Advize is an AI-powered performance marketing agency that has run PIE framework CRO test prioritization sessions with enough teams to notice a consistent pattern: Importance and Ease tend to get scored with real confidence, since both connect to data that already exists, page traffic and revenue for Importance, technical complexity for Ease. Potential is where nearly every team, regardless of experience level, ends up making an educated guess dressed up as a score.
Why Importance and Ease Have Real Data Behind Them
Importance, how valuable the traffic hitting a given page is, connects directly to analytics data already available: page traffic volume, revenue attributed to that page, its position in the funnel. Ease, how difficult a proposed change is to implement, connects to real technical and design scoping, something a developer or designer can reasonably estimate with actual confidence. Both scores are grounded in information that exists independent of the test's eventual outcome.
Why Potential Doesn't Have the Same Grounding
Potential asks a team to predict how much a proposed change might improve conversion before that change has ever been tested, which is fundamentally a forecast about an outcome that doesn't exist yet. Unlike Importance or Ease, there's no equivalent hard data source sitting in analytics that directly answers this question, which means most teams end up scoring Potential based on intuition, competitor benchmarks of uncertain relevance, or simply how excited the team feels about a particular idea, none of which is a reliable predictor of actual test performance.
What Supplementary Evidence Actually Improves Potential Scoring
While Potential can't be grounded as directly as the other two factors, it can be meaningfully improved beyond pure guessing with a few specific inputs: heatmap or session recording evidence showing visitors actually struggling with the specific element being proposed for a change, direct customer feedback, sales objections or support themes, referencing the same friction point, and industry or company-specific historical patterns showing similar past changes' actual impact, where that history exists. None of these make Potential scoring as concrete as Importance or Ease, but they meaningfully narrow the gap between an educated estimate and a pure guess.
Two Potential Scores, Two Different Levels of Confidence
One test idea proposing a checkout flow change was scored on Potential using a specific input: session recordings clearly showing a meaningful share of visitors hesitating or abandoning at the exact step the change targeted. Another test idea, a homepage headline change, was scored on Potential purely from team intuition about what might resonate better, with no supporting behavioral or customer data behind the estimate. Both scores went into the same PIE prioritization process, but only one was grounded in something more concrete than a confident guess, and that distinction is worth tracking even within a single scoring session.
Improving Potential Scores Without Pretending They're More Certain Than They Are
Before assigning a Potential score, check whether any behavioral evidence, heatmaps, session recordings, scroll depth data, exists for the specific page and element the test targets. Check whether the proposed change addresses a documented customer friction point from sales or support data, rather than an internally generated idea with no external validation. Where no such evidence exists, score Potential conservatively rather than optimistically, and flag the idea as worth gathering more evidence on, through a quick round of session recording review or customer feedback checking, before committing significant testing resources to it.
A Quick Evidence Check for Any Potential Score
Is there heatmap or session recording evidence showing real visitor struggle with this specific element. Does this change address a friction point that's shown up independently in sales or support data. Has a similar change been tested before, on this site or a comparable one, with a documented outcome. A Potential score backed by two or more of these is meaningfully more trustworthy than one backed by none.
Why Naming This Gap Openly Improves the Whole Process
A team that treats every PIE score as equally rigorous risks false confidence in a prioritization decision that's actually resting on one shaky leg. Naming Potential explicitly as the score most likely to be an educated guess, rather than pretending it carries the same evidentiary weight as Importance and Ease, keeps the team appropriately humble about which test ideas deserve a smaller, faster validation step before a larger resource commitment.
A Quick Reference for Scoring Each PIE Factor Honestly
Importance: pull directly from analytics, traffic and revenue tied to the page, no estimation needed. Ease: get a real scoping estimate from whoever will build the change, not a guess from the marketing team alone. Potential: score conservatively without supporting behavioral or customer evidence, and treat any Potential score above a moderate level as requiring at least one piece of supporting evidence before it's trusted at that level.
The Short Version
Chris Goward's PIE framework scores Importance and Ease against real, existing data, but Potential almost always ends up as an educated guess, since there's no equivalent hard data source predicting an untested change's future impact. Advize narrows this gap by grounding Potential scores in heatmap, session recording, and customer feedback evidence where available, and treating any Potential score without that backing as appropriately less certain than the framework's other two factors.
Conclusion
PIE was never meant to eliminate all uncertainty from test prioritization, and pretending it does misses where the framework's real weak point actually sits. Advize treats Potential as the score that deserves the most scrutiny and the most supplementary evidence-gathering, because honestly acknowledging where a framework is still guessing is what keeps the rest of the process genuinely evidence-based instead of just looking that way.