Why don't AI crawler visits show up in Google Analytics?

Google Analytics doesn't watch your server. It waits for a browser to load the page, run a small JavaScript snippet, and send a hit back. If nothing runs that snippet, nothing gets counted.

Most AI crawlers behave like a plain HTTP client. They request the URL, read the HTML that comes back, and leave. They don't build the page in a browser engine, so the analytics tag sits in the HTML as unexecuted text.

That means an AI bot report in your analytics will look empty even when GPTBot, OAI-SearchBot or ClaudeBot hit your site thousands of times. The visits happened. Your analytics tool was simply never in the loop.

The same mechanism has a second consequence worth knowing. If your prices, product names or key claims only appear after JavaScript runs, those crawlers can't see them either. Our guide to JavaScript rendering and AI crawlers covers that side of the problem.

Vercel's study of AI crawler traffic (published December 2024, updated February 2025) is the clearest primary source. It found that ChatGPT and Claude crawlers do fetch JavaScript files (11.50% and 23.84% of their requests respectively) but don't execute them. In Vercel's words, they "can't read client-side rendered content."

The 99Visibility evidence base records independent re-testing in June 2026 that found the same for GPTBot, OAI-SearchBot, ChatGPT-User, ClaudeBot, Claude-SearchBot, PerplexityBot, Meta-ExternalAgent and Bytespider. Two crawlers are the documented exceptions:

• Googlebot renders JavaScript, as Google's JavaScript SEO documentation describes.
• AppleBot renders through a browser-based crawler.

Carry two caveats. Vercel's sample came from its own traffic, which skews toward Next.js sites, and Vercel sells server-side rendering. The finding has replicated, but crawler behavior can change, so treat it as accurate as of October 2026 rather than permanent.

Analytics can't see crawlers, but it can see the people who arrive from AI answers. Those are humans in real browsers, so the tag fires normally.

Even that view is partial:

• OpenAI says ChatGPT adds utm_source=chatgpt.com to referral links, but free-tier and mobile app clicks often arrive untagged and land in direct traffic. See why ChatGPT traffic shows up as direct.
• Google AI Overviews emit no distinct referral, so those clicks blend into ordinary Google organic.
• User-triggered fetchers such as ChatGPT-User and Perplexity-User retrieve a page on behalf of a person, but the fetch itself is still a bot request. The human only appears in analytics if they click through.

So analytics answers one question: which humans came from AI surfaces that tag their links. For the full setup, read measuring AI referral traffic in analytics. For crawler activity, you need a different data source.

Your web server and CDN log every request, whether or not a script runs. That makes logs the ground truth for AI crawler activity.

1. Get the logs. Pull raw access logs from your host, or the request logs from your CDN. Many CDNs also offer a bot or AI crawler dashboard.
2. Filter by user agent. Search for the tokens each operator documents: GPTBot, OAI-SearchBot and ChatGPT-User (OpenAI), ClaudeBot and Claude-SearchBot (Anthropic), PerplexityBot and Perplexity-User (Perplexity), Meta-ExternalAgent and Meta-WebIndexer (Meta).
3. Verify the bot. User agents are easy to fake. Cloudflare's bot documentation describes verification by cryptographic signature, a published IP list with a stable user agent, or reverse DNS. OpenAI publishes its search crawler IP ranges at openai.com/searchbot.json, and it warns against relying on IP lists alone, so check user agent and IP together.
4. Read the status codes. A 200 means the crawler got the page. A 403, 429 or a challenge page means something turned it away. The playbook on reading server logs for AI crawler status codes walks through each case.

If you don't have log access, there's a free way to see which AI bots crawl your site.

Logs tell you about access, which is the stage you control. They can show that a retrieval bot never visits, or that your firewall returns 403 to OAI-SearchBot while robots.txt allows it. That second case is a silent failure: nothing in analytics or robots.txt will warn you. Since 15 September 2026, Cloudflare blocks Training and Agent crawlers by default on ad-displaying pages for new domains and existing free-tier customers, so a site can lose access without anyone editing a file.

If you want to test this directly, the AI Crawler Checker requests a page as each AI crawler and shows which ones your firewall turns away.

Logs can't tell you whether a model used or cited your page. Retrieval and citation happen inside the engine, and a visit buys eligibility, nothing more. Read why AI crawlers visit but never cite you before treating crawl volume as a visibility metric. Also separate training bots from search bots: seeing GPTBot but not OAI-SearchBot means different things for ChatGPT search.