AI referral traffic is the set of visits a website receives when a person clicks a link inside an answer written by an AI product, such as ChatGPT, Perplexity, Gemini, Claude or Microsoft Copilot. Analytics tools use the term referral traffic for visits that arrive from a link on another site, and AI referral traffic is one kind of it. The visitor is a human who read an AI answer and decided to open one of its sources.
How does a click in an AI answer become a visit?
AI search products write one answer from several pages and attach links to the pages they used. Those links appear as numbered citations, source cards beside the answer, or inline links in the text. When a person clicks one, the browser loads the cited page, and the website's analytics tool records a new session.
Whether a page gets cited at all is decided before the click. A retrieval step finds candidate pages, a ranking step picks passages, and the model links to the passages it used. How AI search chooses which sources to cite covers that process. Referral traffic comes after it, when a reader acts on a citation.
How do analytics tools recognize AI referral traffic?
Analytics tools classify each visit by where it came from. For AI products they rely on two signals.
- The referrer. When a browser follows a link, it usually sends a header naming the page the visitor came from. A visit with a referrer such as chatgpt.com, perplexity.ai, gemini.google.com, copilot.microsoft.com or claude.ai can be attributed to that product.
- Tracking parameters. Some AI products add a parameter to the link itself, typically utm_source, a tag that analytics tools read from the URL. This works even when the referrer is missing.
Most analytics tools do not yet have a ready-made channel for AI products, so these visits often land in a general "referral" bucket mixed with links from blogs, directories and partners. The usual fix is a custom channel or segment that matches a list of AI domains with a regular expression, which is a short text pattern that tests whether a source name contains any of the listed domains.
Why is AI referral traffic hard to measure accurately?
The number in an analytics report is almost always lower than the real number of visits that started with an AI answer. Several mechanisms cause this.
| Situation | What analytics usually records |
|---|---|
| Click from a desktop browser with the referrer intact | AI referral, correctly attributed |
| Click from a mobile or desktop app that strips the referrer | Direct traffic |
| User copies a URL from the answer and pastes it | Direct traffic |
| Click from Google AI Overviews or AI Mode | Google organic search, with no separate label |
| User reads a brand name, then searches for it | Organic or direct, with no trace of the AI answer |
AI Overviews and AI Mode live on google.com, so in most analytics setups a click from them looks like any other organic click. Google Search Console also reports them together with regular results.
The last row is the hardest to capture. Many AI answers recommend a brand by name without the reader clicking anything. The person may visit days later by typing the name into a search box, and analytics credits that visit to search or to direct traffic.
How is AI referral traffic different from AI crawler traffic?
Both involve AI companies and both appear on a website, so the two are often confused, though they are separate things.
AI crawlers are automated programs that fetch pages for model training, for building a search index, or to read a page during a live answer. Because they do not run the JavaScript that most analytics tags depend on, they rarely appear in tools like Google Analytics. Server logs record them instead, identified by their user agent, the name a program sends with each request.
AI referral visits come from people. A visitor loads the page in a browser, runs the analytics tag, scrolls, clicks and sometimes converts. A crawler fetch can lead to a citation, and a citation can lead to a referral visit, but each is counted in a different place and answers a different question.
| AI crawler visits | AI referral visits | |
|---|---|---|
| Who visits | A bot | A person |
| Where it shows | Server logs, CDN or firewall logs | Web analytics |
| What it indicates | The page can be read by an AI product | The page was cited and someone clicked |
What does AI referral traffic mean for a website?
AI referral traffic is a downstream signal. A page has to be retrievable, then cited, then interesting enough to click after the answer has already summarized it. If the answer covers the question fully, a page can be cited often and still get few clicks. Pages with something the answer cannot provide, such as a tool, a full comparison, pricing details or a way to buy, tend to turn citations into visits more often.
Referral data misses most of the influence, so it works best alongside a direct measure of presence in answers. AI visibility tracking runs a fixed set of prompts across AI platforms and records which brands are mentioned and which pages are cited. Comparing the two shows whether a change in citations is followed by a change in visits. The Searcherries AI visibility tracker records prompts, competitors and cited sources on a schedule for this kind of comparison.
To get a usable AI referral number, a website typically creates a segment or channel for known AI domains and updates that domain list as new products appear. It looks at landing pages and conversions for the segment as well as sessions. It also treats a rise in direct and branded search traffic as a possible sign of AI answers that never produced a recorded click.
