AI Shopping Attribution: Measuring AI-Referred Commerce

The input is a referring domain (chatgpt.com, perplexity.ai, copilot.microsoft.com), a campaign parameter, or another identifier you control. The outputs are classified visits, sessions, checkout events and purchases that meet the attribution rules you set.

A good report states its window and how it treats unknown sources. Purchases with no detectable signal stay unattributed rather than being assigned to AI by assumption. In practice many AI-influenced orders fall into that bucket, because assistants often strip referrers and shoppers return directly later, so the attributed number is a floor.

A practical attribution pipeline has five explicit steps:

1. Classify the entry signal. Record the referrer, campaign parameters, landing URL, timestamp, and available platform identifier.
2. Persist session context. Carry the classified source through pageviews and checkout without replacing a stronger known source with an unknown one.
3. Collect conversion events. Record checkout and order events with the minimum identifiers needed for an approved join.
4. Apply the model. Declare the attribution window, eligible event sequence, cross-device limits, and treatment of direct returns.
5. Deduplicate. Prevent one order from being counted more than once across browser, server, retry, or platform events.

Reports should separate unattributed orders from zero-value outcomes. An order without a detectable signal is unknown for this model; it is not evidence that AI played no role.

Attribution is the last measurement in the chain. Before it come the questions shoppers ask, which products appear in the answers, and the product information changes you made. Read the four together: configured question, assistant, appearance type and position on one side, AI-referred sessions and orders on the other.

A before-and-after rise is directional evidence. It becomes stronger when you track from a baseline, change one thing at a time where you can, and watch the same question set over the following weeks. This is how Paz reports: appearances in AI answers over time next to the AI-referred traffic in your analytics, with the Paz tracking pixel where you want finer detail. See the visibility methodology and the AI visibility service.