Is ChatGPT Sending You Customers? Measure AI Traffic in Matomo

“AI traffic” can mean a person clicking a ChatGPT recommendation or an automated system requesting your content. Keep those separate when reporting results. A chatbot fetching a page is not a customer visit or evidence of a conversion.

Matomo provides AI-related reporting, but availability and collection requirements depend on the report and your installation. Start by naming the question you want to answer.

Are people arriving from AI recommendations?

Look at acquisition and referrer reports for recognised AI assistant sources. This measures attributed arrivals when the necessary referrer information is available. It does not reveal every time your brand was mentioned or every person who saw a recommendation.

Record the date range, referred visits, landing pages and relevant goals if they are configured. A click is an acquisition signal; whether it produced a useful outcome is a separate question.

Are AI agents accessing the site?

Matomo’s AI Agent reporting addresses automated activity detected through supported signals. Interpret it separately from the people referred by assistants. Do not add the two categories together and call the result “customers from AI.”

Are chatbots requesting content?

Some chatbot reporting requires server-side integrations rather than ordinary browser tracking. Matomo documents integrations for environments including Cloudflare and CloudFront. Follow the instructions for your actual infrastructure before expecting data.

An empty chatbot report might indicate missing integration or no detected activity. It does not establish that AI tools never access or mention your site.

Build a small AI acquisition report

For each completed month, record attributed assistant referrals, the pages they reach and the business outcomes you can measure. Compare source volumes and outcomes separately. Small samples can produce unstable conversion rates; keep the underlying counts visible.

For content decisions, ask: which pages receive attributed arrivals, what do those pages offer, and what action do visitors take next? Improve the useful next step rather than chasing a raw chatbot-request count.

Where Savigny helps

Savigny’s current tools retrieve traffic sources, referring websites and leading pages. They do not expose dedicated AI Agent or AI Chatbot reports, and they do not provide goal or conversion reports.

When your Matomo referrer data contains identifiable AI domains, an assistant may help summarise those returned rows. Ask it to state the domains and limitations rather than infer an AI category from ordinary direct traffic:

For the selected website and completed date range, inspect referring
websites. Identify any returned domains that correspond to AI
assistants. Show their original visit counts and flag that referrer
attribution is incomplete. Do not infer chatbot requests, brand
mentions or conversions from these results.

Check the relevant reports directly in Matomo for a complete AI measurement workflow. Matomo’s terminology guide explains attribution and detection limits.

What this does not measure

Matomo cannot turn a missing referrer into proof of a particular assistant recommendation. Nor does a content request prove that an answer cited the page. Keep visibility, referral visits and outcomes as separate measures.

If your main goal is to use AI to analyse your existing website traffic, follow the ChatGPT connection guide. That workflow solves a different problem from measuring AI-originated traffic.