ICE scoring

ICE ranks experiment ideas by three factors, each 1-10. Product is the score.

Impact

How big if it works?

Confidence

How sure are you it'll work?

Ease

How fast can you ship and test?

Output

Rank backlog by ICE score. Run top-scored first.

Critique

Subjective scoring. Use for relative ordering, not absolute decisions.

What to do with this

• Score every experiment before committing engineering time, ICE prevents the team from chasing flashy-but-low-impact ideas
• Weight confidence honestly, inflated confidence scores hide the real ranking
• Use ICE for relative ranking within the backlog, not absolute prioritization across teams
• Review ICE scores weekly, context changes (new data, failed test) shift scores and order
• Don't score alone, team discussion surfaces differing perspectives on impact and confidence