AI Referrals vs AI Crawlers
Two kinds of AI traffic reach a website, and they answer different business questions. One is a person. The other is software. Add them together and you get a number that describes neither, so here is how to keep them apart and what each one is good for.
The difference at a glance
| AI referral | AI crawler | |
|---|---|---|
| Who it is | A person who clicked a link inside an AI assistant | Software that fetches your page for an AI company |
| What it tells you | An assistant pointed someone to you, and that person acted on it | An AI company is reading your content |
| Typical signal | A referrer such as chatgpt.com, an AI app's built-in browser, or an AI tag in the link | The crawler's own name in its user agent, or a known crawler network address |
| Can it become a customer? | Yes. It can call, book, buy or fill out a form | No. It never buys anything |
| Where it appears in AppWT Analytics | AI referrals, by assistant | AI crawlers, by crawler, apart from people |
Why they must stay separate
Picture a month with 40 AI referrals and 400 AI crawler hits, and 4 sales from the AI referrals. Add the two and you report 440 "AI visits" with a sales rate under 1%, which looks like failure. Keep them apart and you see 4 sales from 40 people, a 10% rate, plus a separate sign that AI companies are reading your pages.
The mixed number hides the result you care about and blurs the signal you were trying to measure. AppWT Analytics records the two as different source types for exactly this reason.
Three kinds of AI crawler
- Training crawlers collect pages to build or update a model, for example GPTBot, ClaudeBot and CCBot.
- Search and answer crawlers index pages so an assistant can cite them in answers, for example OAI-SearchBot and PerplexityBot.
- Fetchers triggered by a user read a page because a person asked the assistant a question at that moment, for example ChatGPT-User. A person is behind the request, but the visit itself is software. The blog post on user-triggered AI fetchers covers this third bucket in depth.
Reading the dashboard
In the Traffic Sources report, AI referrals and AI crawlers are separate source types, each broken down by assistant or crawler. The AI Sources card at the top of the dashboard adds the two together today, so when you want people only, read the AI Referrals number instead of that card.
What each number can prove
- AI referrals prove that people arrived from an assistant. They do not prove how often the assistant mentioned you without a click.
- AI crawler hits prove that a crawler fetched a page and ran the tracking script. They do not prove that the page will be cited, and they undercount crawlers that never run JavaScript.
- Neither counts a link someone copied out of an AI answer and pasted into a browser. That visit carries no signal and lands in direct traffic.
Go deeper
- AI bot hits vs AI referred visitors: how to separate them
- Measuring the ratio of AI crawler hits to AI referrals
- How AppWT detects AI traffic, with every check and its limits
- How to track ChatGPT traffic to your website
Common questions
Is an AI crawler visit a sign that ChatGPT will recommend my business?
Not on its own. A crawler visit shows that an AI company fetched your page. It does not show that the page will be quoted, cited or recommended to anyone.
Should AI crawler hits count toward my visitor total?
No. Crawlers are software, so counting them as visitors inflates traffic and lowers every conversion rate you calculate. Keep them in their own report.
Why does my server log show far more AI bots than my analytics?
A tracking script only records software that runs it. Many crawlers download a page without running JavaScript, so the server log, which records every request, always shows more of them.
See both numbers for your site
30-day free trial, card required. Then $9.97 a month or $77 a year, per site. Unlimited pageviews. Cancel anytime.
Get Started Free