Decision journal / Measurement and learning
Influenced revenue vs incremental impact
Revenue that follows an experience can be relevant without proving that the experience created extra revenue.
A shopper sees a recommendation, later buys and the order falls inside an attribution window. The relationship matters, but several different claims could describe it. Choosing the strongest-sounding label is not the same as choosing the most accurate one.
Measurement should climb an evidence ladder and stop at the highest level the method genuinely supports.
Build the journey before the claim
Assignment means an experience was selected. Eligibility means it could be shown. Delivery means your storefront confirmed a real shopper saw it. Interaction adds a pre-defined action. An outcome is a server-observed event such as a Shopify order.
Without delivery, there is no valid experience-performance learning. Preview, staff and design-mode activity must stay outside commercial populations.
Use an evidence ladder
Recorded says the source captured an event. Shown before outcome says verified delivery preceded it. Interaction assisted adds a recorded engagement. Associated applies a declared attribution rule. Incremental requires a valid controlled or accepted causal comparison.
Keep the window, population, method version and deduplication visible. Refunds can revise later value without rewriting the original event.
A practical route
- Verify real delivery. Tie the exact experience and version to an eligible session.
- Link the outcome. Use a declared window and priority rule.
- Deduplicate. Prevent one order from being counted across overlapping claims.
- Choose honest language. Use recorded, assisted, associated or incremental according to the method.
Keep these checks visible
- Expose denominators
- Separate gross and net value
- Recommend a test when the decision requires causal evidence
Honest outcome language does not weaken a result. It tells your team exactly how much decision weight the evidence can carry.
Go deeper
Continue this line of thinking.
Read every Measurement and learning decision brief →Editorial path
Measurement and learning
Check what was shown, what happened next and what the evidence can support.
- Influenced revenue vs incremental impact
- Turn a Shopify test into useful learning
- How to know whether a storefront change was delivered