ICE Scoring Model Explained: Formula, Examples and a 4-Week Rollout

8 min read · 2026-10-10

The ICE scoring model is a quick way to rank ideas, features or experiments. You score each idea from 1 to 10 on three criteria: Impact, Confidence and Ease. Then you multiply the three numbers, and the idea with the highest score goes first. ICE is often credited to Sean Ellis, the person who coined the term "growth hacking". It is now common in product and marketing work.

ICE helps when your list of ideas is longer than your team's time. It does not give you the "right" answer. It gives you a shared, fast and visible way to compare options, so you can stop debating and start building. Below you will find the formula, a worked example and a 4-week roadmap to bring ICE into your team's planning.

The roadmap at a glance

Goal: Replace gut-feel prioritization with ICE scores and turn the top-ranked ideas into a visible roadmap. Duration: 4 weeks

  1. Prepare the Backlog (Week 1)

    Get every candidate idea into one list, written the same way.

    • Collect ideas from your backlog, support tickets, sales notes and team suggestions.
    • Remove duplicates and merge ideas that solve the same problem.
    • Write each idea as one sentence: what changes and for whom.
    • Pick the one metric the ideas should move this quarter, such as activation or retention.

    Milestone: One shared list of ideas, each tied to the same target metric.

  2. Define Your Scoring Scale (Week 1)

    Agree on what each score means before anyone scores.

    • Write a short definition for Impact scores 1, 5 and 10 against your target metric.
    • Define Confidence levels by evidence type: opinion, user feedback, data, past test result.
    • Define Ease in team effort. For example, 10 means under a day and 1 means more than a quarter.
    • Decide whether you multiply the scores or average them, and stick to it.

    Milestone: A one-page scoring guide that the whole team has read.

  3. Score the Ideas (Week 2)

    Give every idea an honest score you can compare with the others.

    • Have two or three people score each idea on their own first.
    • Compare scores and discuss only the ones with a gap of 3 points or more.
    • Note the evidence behind each Confidence score next to the idea.
    • Calculate the final ICE score for each idea.

    Milestone: Every idea has an Impact, Confidence, Ease and total score, with notes.

  4. Review and Pick (Week 3)

    Turn the ranked list into a short list you can commit to.

    • Sort ideas by ICE score, highest first.
    • Check the top ten for dependencies, legal limits or strategic gaps.
    • Move low-Confidence, high-Impact ideas into a "test first" group.
    • Agree on the ideas that fit your capacity for the next cycle.

    Milestone: A signed-off short list of ideas for the next cycle.

  5. Build the Roadmap and Rescore (Week 4 and ongoing)

    Make the decision visible and keep the scores fresh.

    • Place the chosen ideas on a roadmap by phase, with clear milestones.
    • Share the roadmap and the scoring guide with stakeholders.
    • Set a rescoring date, such as every month or after each release.
    • Update Confidence scores when new data or test results arrive.

    Milestone: A published roadmap and a date in the calendar for the next rescoring.

What Impact, Confidence and Ease Mean

Each letter answers one simple question. Keep these definitions short and shared, because vague definitions are the main reason ICE scores drift.

Confidence is the letter most teams get wrong. It is not "how much we like the idea". It is a score for the quality of your evidence. Product coach Itamar Gilad built a Confidence Meter around this idea: it links Confidence to the type of evidence you have, from your own conviction up to real usage data. An idea backed by a user test should score higher than an idea backed by one opinion in a meeting.

  • Impact: How much will this idea move our target metric if it works? A 10 changes the metric a lot. A 1 barely touches it.
  • Confidence: How sure are we that the Impact score is right? A 10 is backed by test results or strong data. A 1 is a hunch.
  • Ease: How easy is it to build and ship? A 10 takes very little effort. A 1 needs a large team for a long time.

The ICE Formula, With a Worked Example

The standard formula is ICE score = Impact × Confidence × Ease. With scores from 1 to 10, the result ranges from 1 to 1,000. Some teams average the three numbers instead, which gives a result from 1 to 10. Both work. Multiplying spreads the scores further apart, so differences are easier to see.

Here is an example for a small software team whose target metric is activation:

  • Onboarding checklist: Impact 8, Confidence 6, Ease 7. Score 8 × 6 × 7 = 336.
  • Dark mode: Impact 3, Confidence 8, Ease 6. Score 3 × 8 × 6 = 144.
  • New integration with a chat tool: Impact 7, Confidence 4, Ease 3. Score 7 × 4 × 3 = 84.

What the Worked Example Shows

The onboarding checklist wins. Dark mode comes second even with low Impact, because the team is sure about it and it is easy to build. The integration might be big, but the evidence is thin and the work is heavy. A good next step is a cheap test, such as a waitlist or a few user interviews. That raises its Confidence before anyone commits weeks of work.

When to Use ICE, and When Not To

ICE is built for speed. It works best when you have many small or medium ideas and need a ranked list today, not next month. Growth experiments, marketing tests, onboarding tweaks and backlog clean-ups are good fits. It is also useful for solo founders and small teams, because one person can score twenty ideas in an hour.

It is a weaker fit for large, long bets. If one idea is a six-month platform rebuild and another is a button change, the same 1 to 10 scale can hide big differences. ICE also ignores how many users an idea reaches. If reach matters a lot in your product, many teams switch to RICE. RICE adds Reach as a fourth factor and uses effort instead of ease.

  • Use ICE when you need a ranked list quickly.
  • Use ICE when the ideas are similar in size.
  • Use ICE when the team agrees on one target metric.
  • Choose another method when ideas vary hugely in size or risk.
  • Choose another method when reach differs a lot from one idea to the next.
  • Choose another method when decisions must be defended to many stakeholders with detailed data.

How to Run Your First ICE Scoring Session

A first session should take about an hour for a list of 15 to 30 ideas. Send the list and the scoring guide one day ahead so people arrive ready. Ask each person to score alone before the meeting. This stops the loudest voice from setting every number.

In the meeting, show all scores side by side in a shared sheet in a tool like Excel, Google Sheets or Notion. Skip the ideas where everyone agrees. Spend your time only on the ideas with big gaps between scorers. Those gaps usually reveal a hidden assumption, like a different reading of what "Impact" means or missing facts about effort.

End the session with three outputs:

  • A ranked list with final scores.
  • A short list of ideas you will commit to.
  • A "test first" list of promising ideas with low Confidence.

How ICE Scores Feed Your Roadmap

An ICE ranking is a list, not a plan. The roadmap is where you decide the order, the timing and the milestones. Take the top ideas and group them into phases based on dependencies and team capacity. For example, the onboarding checklist might sit in phase one, a quick test for the integration in phase two, and the full integration in phase three if the test succeeds.

Keep the scores attached to each roadmap item. When a stakeholder asks why an item comes first, you can point to the score and the evidence. When a test result arrives, update Confidence, rescore, and move items if needed. That loop is what keeps a roadmap honest over time.

Common mistakes to avoid

  • Scoring without a written scale, so Impact means different things to different people. Fix it with a one-page guide that defines scores 1, 5 and 10.
  • Treating Confidence as enthusiasm, which inflates weak ideas. Fix it by tying each Confidence score to a named piece of evidence.
  • Letting one person score everything alone in a team setting, which hides blind spots. Fix it by having two or three people score separately first.
  • Comparing ideas of very different sizes on the same scale, which favors small tweaks. Fix it by splitting big ideas into smaller, testable steps.
  • Scoring once and never again, so the ranking goes stale. Fix it by setting a fixed rescoring date after each release or each month.
  • Following the score blindly, even when a strategic or legal need says otherwise. Fix it by reviewing the top ten with judgment before you commit.

Frequently asked questions

What does ICE stand for in the ICE scoring model?

ICE stands for Impact, Confidence and Ease. Impact is how much an idea moves your target metric. Confidence is how sure you are about that Impact. Ease is how little effort it takes to ship.

How do you calculate an ICE score?

Score each idea from 1 to 10 on Impact, Confidence and Ease. Then multiply the three numbers to get a score between 1 and 1,000. Some teams average the three instead. Pick one method and use it for every idea.

What is the difference between ICE and RICE?

RICE adds Reach, the number of people an idea affects in a set period. It also uses Effort, usually in person-weeks, instead of an Ease score. ICE is faster and simpler. RICE is more detailed and works better when reach varies a lot between ideas.

How often should you rescore ideas with ICE?

Rescore whenever new evidence arrives, such as a test result or fresh user data. On top of that, set a regular rhythm, for example after each release or once a month. Regular rescoring keeps the roadmap in line with what you have learned.

Generate this roadmap with AI