Working · 2026-11PrimaryGrowth opsProcess

ICE scoring for test ideas

Rank experiment ideas by scoring each 1 to 10 on impact, confidence and ease, then work from the top.

Also called: Impact Confidence Ease

Claim and evidence

The claimWe use something that we call an ICE Score.

GrowthHackers · saastr.com · undated

What the numbers show

No published number for this tactic has passed our word-for-word check yet, so there is no chart.

Why it works

Diagram, not data
ImpactConfidence (from data)EaseICE score per ideaWork from the topof the backlog
ImpactConfidence(fromdata)EaseICE score per ideaWork from the top of the backlog

A quick shared score forces the team to compare ideas on the same axes and keeps the backlog moving. Confidence is meant to reflect how much data supports the idea, so evidence-backed ideas rise.

How to run it

  1. List every idea in one backlog
  2. Score impact, confidence and ease from 1 to 10
  3. Define confidence as supporting data, not gut feel
  4. Average or sum the scores and sort
  5. Re-score after each result

Ways to run it

Design options to test, not promised results.

  1. Option AClassic ICE

    Score each idea on impact, confidence and ease and sort the backlog, as GrowthHackers' team described.

  2. Option BEvidence-based confidence

    Allow a high confidence score only when data or research supports the idea.

  3. Option CRe-score after results

    Update related ideas' scores after every test so new evidence reshapes the queue.

Timeline

  1. 12016 first documented: Sean Ellis (GrowthHackers), SaaStr Annual talk transcript
  2. 22016 GrowthHackers — Sean Ellis describes using an ICE score on the team's growth backlog
  3. 32026-11 status checked: Working

Dated examples

  1. 2016

    GrowthHackers

    Sean Ellis describes using an ICE score on the team's growth backlog

    “We use something that we call an ICE Score.”
    ✓ quote checked word-for-word on the page

    saastr.com · undated

Does it still work?

Workingas of 2026-11

Widely used for fast triage; critics say the inputs are subjective (see PXL). source

Risks

  • Scores are subjective and gameable
  • Confidence inflates for the ideas people already like

Sources

  1. saastr.com · SaaStr · Gretchen DeKnikker · undated
    “Scoring on a scale of 1 to 10, 10 being something that's really good.”
    ✓ quote checked word-for-word on the page
  2. cxl.com · CXL · Peep Laja · 2016-09-27
    “It’s that broad criteria such as potential, impact, and confidence can leave too much room for subjective scoring.”
    ✓ quote checked word-for-word on the page

Change log

  • 2026-11-20 — harvest v0.1 (ops-ai)