If you’ve opened Google Analytics hoping to see how many AI agents — ChatGPT, Perplexity, Gemini — are hitting your store, you’ve probably found nothing. That’s not a setting you missed. GA is structurally blind to AI agents, for two reasons, and the fix lives somewhere else entirely.

Reason 1 — GA4 silently filters known bots

GA4 automatically excludes traffic from known bots and spiders, using the IAB/ABC International Spiders and Bots List. This happens by default, you can’t turn it off, and you can’t view what was removed — there is no “bot report.” (The old Universal Analytics at least had an Exclude all hits from known bots and spiders checkbox; UA was shut down in 2023, and GA4 didn’t replace it with anything visible.)

So even the agents GA could recognize are dropped before they reach any report.

Reason 2 — the bigger one: GA runs on JavaScript

This is the part most people miss. GA is a client-side tag (gtag.js): it only records a visit if the page’s JavaScript executes in the visitor’s browser. But most AI crawlers don’t run JavaScript — GPTBot, ClaudeBot and PerplexityBot fetch your raw HTML and move on. They never execute gtag.js, so they never fire a hit.

The consequence is blunt: even with no bot filtering at all, client-side analytics would still miss the agents you care about. You can’t measure, in a JS tag, traffic from clients that don’t run JS.

What actually sees agent traffic — go server-side

Agent visits are real HTTP requests, so they show up where requests are logged, before any JavaScript:

  • Server access logs. Filter by user-agent for GPTBot, ClaudeBot, PerplexityBot, Google-Extended, Meta-ExternalAgent, Applebot. This is the ground truth of who fetched what.
  • Your CDN / WAF dashboard. If you sit behind Cloudflare, its bot and AI-crawler analytics already break traffic down by AI company — no tag required.
  • Log-based analytics tools that ingest server or edge logs instead of a browser tag.

One caveat: user-agent strings can be spoofed, so for anything load-bearing, verify against the crawler’s published IP ranges (OpenAI, Anthropic and others publish theirs) rather than trusting the UA alone.

Why this matters for your store

Agent traffic is the leading indicator of whether ChatGPT, Perplexity and Gemini can even reach you — and, increasingly, whether they can buy from you. If you’re steering by GA, you’re flying blind to the exact audience that’s growing. Seeing which agents crawl you, and what they fetch, is the first step; making sure they can actually complete a purchase is the next one — which is what an AgentReady scan checks.

Sources

  • Google Analytics 4 Help — “[GA4] Known bots and spiders” (automatic bot exclusion, IAB/ABC International Spiders and Bots List).
  • Cloudflare — AI-crawler / bot analytics documentation.
  • Crawler operator docs — OpenAI (GPTBot), Anthropic (ClaudeBot), Perplexity (PerplexityBot), including published IP ranges for verification.

As of July 2026. Verify the specifics against each vendor’s current docs before acting.