Implementation worksheet · 5 min read
A Campaign Attribution Window Decision Worksheet
Choose the window from your observed time-to-conversion rather than from convention: pull the distribution of time between first qualifying touch and conversion, and set the window where the curve flattens — commonly a few days for transactional products and several weeks for considered B2B purchases. Then write down four things: the window per campaign type, whether it is click-based or view-based, what happens when two campaigns both fall inside it, and the date the policy took effect. That last item is what protects you, because changing a window retroactively re-writes historical numbers and makes every previous report unreproducible.
Teams argue about attribution models while the window — usually inherited from a tool default — silently determines the answer. A seven-day window on a product with a thirty-day consideration cycle will under-report every campaign equally, which looks like consistency rather than error.
Put it into practice
1. Plot your own time-to-conversion
From qualifying touch to conversion, across at least a quarter. The shape of that curve is the argument; anything else is a preference.
2. Set the window where the curve flattens
Capturing 80-90% of conversions is usually the right trade. Extending further adds noise and credits touches that plausibly had nothing to do with the outcome.
3. Differentiate by campaign type
A password-reset nudge and a win-back campaign do not share a conversion horizon. One window across all campaign types is simple and wrong in both directions.
4. Decide the overlap rule
When two campaigns fall inside the window for one conversion: first touch, last touch, split, or both credited in separate reports. Any of these is defensible; not deciding is not.
5. Version the policy with a date
Record when it changed and never restate history under the new window without labelling it. Silent retroactive changes are how a quarter of reporting becomes unreproducible.
Window decision worksheet
Copy this structure into your review document and record your observed result for each row.
| Campaign type | Observed 80th percentile | Window chosen | Overlap rule |
|---|---|---|---|
| Activation / onboarding | |||
| Feature adoption | |||
| Win-back | |||
| Renewal / expansion | |||
| Transactional nudge |
A failure worth checking
The retroactive extension: the window is widened from 7 to 30 days to 'capture more of the effect', historical reports are regenerated, and every past campaign improves at once. Nobody did anything differently. The programme now has two irreconcilable versions of its own history, and the next genuine improvement is indistinguishable from another methodology change.
Common questions
Should view-based touches count at all?
For email and in-app messaging, an open is weak evidence of influence and a click is strong evidence. If you count views, report them separately rather than blending — the blended number is the one people stop trusting.
How often should the window be revisited?
Annually, or when the product's buying cycle genuinely changes. More often than that and you are tuning the metric rather than measuring with it.
Basis and scope
This is a proposed implementation method using illustrative examples, not a measured benchmark or a customer case study. Prepared with AI assistance. Validate product-specific behavior against current documentation and your own test environment.