Step-by-Step Guide

ICE Scoring Template: How to Set Up ICE for Your Team

Five numbered steps to rank your backlog with ICE. Formula and 1-10 scoring rubrics included.

Last updated: June 2026

Download the free ICE scoring template

Columns for Impact, Confidence, and Ease on 1-10 scales with three worked examples and live ICE score formulas in the sheet. Opens in Excel, Google Sheets, and Numbers with the score formulas already wired up. No email required.

1
List every candidate item you want to score

ICE is built for speed, so score a fixed batch in one short session. Sprint planning and growth experiment reviews are the usual moments.

Pull ideas from your backlog, experiment board, and recent customer feedback.
Cap the batch at 10-20 items. ICE on 20 items takes about 15 minutes; ICE on 100 takes all afternoon.
Merge duplicates into one entry with a clear, specific name.
Write a one-line description for anything ambiguous so everyone scores the same thing.

Formula

Candidate list = one batch of clearly named initiatives, scored in one sitting

Pro tip: Set a timer. Ten minutes for ten items forces intuitive scoring, which is the point of ICE. If a score needs a debate, mark it and move on.

2
Score Impact (1-10)

Impact is how much the item moves your goal metric if it works. Score it 1-10 against a shared rubric, not gut feel alone.

10 = transformational change to a key metric. 5 = moderate, noticeable improvement. 1 = barely measurable.
Anchor scores in a metric: a checkout fix expected to lift conversion 20% is not the same Impact as a copy tweak.
Force ranking keeps scores honest. In a batch of 10, allow at most 2 items to score 8 or higher.
Default new or unclear items to 5 to avoid anchoring high.

Formula

Impact = 1-10 score for how much this moves the needle

Pro tip: If everything in your batch scores 6-8, you are not prioritizing. Spread scores across the full 1-10 range or the ranking tells you nothing.

3
Score Confidence (1-10)

Confidence is how sure you are that the Impact score is real. It is the honesty check that stops gut feelings from topping the list.

10 = backed by data, A/B test results, or direct user research. 5 = educated guess. 1 = pure speculation.
Require evidence for any Confidence above 7: a link to research, analytics, or a comparable shipped feature.
Untested assumptions default to 5, no matter how excited the room is.
If more than half your batch sits at Confidence 4 or below, run validation experiments before building anything.

Formula

Confidence = 1-10 score for evidence quality behind the Impact estimate

Pro tip: Gut-feeling items often get scored 8-9 and perform like 3-4. Asking 'what evidence supports this?' out loud before each Confidence score fixes most of it.

4
Score Ease (1-10)

Ease is how little work the item takes to ship. Note the direction: a HIGH Ease score means LOW effort. 10 is a same-day change, 1 is months of complex work.

10 = ship today, trivial change. 7-8 = a few days. 5 = a few weeks. 1-3 = months of complex development.
Include design, QA, and review time, not just the engineering estimate.
Calibrate against history: if your past "Ease 8" items took two weeks, score similar items lower this time.
When in doubt between two Ease scores, pick the lower one.

Formula

Ease = 1-10 score, where 10 = minimal work and 1 = months of effort

Pro tip: Ease inflation is the number one reason ICE predictions fail. Teams consistently underestimate complexity, and the math rewards it because Ease multiplies the score.

5
Compute the ICE score and rank the batch

Multiply the three scores together. The result ranges from 1 to 1,000, and the ranking within your batch is what matters, not the absolute number.

Example: Impact 7 x Confidence 8 x Ease 9 = 504. Impact 9 x Confidence 4 x Ease 3 = 108.
Sort descending and discuss the top items as a team before committing.
Items with high scores AND high Ease (7+) are quick wins. Batch them into a sprint for fast momentum.
Re-score on a regular cadence, weekly or per sprint. Stale scores reflect stale assumptions.

Formula

ICE Score = Impact x Confidence x Ease

Pro tip: The score is an input to the conversation, not the decision. If the ranking surprises you, check whether the scores are wrong or your intuition is. Either answer is useful.

Rank Your Backlog in the Browser

Skip the spreadsheet. The free ICE calculator scores with keyboard shortcuts, ranks as you type, and flags inflated Ease scores automatically.

Open Free ICE Calculator

Frequently Asked Questions

Who created ICE scoring?

ICE was popularized by Sean Ellis, the growth practitioner behind GrowthHackers, as a rapid prioritization method for growth teams running large experiment backlogs. Its simple 1-10 scales were a deliberate choice: any team member can score in seconds, which keeps experiment velocity high.

How is ICE different from RICE?

RICE adds a fourth factor, Reach, which quantifies how many users an initiative affects, and it divides by Effort instead of multiplying by Ease. ICE drops Reach entirely and uses simple 1-10 scales for all three inputs. The trade is precision for speed: RICE rewards careful estimation, ICE rewards fast consensus. Many teams use ICE for experiments and switch to RICE for formal roadmap planning.

What's a good ICE score?

There isn't one. ICE scores are relative rankings, not grades. A score of 600 beats a score of 300 within the same batch, but there is no universal threshold an item must clear. The maximum possible score is 1,000 (10 x 10 x 10), and most real items land in the low hundreds. Always compare scores against other items scored in the same session by the same people.

What's the difference between Ease and Effort?

Direction. Ease scores high when the work is small: a 10 means you can ship today, a 1 means months of complexity. Effort, the E in RICE, works the opposite way and sits in the denominator: more person-months means a lower score. Both punish big projects, but mixing up the direction when scoring is a common mistake. In ICE, easy items get high numbers.

When should I use ICE instead of RICE?

Use ICE when you don't have reach data or don't have time to gather it: early-stage products with few users, internal tools without user counts, growth experiment backlogs with 50+ ideas, or sprint planning sessions where you need a ranked list in 15 minutes. Once you have real user data and the stakes justify the extra estimation work, RICE gives you a more defensible ranking.