Are ChatGPT and Claude Sending You Leads? How to Track AI Website Referrals

Decide which question the report should answer

A business owner asks whether being recommended in AI chat is producing work. The website report shows a few visits from an AI-related domain. That is a useful starting point, but it cannot establish how often the business was mentioned, what every visitor saw or whether a signed project began with that recommendation.

Start with a narrower question: which visits have an identifiable AI-related source, which pages did those visitors reach and which recorded inquiries followed? Keep search exposure, website behaviour and the team’s qualification decisions in separate columns. This lets you improve a real page without inventing a complete attribution story.

Google’s Traffic acquisition documentation describes session source and medium for understanding where visits originate. Its traffic-source scope guide distinguishes first-user, session and event scopes. Choose one consistent basis for the report rather than comparing unlike dimensions.

Build the source list from observed data

In your analytics tool, inspect the actual source values recorded for a comparable period. Look for visits associated with ChatGPT, Claude and other assistants, then verify what each value represents. Domain names are examples to investigate; do not assume every provider, app and browser supplies an identical referrer.

If you use GA4, start with Traffic acquisition and Session source / medium. Preserve an unfiltered view before creating a comparison. Keep paid campaigns, ordinary search visits and identifiable assistant referrals distinguishable so a campaign change does not quietly become an apparent organic-AI gain.

Check redirects and completion signals

Follow a real, publicly available assistant link to your site when you have one. Note whether the journey changes domains or redirects to a different page. A developer can check what happens to referral and campaign information along that route. Fix confirmed losses in the journey before interpreting a change as declining interest.

Google’s campaign URL guide explains how campaign parameters populate acquisition reporting. Use consistent tags for campaigns you control. Do not add artificial campaign labels to internal service links or manufacture an AI source for traffic whose origin you cannot establish.

The inquiry event also deserves a test. A click on ‘Send’ can happen before validation fails or delivery is rejected. Prefer a defined success event that follows the system accepting the inquiry, then keep the sales qualification outcome in the business’s working records. Test with a clearly labelled internal inquiry when the production owner authorizes delivery.

Treat missing attribution as missing attribution

Someone might read a recommendation, copy the address into another browser and return later from a bookmark. A website session report cannot reconstruct that full history from the landing page alone. A direct visit is not proof that the visitor first discovered the business directly, and it is not proof that an AI assistant influenced them.

An optional ‘How did you hear about us?’ question can add context. Allow a free-text answer and retain the customer’s wording. Treat it as a self-reported source rather than silently overwriting analytics. When both signals exist, they may describe different parts of the same customer journey.

Turn the observation into a page improvement

Suppose identifiable assistant referrals arrive on a website-cost article while the inquiries ask what is included in a redesign. That is a reason to inspect the connection between the article, the redesign explanation and contact. It is not enough evidence to rewrite every page around the assistant’s name.

Review a comparable period, annotate substantive changes and read the actual inquiry themes. Our Google AI-search measurement guide covers the exposure side. This referral report covers visits and outcomes, which answer a different question.

If the report reveals a confusing landing page or an unreliable inquiry path, discuss the website journey with Menerai. For workflows that need shared qualification records or integrations, our custom software service can help define the operating requirements before a build is proposed.