Glossary
Revenue attribution: definition and examples
Revenue attribution assigns credit for a verified payment to an observed acquisition touch. It needs two separate decisions: which visitor made the payment, and which of that visitor’s visits receives credit.
Match the buyer, then choose a touch
AnalyticFast first tries checkout metadata containing the visitor ID. Existing return-page claims, provider customer correlation, explicitly identified email and constrained timing evidence can supply fallback matches. These signals have different confidence; timing is an inference, not an identifier. A signed webhook verifies a payment but does not by itself reveal its original traffic source.
After matching a visitor, the default model credits the latest observed non-direct visit before payment. The first-touch model credits the earliest observed visit. Changing models changes source allocation, not the amount the customer paid.
A worked example
A visitor arrives from a tagged X post, returns through a Google result, then types the URL directly and buys for $40. With a successful metadata handoff, first touch credits X; last non-direct touch credits Google. The direct return does not replace Google in the default model. A later $10 refund reduces the payment’s net contribution to $30.
What attribution cannot establish
Credit is not proof of causation. A missing referrer, deleted browser storage, a second device or a missing checkout handoff can hide part of the journey. Keep unattributed revenue visible instead of forcing every dollar into a campaign. Analytics revenue also differs from a payout after fees, taxes or settlement timing.
Related definitions
Install AnalyticFast and connect verified revenue to use these definitions with your own data.
Questions, answered.
Does a signed payment webhook guarantee a known source?
No. It establishes the financial event. Source attribution also requires a usable match to a tracked visitor and visit. If that context is missing, the payment can remain unattributed.
Is attribution a causal measurement?
No. It is a rule for assigning observed credit. To estimate incremental impact, use a suitable experiment or other causal method alongside attribution.