The ICE Framework is a prioritisation method used by growth and CRO teams to rank experiment ideas and feature changes before committing to building or testing them. ICE stands for Impact, Confidence, and Ease. Each idea is scored on these three dimensions, typically on a scale of 1–10, and the scores are averaged to produce an ICE score that guides which experiments get scheduled first.
ICE Score = (Impact + Confidence + Ease) / 3
- Impact: How much will this change move the key metric if it works? (1 = negligible, 10 = transformational)
- Confidence: How sure are you the change will have this impact, based on data, research, or precedent? (1 = pure gut feel, 10 = strong data-backed evidence)
- Ease: How easy is this to implement in terms of engineering effort, design, and time? (1 = months of work, 10 = can go live today)
An idea scoring 9 on Impact, 7 on Confidence, and 3 on Ease gets an ICE score of 6.3. An idea scoring 6 on all three gets 6.0. Despite lower impact, the second idea might ship first if the team is resource-constrained.
Why ICE Framework Matters for Ecommerce
Ecommerce and CRO teams almost always have more experiment ideas than bandwidth to test them. Without a scoring system, the loudest person in the room or the most recent customer complaint drives the roadmap — neither is a good proxy for revenue impact. ICE creates a shared, transparent language for prioritisation that keeps the team focused on high-value experiments. For Shopify stores where engineering time is limited, the Ease component is particularly valuable because it surfaces quick wins that can ship in days.
Real-World Example
The CRO team at a D2C skincare brand (similar to mCaffeine) has 15 experiment ideas on their backlog. One idea is to add a trust badge near the buy button — Impact: 6, Confidence: 8, Ease: 9 → ICE: 7.7. Another idea is to rebuild the product page layout entirely — Impact: 9, Confidence: 5, Ease: 2 → ICE: 5.3. The trust badge test ships first because it has a higher ICE score, even though the page rebuild has higher potential impact. Three weeks later, the badge test returns a result. Now the team is one learning richer and the page rebuild can be re-evaluated with new data.
How to Improve / Optimize ICE Scoring
- Score as a team, not solo: Have product, CRO, and engineering score ideas independently, then average the scores to reduce individual bias.
- Be conservative with Confidence: Teams consistently over-estimate their certainty. If you don't have data or a proven precedent, cap Confidence at 5.
- Revisit scores when new data arrives: A test result, a session recording insight, or a customer interview can change the Confidence score for related ideas significantly.
- Combine with effort estimates: Have engineering provide a rough hour estimate alongside your Ease score to catch cases where your perception of ease was off.
- Don't treat ICE as the only input: Very high-Impact ideas that score low on Ease may still deserve a place on the roadmap — use ICE to start the conversation, not end it.
ICE Framework in A/B Testing
ICE is specifically designed to prioritise what gets tested. Once an idea clears the ICE threshold and enters the experiment queue, it needs a formal hypothesis, success metric, and sample size calculation before the test launches.
Run smarter A/B tests with CustomFit.ai — 14-day free trial, no credit card required.