Conversions Matter | Frameworks
PEEP: A practical framework for evidence-led reasoning
Point. Evidence. Explanation. Prediction.
A simple way to make sense of information, explain your reasoning and update what you believe when new evidence comes along.
Most analytics reports are very good at telling you what happened. Fewer explain why it might have happened. Fewer still tell you what you should expect to see next.
PEEP helps bridge that gap. Not by making every observation into a complicated investigation, but by making the reasoning behind meaningful insights visible and testable.
Four steps. Better reasoning.
Point
What am I observing, proposing or trying to understand?
Evidence
What do I know? What supports or challenges my point?
Explanation
What's the most reasonable explanation of the evidence?
Prediction
If that explanation is reasonable, what should happen next?
These aren't rigid steps. New evidence might change your explanation, challenge your prediction or even overturn the original point.
The aim isn't to prove yourself right
It's to develop the most reasonable explanation the available evidence supports.
That means looking for information that challenges your assumptions, considering other explanations and making predictions that could turn out to be wrong.
When a prediction fails, you've learned something. You might have ruled out one explanation, uncovered a missing variable or opened up an entirely new avenue of investigation.
The important thing is to update what you think rather than defend what you thought.
Where PEEP earns its keep
Weekly and monthly analytics reporting
You don't need a PEEP record for every metric movement. Use it for the findings that matter.
Record what you think is happening, the evidence supporting that explanation and what you expect to see next.
At the following reporting cycle, revisit the prediction. Did it hold up? Has the evidence changed? Do you need a better explanation?
Over time, reporting becomes a record of evolving understanding rather than a series of disconnected observations.
CRO insights and hypotheses
PEEP helps connect behavioural observations to plausible explanations and expected outcomes.
That reasoning can inform a test hypothesis, additional research, a direct fix or a decision to monitor performance.
PEEP doesn't tell you that everything needs an A/B test. It helps explain why you believe something and what you'd expect to happen if you're right.
From observation to explanation
Consider a monthly analytics report showing that mobile conversion has fallen by 15%.
That's a fact worth investigating, but it isn't an explanation.
The following example is illustrative, not actual client data.
Point: Mobile conversion is down 15%. What's driving it?
Evidence: Paid social traffic has increased, returning visitor conversion is broadly stable, and desktop hasn't experienced the same decline.
Explanation: The changing acquisition mix may be contributing to the aggregate decline, rather than a general deterioration in the mobile experience. A mobile-specific issue remains possible.
Prediction: If traffic composition is the main driver, conversion within comparable audience and channel segments should remain relatively stable. If those segments have also declined, we need to reconsider the explanation.
What did we observe? Did the prediction hold? What do we now believe, and what evidence should we examine next?
That's where PEEP becomes more than a way of writing commentary. It creates a record of why you believed something and gives you a reason to revisit it.
Think informally. Record selectively.
You can use PEEP in your head, in a conversation or while working through an unfamiliar problem. It doesn't need a form or a spreadsheet.
Formal documentation is most useful when an insight matters enough to influence a decision or deserves to be revisited later.
In a monthly report, that might mean recording three meaningful PEEP insights rather than writing one for every chart.
The level of documentation should match the importance of the question.
Where PEEP came from
My history teacher used to tell us to PEE in essay questions.
Make a Point. Provide Evidence. Give an Explanation.
Years later, working in digital analytics, I kept seeing analysts present observations as insights.
"Bounce rate increased by 10%."
That's evidence. It tells us something happened. It doesn't explain why.
As I moved further into experimentation and conversion optimisation, I added another P: Prediction.
If our explanation is reasonable, what else should we expect to observe? What happens if we intervene? What happens if we do nothing?
That's how PEEP evolved. A familiar approach to constructing an argument became a simple way of investigating, explaining and revisiting how things work.
Standing on the shoulders of established reasoning
PEEP isn't a new scientific method, and it doesn't pretend to be.
It simplifies ideas that have existed for a long time.
Abductive reasoning helps us construct plausible explanations from observations.
Scientific reasoning encourages us to derive predictions from explanations and evaluate them against evidence.
Bayesian reasoning provides a formal approach to updating beliefs as evidence changes. PEEP doesn't require Bayesian calculations.
Systems thinking reminds us to consider relationships, interactions and dependencies rather than treating every observation in isolation.
PEEP draws on principles consistent with these approaches. It doesn't replace sound research, causal inference or statistical analysis.
It makes evidence-led reasoning easier to remember, apply and communicate.
Try it in your next report
Pick one finding that matters.
Write down the point, the evidence, your most reasonable explanation and a prediction you can revisit.
Next month, come back to it.
If the evidence disagrees with you, change the explanation.
