Start with the decision, then the model
A trading meeting, a creative review and an investment decision ask different questions. A trading team may need the order total after cancellations and returns. A creative team may compare ads inside one platform. A budget owner may need evidence of additional demand. Give each discussion a named measurement basis instead of expecting one column to answer every question.
For a fashion business, returns, delayed purchases and seasonal collections add context. A sale assigned to a channel this week may relate to a product discovered earlier, and its final commercial value may change after a return. Write down the date basis and revenue definition alongside the model.
Keep three views distinct
Use a consistent order or financial source to describe the business result. Use an explicitly named attribution report to describe how credit is allocated. Use a suitable experiment when the question is whether an action caused additional purchases. These views can inform one another without becoming interchangeable.
When multiple platforms report revenue against the same purchase, adding the reports does not produce a new company revenue total. Nor does reducing their totals proportionally establish which individual orders overlap. That requires suitable transaction-level evidence.
Choose a useful starting point
Begin with one closed reporting period, a clear purchase definition and a question the team can act on. Check currency, timezone, account filters, channel mapping and the treatment of returns. Record which fields are unavailable rather than replacing uncertainty with zero.
The guides below separate simple allocation rules, platform-specific behaviour and the distinction between attribution and incrementality. Start with the page matching the decision in front of you; a more complex model is not automatically a more useful answer.
One €120 order, three ways to allocate credit
Worked example: a shopper discovers the shop through organic search, returns through a paid search ad and finally buys after clicking an email. Assume all three touchpoints are observed and eligible within the same agreed window. The order value used in this example is €120.
| Model | Organic search | Paid search | |
|---|---|---|---|
| First click | €120 | €0 | €0 |
| Last click | €0 | €0 | €120 |
| Linear | €40 | €40 | €40 |
Each row still totals €120: the model changes the allocation, not the order. Do not add the rows or sum platform claims. Data-driven attribution cannot be calculated from this journey alone. To establish whether a channel caused extra sales, use a suitable incrementality experiment.
Use this at work
- Name the business question and decision owner.
- Define the purchase value, reporting period and source.
- Inspect the attribution rule, scope and available interactions.
- Record what the comparison cannot establish.
Reference material
Platform guidance checked 3 October 2026. Examples and working checklists are Faccelerate editorial illustrations.