What analytics tools show traffic coming from ChatGPT and Perplexity? What we use
What analytics tools show traffic coming from ChatGPT and Perplexity? How we combine GA4 with Promptwatch visitor analytics and crawler logs for client reporting.
When clients ask which analytics tools show traffic from ChatGPT and Perplexity, we start by asking which traffic they mean. There are three different things hiding under that phrase, and a report that mixes them up will mislead.
We run client programs on Promptwatch, and we pair it with GA4 for the client's own reporting. Here is how the pieces fit.
Three kinds of "AI traffic"
Referral visits. A person clicks a link inside an AI answer and arrives on the site. This is what most clients mean.
Crawler hits. A bot fetches a page. GPTBot, OAI-SearchBot, ChatGPT-User, ClaudeBot and PerplexityBot are examples. These are not visitors, and web analytics usually does not count them. They show up in server or CDN logs.
Mentions with no click. The AI names the brand in an answer and nobody follows the link. Analytics cannot see it at all.
A client who only sees the first kind will undercount their AI footprint. A client who confuses the second with the first will overcount it. So we report all three, labeled separately.
Step one: GA4, because it is free
We set up a custom channel group in GA4 that collects the known AI referrers, so ChatGPT, Perplexity and others appear as their own line instead of being scattered across referral sources. It takes an hour, costs nothing, and gives the client a number in a tool they already use.
The caveats are real, and we tell clients up front. Whether a referrer is passed depends on the platform and the way the link was opened, so some AI visits land in direct traffic. The referrer list needs maintenance as new engines appear. And GA4 has no idea about prompts, citations or crawlers. It shows that a session happened, not what produced it.
Step two: dedicated AI traffic attribution
For a fuller view we use Promptwatch's visitor analytics. It counts AI-referred traffic and conversions from ChatGPT, Perplexity, Gemini, Claude, Copilot, DeepSeek, Kimi, Mistral and others, through a lightweight script, a GTM template or the WordPress plugin. The setup is small, and the data is labeled by engine.
What makes it more useful than a channel group is the context. The same workspace holds the prompt panel and the citation reports, so when AI-referred visits rise on a page, we can check whether that page's citations also rose. We stop guessing and start correlating, though we are careful to say correlation is what it is.
Plan note: Essential ($95 a month) includes 200K visitor events, Professional ($245) includes 1M, and agency plans start at 500K on Kick-off ($199).
Step three: crawler logs
Visits tell you people arrived. Crawler logs tell you whether the engines can read the pages in the first place. Agent Analytics logs the AI bots in real time with IP verification, so spoofed user agents are filtered out, and it shows errors with failure explanations. It adds a crawl-to-citation path with a citation rate per page.
Crawler logs need a log source: Cloudflare (Logpush or an auto-deployed Worker), CloudFront, Fastly, Vercel, Netlify, Akamai, Google Cloud CDN, custom HTTP or the WordPress plugin. They start on Professional, so they are not on Essential. We scope this with the client's developer early, since it is the longest step in setup.
360 Insights then joins crawler logs, citation clicks and AI-referred visits for each page. For client calls, this is the view that links the three kinds of traffic.
What other tools do
We try not to generalize about other products, but some patterns are worth knowing. Some trackers include a simple AI traffic or referral module, usually on mid-tier plans. Some are built mostly around crawler and agent analytics, with a small prompt feature. Some SEO suites report AI mentions and cited sources but not visits to your site. Cloudflare reports crawler activity on every plan, and referral data requires a paid plan. When a client already uses one of these, we ask what it counts before we rely on it.
How we report it
The monthly summary has three lines.
- AI-referred visits and conversions, by engine, with a note on how small the sample is.
- Crawler activity and errors fixed.
- Visibility and citations for the prompts that matter.
We add a standing sentence: referrer data is incomplete, so these numbers are a floor, and visibility often runs ahead of clicks.
Case studies, used carefully
Promptwatch publishes customer case studies, and we cite them with care. TreeHouse, which runs Pararius and Huurwoningen.nl, reports that AI search became about 25% of organic conversions, 99% of it from ChatGPT, after crawler access was fixed. Crisp reports that AI-referred traffic converted at twice the rate of traditional channels. These are their results, and we never present them as a forecast for a client.
Where to start
Set up the GA4 channel group this week. Then add a dedicated tool when you want to see engines, conversions and crawlers in one place. Check crawler access before you invest in new content, because a page the bots cannot read will not earn visits.