PIE Framework

The PIE Framework is a structured method for prioritizing which A/B tests and CRO experiments to run first, developed by Chris Goward at WiderFunnel.

Each test idea is scored across three dimensions:

• Potential — how much improvement is possible?
• Importance — how significant is this page or element to the business?
• Ease — how simple is it to implement and run?

PIE Score = (Potential + Importance + Ease) ÷ 3

Tests are ranked by PIE score and run in order from highest to lowest.

Scoring Each Dimension

Potential (1–10)

How much room for improvement exists on this page or element?

Signals of high potential:

• Conversion rate significantly below industry benchmark
• High exit rate on pages that should retain visitors (>60%)
• Rage clicks or dead clicks visible in session recordings
• User testing reveals persistent confusion at this step
• Visitor survey responses mention frustration with this element
• Heatmaps show visitors ignoring the primary CTA

Score 9–10: Page or element has obvious, severe problems. Major improvement opportunity. Score 5–6: Some issues present but page performs reasonably well. Score 1–2: Page already performing well, limited upside.

Importance (1–10)

How much revenue or traffic flows through this page or element?

Signals of high importance:

• High monthly sessions (this is where most of your audience lands)
• Direct connection to conversion (checkout, pricing page, primary CTA)
• High revenue per visitor (paid traffic landing page)
• Part of the critical path — visitors must pass through this page to convert

Score 9–10: Your single highest-traffic or highest-revenue page. Every 1% improvement here multiplies across thousands of monthly visitors. Score 5–6: Moderate traffic; contributing to conversion pipeline but not the primary bottleneck. Score 1–2: Low traffic, niche page with minimal revenue impact.

Ease (1–10)

How easily can this change be implemented and tested?

Score 9–10: Copy change, button text update, trust badge addition — minimal development, deployed in hours. Score 5–6: Layout change, new section, modest design work — few days development. Score 1–2: Full page redesign, new checkout flow, major back-end change — weeks of development and QA.

PIE Scoring Example

Test idea	Potential	Importance	Ease	PIE Score
Rewrite checkout page headline	8	9	9	8.7
Add trust badges to checkout	7	9	8	8.0
Simplify homepage navigation	7	8	5	6.7
Add social proof to pricing page	6	8	7	7.0
Redesign product category page	6	7	3	5.3
A/B test blog CTA copy	4	3	9	5.3
Reduce form fields on lead gen page	8	6	8	7.3

The checkout headline rewrite scores first: high potential (headline is a known high-leverage element), highest importance (checkout is where purchases happen), and easy to implement. The blog CTA test scores identically to the category page redesign but represents dramatically less potential revenue — a reminder that PIE score alone doesn’t tell the whole story.

PIE vs ICE Framework

Framework	Dimensions	Best for
PIE	Potential, Importance, Ease	Identifying which pages to focus on first
ICE	Impact, Confidence, Ease	Ranking specific hypotheses by certainty

Many CRO programs use PIE to select pages and ICE to prioritize individual tests within a page.

ICE scoring in practice:

• Impact: expected CVR lift if the hypothesis is correct (1–10)
• Confidence: strength of evidence backing the hypothesis (1–10)
• Ease: implementation effort (1–10)

A hypothesis with strong research backing (session recordings, user testing, heatmaps all pointing to the same problem) scores high on Confidence. A hypothesis based on a gut feeling scores low. ICE builds research-backing into the prioritization system.

Data Sources for PIE Scoring

PIE Dimension	Data source	What to look for
Potential	GA4 exit rate	>60% exit rate = high potential
Potential	Session recordings	Rage clicks, form abandonment patterns
Potential	CVR vs industry benchmark	50%+ below average = high potential
Potential	Heatmaps	Ignored CTAs, scroll depth drop-offs
Importance	GA4 page traffic	Monthly sessions count
Importance	Revenue flow	Pages in direct checkout path
Importance	Ad spend destination	Paid traffic landing pages score highest
Ease	Development estimates	Time to implement + QA

Grounding each dimension in actual data before scoring reduces the team disagreement that undermines PIE’s usefulness as a prioritization tool.

Building and Maintaining a PIE Backlog

A well-run PIE backlog follows this cycle:

1. Discovery — Pull low-CVR pages from GA4, identify rage clicks in Hotjar/FullStory, review user survey responses
2. Hypothesis formation — Convert each problem into a testable hypothesis: “If we [change X], then [metric Y] will improve because [reason Z]”
3. Scoring — Score each hypothesis on PIE (or ICE) using data anchors, not intuition
4. Ranking — Sort by score, review top 5 for feasibility
5. Execution — Run the highest-scored test that’s ready to build
6. Review — Quarterly re-score of inactive backlog items as site data changes

A backlog of 10–30 items is manageable. Beyond 30, low-priority items become stale and the backlog loses credibility as a prioritization tool.

Limitations of PIE

PIE is a subjective scoring system — different team members will score the same test differently. Reduce subjectivity by:

• Anchoring Potential scores to actual analytics data (exit rate, CVR relative to benchmark)
• Anchoring Importance scores to actual traffic or revenue numbers
• Having multiple team members score independently and averaging

Common PIE mistake: Over-weighting Ease. A high-importance, low-ease test (checkout redesign) will often produce more revenue than 10 high-ease, low-importance tests (blog sidebar CTAs). Ease is a tiebreaker between otherwise equal options — not a reason to skip high-importance work.

Weighted PIE: Some teams assign weights to dimensions (e.g., Importance × 2, Potential × 1.5, Ease × 1) to prevent Ease from dominating the ranking. This is a reasonable modification for teams where Ease scores consistently crowding out high-importance, high-potential tests.

PIE is a structured way to have an informed argument about priorities — not a mathematical truth.

PIE in the Context of a Full CRO Programme

PIE is most effective when paired with a strong research phase. Running PIE scoring without analytics data, session recordings, and user research produces subjective guesses formatted as numbers. Running it with solid data produces informed prioritization.

The research → PIE scoring → testing → learning → research cycle is the engine of a systematic CRO programme. For running tests once priorities are set, see A/B Testing Best Practices. For how PIE fits within the broader CRO strategy, see the full methodology overview.