Define the evidence

When analytics attributes a visit to an AI service, inspect the available context and the tool’s definition before drawing conclusions. The visit may be useful, but it does not prove that an AI answer recommended the product or that the visitor became a customer. Keep referral traffic separate from manually observed citations and from self-reported discovery. Each offers a different piece of evidence about how the publication is being encountered.

Make it operational

Check the landing page, relevant on-site actions and any qualified enquiry that can appropriately be connected to the visit. Avoid sending private enquiry text into analytics to make the connection easier. Ask an optional discovery question in the enquiry process if that information would help. Report attribution limitations and small sample sizes plainly. Use the findings to improve the usefulness of the landing article and next step, rather than treating a new source label as proof of a complete acquisition channel.

Imagine this

Hypothetical example

A guide receives visits attributed to an AI service. One visitor later reports that an AI answer helped them find the article, while other visits have no further context. The team records the supported observations and does not assume every session represents a recommendation or a commercial lead.

Make it practical.

Use this checklist as you read. Checks are not stored.

References & further reading

Further reading: Search Console generative AI performance report ↗

Published by Tap Tap Go. Product availability and eligibility vary by programme and market. About our stories.