PIE scoring ranks test ideas on three dimensions: Potential, Importance and Ease. Score each from one to ten, average the three, and run the highest first. Its real value is not the arithmetic. It is that four people can disagree about a score in ten minutes rather than arguing about taste for an hour.
What each letter means
Potential. How much room for improvement exists on this page. A page already performing well has less headroom than one that is visibly broken, regardless of how much traffic each gets.
Importance. How valuable the traffic on that page is. Not volume alone. A page with fewer sessions but higher purchase intent outranks a busy page full of bouncing visitors.
Ease. How difficult the change is to implement and test. This includes political difficulty, which most teams leave out and then get blocked by.
Average the three. Run the top score first. Re-score after every concluded test.
A worked example
| Test idea | Potential | Importance | Ease | PIE |
|---|---|---|---|---|
| Show shipping cost on the PDP | 9 | 9 | 8 | 8.7 |
| Add post-purchase upsell | 8 | 7 | 9 | 8.0 |
| Test free shipping threshold value | 9 | 8 | 6 | 7.7 |
| Rebuild collection page filters | 7 | 8 | 3 | 6.0 |
| Change hero image on homepage | 3 | 5 | 9 | 5.7 |
| Button colour on PDP | 1 | 9 | 10 | 6.7 |
Look at the last row carefully, because it exposes the framework’s main weakness. Button colour scores 6.7 and outranks a filter rebuild, purely because it is trivially easy and sits on an important page. Nobody should run it.
Where PIE breaks, and how to fix it
Ease is weighted too heavily. Averaging the three dimensions equally means a very easy, very low-potential test can float to the top. The fix is to treat Ease as a tie-breaker rather than an equal input: rank by Potential and Importance, then use Ease to order within similar scores.
Potential is guessed. Scores get assigned on instinct. Anchor them to evidence instead: a 9 requires an observed user failure in session recordings or a documented cause, a 5 requires an analytics pattern, a 1 is somebody’s opinion.
It ignores confidence. PIE has no dimension for how strong the evidence is. If you only add one thing to the framework, add that.
This matters because most tests fail regardless. ConversionTeam’s audit of 2,288 tests found 19.1% reached statistical significance per test, and Optimizely’s analysis of more than 127,000 experiments puts the average win rate near 12%. When four in five attempts produce nothing conclusive, choosing better is the only lever that changes your annual output.
PIE against ICE
| PIE | ICE | |
|---|---|---|
| Dimensions | Potential, Importance, Ease | Impact, Confidence, Ease |
| Handles evidence quality | No | Yes, via Confidence |
| Handles traffic value | Yes, via Importance | Not explicitly |
| Best for | Page-level prioritisation | Hypothesis-level prioritisation |
In practice the strongest version borrows from both: Potential, Importance, Confidence, with Ease as a tie-breaker. That keeps traffic value in the model and adds evidence quality, which is the dimension PIE is missing.
Scoring honestly
Score as a group, not alone, and make people say their number before hearing anyone else’s. Otherwise the most senior voice sets the anchor and everyone converges on it, which reproduces exactly the problem the framework was meant to solve.
Write one sentence of justification per score. It takes seconds and it exposes the scores that are opinions wearing numbers.
And re-score after every concluded test. A roadmap written once and followed unchanged is a schedule. Every result, including the losses, should change what you believe and therefore what sits at the top. Five minutes on the top five items, once a fortnight, is enough.
What a well-ranked roadmap produces
The clearest example we have seen was not a clever test. At a DTC supplements brand, a shipping threshold test produced two million dollars in profit. It reached the top of the roadmap because it scored high on every dimension: large potential, high-traffic exposure, clear analytics evidence and moderate build cost.
Nothing about it was ingenious. It was ranked correctly, which is most of the job.
Want a ranked test roadmap for your store?
Parah Group builds and runs CRO programs with every idea ranked by expected business impact, confidence and effort.
Frequently asked questions
What is PIE scoring?
A prioritisation framework that scores test ideas on Potential, Importance and Ease, each from one to ten, then averages them to produce a ranking.
What is the difference between PIE and ICE?
PIE uses Potential, Importance and Ease. ICE uses Impact, Confidence and Ease. ICE accounts for evidence quality; PIE accounts for traffic value.
Is PIE scoring accurate?
It is a structure for argument rather than a calculation. Its value is that it makes disagreements explicit and quick, not that the number is precise.
How often should a roadmap be re-scored?
After every concluded test, on the top five items only. Roughly five minutes a fortnight.
If you want a roadmap ranked on evidence rather than on who spoke last, that is what our CRO Program produces, and A/B testing covers how the results feed back into it.
Frequently asked questions
What businesses benefit most from the PIE Strategy?
Any business juggling multiple priorities, especially e-commerce brands and SaaS companies, can benefit from PIE.
What tools can I use to score tasks?
Platforms like Airtable, Trello, or Google Sheets can simplify PIE scoring.
How often should I revisit PIE scores?
Reevaluate quarterly or after major business shifts.
Can I use PIE for non-marketing tasks?
Absolutely! PIE works for prioritizing product development, operational improvements, and more.