How to prioritize growth experiments
Learn how to prioritize growth experiments using proven frameworks like ICE, RICE, and PIE. Stop running random tests. Start focusing on the experiments most likely to drive results.
Why prioritization matters
The average growth team has 50+ experiment ideas but can only run 2-4 at a time. Without rigorous prioritization, teams waste time on low-impact experiments while high-impact opportunities sit in the backlog. Good prioritization is leverage: it can multiply your experimentation ROI several times over without adding a single extra test.
The frameworks: ICE, RICE, and PIE
Three proven scoring frameworks. Start with ICE, then evolve based on what your team needs.
ICE Score
Impact, Confidence, Ease. The most popular framework for growth experiments. Simple, fast, and effective for most teams. Formula: ICE = (Impact + Confidence + Ease) / 3.
- Impact: how much will this move the needle? (1-10)
- Confidence: how sure are we this will work? (1-10)
- Ease: how easy is this to implement? (1-10)
Best for growth teams, startups, marketing experiments. Pros: simple and fast, easy to explain. Cons: subjective scoring, ignores dependencies.
RICE Score
Reach, Impact, Confidence, Effort. A more rigorous framework that adds Reach as a multiplier, common with product teams. Formula: RICE = (Reach times Impact times Confidence) / Effort.
- Reach: how many users per quarter?
- Impact: impact per user (0.25 to 3)
- Confidence: percentage confidence
- Effort: person-months of work
Best for product teams and larger initiatives. Pros: more objective with Reach, accounts for effort. Cons: needs more data, can favor small optimizations.
PIE Framework
Potential, Importance, Ease. Built for conversion rate optimization and website testing. Formula: PIE = (Potential + Importance + Ease) / 3.
- Potential: how much room for improvement? (1-10)
- Importance: how valuable is this traffic? (1-10)
- Ease: how easy is the test? (1-10)
Best for CRO teams, website optimization, e-commerce. Pros: purpose-built for CRO, considers traffic value. Cons: less applicable outside CRO, ignores confidence.
Worked example: scoring five experiments
The same backlog scored on ICE and RICE. Notice how the two frameworks rank the experiments differently.
| Experiment | Impact | Confidence | Ease | ICE | RICE |
|---|---|---|---|---|---|
| Onboarding email sequence test | 7 | 8 | 9 | 8.0 | 16,000 |
| Pricing page redesign | 9 | 6 | 4 | 6.3 | 5,000 |
| Homepage headline test | 5 | 7 | 10 | 7.3 | 64,000 |
| Feature tutorial videos | 6 | 5 | 3 | 4.7 | 750 |
| Checkout simplification | 8 | 9 | 6 | 7.7 | 6,400 |
ICE favors the high-confidence onboarding test; RICE favors the high-reach headline test. The 'right' answer depends on whether reach or confidence matters more to you right now.
Common prioritization mistakes
Five patterns that quietly undermine a scored backlog, and how to fix each.
01. Optimizing for Easy over Impact
Teams often prioritize easy wins, but a series of 1% improvements rarely compounds to significant growth.
Fix: Reserve 20-30% of capacity for 'big bet' experiments with higher risk but higher potential impact.
02. Overconfident Scoring
Teams consistently overrate their confidence, leading to disappointing experiment results.
Fix: Use historical data to calibrate. If 80% confidence experiments only win 40% of the time, adjust scoring.
03. Ignoring Dependencies
Some experiments block or enable others. Pure prioritization ignores this.
Fix: Identify 'unlock' experiments that enable future tests and weight them accordingly.
04. Not Revisiting Priorities
Priorities change as you learn. A backlog scored 3 months ago may be stale.
Fix: Re-score top experiments monthly. Learnings from recent experiments should inform scores.
05. Scoring in Isolation
One person's 8 Impact is another's 5. Inconsistent scoring undermines the framework.
Fix: Establish scoring calibration sessions and reference examples for each score level.
Frequently asked questions
How to prioritize growth experiments based on data insights?
Prioritize experiments using data by: 1) Analyze funnel metrics to find the biggest drop-offs, these are high-potential experiment areas. 2) Use historical experiment data to calibrate confidence scores. 3) Calculate potential reach from actual user data, not estimates. 4) Weight by revenue impact. 5) Use cohort analysis to identify user segments with the biggest improvement opportunities. 6) Track experiment velocity by area.
What growth experiment frameworks do top marketing agencies recommend?
Top agencies recommend these frameworks: 1) ICE Score, most popular for its simplicity. 2) RICE Score, used by product teams. 3) PIE Framework, specifically for CRO. 4) Opportunity Scoring from JTBD methodology. 5) Custom frameworks combining elements. Most agencies recommend starting with ICE, then evolving based on what works for your team.
What is ICE scoring and how do I use it?
ICE scoring is a prioritization framework where you rate each experiment idea on three dimensions (1-10 scale): Impact, Confidence, and Ease. Calculate the ICE score by averaging: (Impact + Confidence + Ease) / 3. Higher scores = higher priority. Tips: be honest about confidence, don't inflate ease to get your idea prioritized, and calibrate scores across the team for consistency.
How do I balance quick wins vs. big bets in my experiment backlog?
Balance quick wins and big bets using a portfolio approach: 1) Allocate capacity, reserve 70% for optimization and 30% for innovation. 2) Define what makes a 'big bet'. 3) Run quick wins in parallel. 4) Give big bets adequate sample size and time. 5) Learn from both. 6) Adjust the ratio based on stage. Early-stage companies may want 50/50; mature products might be 80/20 toward optimization.
How do I get stakeholder buy-in for my prioritized experiments?
Get stakeholder buy-in by: 1) Make the scoring transparent. 2) Include stakeholders in scoring calibration sessions. 3) Show the data behind priority decisions. 4) Connect experiments to OKRs and business goals explicitly. 5) Share results regularly. 6) Acknowledge and address HiPPO ideas. 7) Reserve capacity for stakeholder experiments. 8) Celebrate wins AND learnings from failures. Track the full loop, not only invites sent.
