teams running frequent experiments who need to triage a steady stream of company-sourced ideas
Not ideal for
Situations where a single high-stakes bet warrants deep estimation rather than a quick relative score
From the transcript
“in order to be able to effectively run a high velocity testing program you need to be able to Source ideas from across the company…”
“if you have a systematic way of being able to compare ideas it's it's more likely that people be able to get it and they'll…”
“reach is a is a super important part of impact and so like I think it's already factored in the eye in ice”
From the episode
The original growth hacker reveals his secrets
Sean Ellis (author of “Hacking Growth”)