This calculator scores a list of ideas by impact, confidence and ease, each on a scale of 1 to 10, then ranks them as the numbers change. Teams use two formulas, impact times confidence times ease or the average of the three, and the toggle below switches the table between them without re-entering a single number.
| Idea | Impact | Confidence | Ease | Score | Remove |
|---|
The rows above are examples. Replace the names and numbers with your own ideas.
How do I choose each input
Three numbers, each 1 to 10. All three are a guess.
Impact is how much this idea moves the goal it is meant to move, if it works. A 3 is a tweak nobody outside the team will notice. A 6 is a feature a reasonable share of users would use and mention. A 9 is a change to what the product fundamentally does, the kind that shows up in a review.
Confidence is how sure the team is that the impact guess is real, as evidence rather than enthusiasm. A 3 means a hunch with no data behind it. A 6 means a pattern seen in support tickets, competitor moves or a small test. A 9 means a result already measured on your own product, a prior launch of something close enough to count.
Ease is how cheap this is to ship, where high is easy. A 3 is a project that touches several systems and needs design work before anyone can start. A 6 is a week or two of ordinary engineering. A 9 is a configuration change or a small patch a single person finishes this week.
The trap common to all three is scoring from the inside of the idea rather than against the rest of the list. An impact of 9 on every row means the scale stopped telling ideas apart. Score each input relative to the other rows on the same list, not against an abstract ceiling.
Product or average
The product multiplies three weak scores into one very weak score, because a 3 anywhere drags the whole result down by two thirds. The average smooths that out, so an idea with one bad input and two strong ones still lands in the middle of the list.
Use the product when a single fatal flaw, usually low confidence, should sink an idea outright. Use the average when the three inputs are closer to three votes on the same decision and no single low score should disqualify an idea by itself. The toggle above runs both from the same three numbers.
A worked example
justRead, a reading app from the same people who build VoteFirst, runs its own public feature board on VoteFirst. Three of the open requests on that board, as it read on 2026-10-02:
- Sync to Goodreads, 25 votes
- Apple Watch integration, 21 votes
- Convert PDF to reflowable EPUB, 12 votes
The titles and vote counts are real. The impact, confidence and ease numbers below are illustrative guesses, not scores the justRead team gave.
Sync to Goodreads: impact 6, confidence 7, ease 4.
product: 6 × 7 × 4 = 168
average: (6 + 7 + 4) ÷ 3 = 5.7
Apple Watch integration: impact 5, confidence 5, ease 3.
product: 5 × 5 × 3 = 75
average: (5 + 5 + 3) ÷ 3 = 4.3
Convert PDF to reflowable EPUB: impact 7, confidence 4, ease 2.
product: 7 × 4 × 2 = 56
average: (7 + 4 + 2) ÷ 3 = 4.3
Both formulas put Sync to Goodreads first. They split on second place. The product ranks Apple Watch integration above the EPUB conversion, because a low ease costs more under multiplication, and the average puts the two level at 4.3. The board's own vote counts, 25, 21 and 12, give the same order as the product.
When ICE is the wrong tool
ICE has no reach term, so an idea for ten users and an idea for ten thousand score alike when the three inputs match. When the audience differs a lot between rows, the RICE calculator puts a reach number in front of the rest.
Impact and confidence are both guesses about demand. Once a request sits on a public board, demand stops being a guess. In the example above, justRead's vote counts put the three requests in the same order the product formula did, and those counts came from people who asked. VoteFirst ranks requests by a demand score built from votes, comments and views, set against an effort estimate, which leaves ease as the only number anyone has to guess.
ICE stays the quicker tool for ideas the team generates itself, such as growth experiments, where nobody outside has voted yet.
Keeping two scorers from drifting apart
The same three numbers mean different things to different people unless the team writes down what a 3, a 6 and a 9 look like before anyone scores anything, the way the anchors above do. Two people independently scoring the same idea and landing within two points of each other on all three inputs is a sign the anchors are working. A gap wider than that on even one input means the conversation that follows the score matters more than the score itself, and the right fix is to compare notes on what evidence each person used, not to average away the disagreement and move on.
A list scored by one person on a Tuesday and rescored by the same person a month later drifts the same way, usually because the list has grown and the top of the scale has quietly gotten more competitive. Rescoring the whole list together when it is revisited, rather than only adding new rows to the bottom, keeps old and new ideas on the same ruler.