1001 SEO Media
All posts
By 1001 SEO Mediaai referral trafficchatgpt trafficanalytics

Which tools measure traffic coming from ChatGPT and Perplexity?

How we measure traffic from ChatGPT and Perplexity for clients: what GA4 can and cannot show, and why we pair referral data with crawler logs and conversions.

Almost every client report we build now has a slide on traffic from ChatGPT and Perplexity. The question is which tool produces a number you can defend. Here is what we have learned about it.

Start by separating three numbers

When a client says "AI traffic", they might mean one of three things, and each needs a different instrument.

  1. Human visits that arrive from an AI answer. A person clicked a citation and landed on the site.
  2. AI crawler activity. Bots such as GPTBot, OAI-SearchBot, ChatGPT-User and PerplexityBot requesting pages. This is not human traffic, but it tells you whether you can be cited at all.
  3. Conversions from those visits. What the human did next.

If you blend 1 and 2 you will overstate the channel. If you stop at 1 you will not know why it is small. We report all three, separately.

What GA4 can do

Google Analytics 4 is where most teams start, and it is a fair starting point. You can build a custom channel group or exploration that matches AI referrers by source and compare sessions and conversions against other channels. For a site with modest traffic, that is often enough for a first look.

The limits are practical. Someone has to maintain the source list as engines change. Some AI visits may arrive without a referrer and land in direct. And it only sees clicks. It cannot tell you which prompts you appear in, which pages are cited, or whether a crawler is blocked. When a client's AI line is flat, GA4 cannot say whether the problem is visibility, access or content.

Why we add a visibility platform

The platform we run client programs on is Promptwatch, and the reason is that its visitor analytics sits next to the cause of the traffic. It counts AI-referred visits and conversions from ChatGPT, Perplexity, Gemini, Claude, Copilot, DeepSeek, Kimi, Mistral and others, using a lightweight script, a GTM template or the WordPress plugin. That covers number 1 and number 3.

For number 2, Agent Analytics shows real-time, IP-verified AI crawler logs, so spoofed user agents are filtered out. We can see which pages each bot fetched, errors with explanations, and a citation rate per page. Then 360 Insights joins crawler logs, citation clicks and AI-referred visits per page, which is the view we open when a client asks "why is this page not getting AI traffic?" The answer is usually in one of three places: the bot never fetched it, the engine fetched it but did not cite it, or it was cited but the click went nowhere.

Practical notes for scoping. Visitor events are 200K a month on Essential, 1M on Professional and 10M on Business. Crawler logs start on Professional. Crawler logs can come from Cloudflare, AWS CloudFront, Fastly, Vercel, Netlify, Akamai, Google Cloud CDN, custom HTTP or the WordPress plugin, so check which route your stack allows before you commit.

When a cheaper or narrower tool is fine

We do not pretend one tool fits everyone.

  • A small ecommerce store that only needs revenue by source can get by with an attribution tool that joins sessions to payments.
  • A site on Cloudflare can look at the AI crawler dashboard that comes with it for a quick view of which bots are fetching pages.
  • A market-level traffic estimator is useful for competitor context, but treat estimates for small sites as directional, not exact.

What none of those do is run the prompts, so none can tell you why you are or are not cited.

What the numbers will and will not tell you

Be careful about how you present this to a client.

  • AI-referred traffic is only the clicks. Plenty of buyers read an answer and never click. A flat line does not mean you are absent. Pair it with prompt tracking to see mentions that did not produce visits.
  • The share varies a lot by site. One published case study, from Promptwatch's customer TreeHouse (Pararius, Huurwoningen.nl), reported AI search at about 25% of organic conversions, almost all from ChatGPT, after fixing crawler access. That is one customer's result. Do not promise it to another site.
  • Conversion rate is the number to watch. Crisp's published case reported AI-referred traffic converting at twice the rate of traditional channels. Again, that is their result.

A simple setup we recommend

  1. Install the visitor script, GTM template or plugin so AI visits and conversions are tracked from day one.
  2. Connect crawler logs through your CDN.
  3. Track a panel of buyer prompts so you see mentions that never become clicks.
  4. Review the per-page join monthly and fix the weakest step.

If you want to set this up without stitching three tools together, Promptwatch is where we would begin. The free Explore tier is ChatGPT-only and will not show crawler logs or the full traffic view, but it is enough to see what a prompt panel looks like before you pay.

Related articles