Why GA4 Can't Show You AI Citations
- GA4 has no category for AI crawlers, so their visits land in noise or get lumped in with bots you can't act on.
- GA4 relies on cookies and client-side scripts, which most AI crawlers ignore or block.
- GA4 can't tell a live ChatGPT fetch from an indexing bot, so you never see which queries drive AI referrals.
- You're managing discovery with one hand tied behind your back if you can't see AI activity at all.
- Kymo separates human visitors, AI answers, indexing crawlers, and training crawlers into one dashboard, so you can actually see what changed.
GA4 can't show you AI citations because it never classifies AI traffic in the first place. It was built for a web where Google sent you humans, and it does that job fine. But search is no longer just Google.
The rise of AI answers means a growing share of your discovery happens when a chatbot reads your site and decides whether to mention you. That activity is invisible in GA4, and you cannot improve what you do not measure.
GA4's blind spot is structural, not a settings issue
GA4 does not have a dimension for "AI crawler". It has a "Bot" category, but that's a bucket designed for filtering junk, not for understanding opportunity. When AI services crawl your site, their user agents appear as undefined or as generic bots, and GA4 gives you no way to understand what they are doing there.
This matters because not all crawlers are the same. A live fetch from PerplexityBot happens because one person asked a question and Perplexity checked your page for an answer. That is a potential customer. GPTBot (OpenAI's training crawler) is reading your content to train a model; it may shape thousands of future answers. Google-Extended (a robots.txt control token, not a crawler) controls whether your content influences Gemini answers.
Those are three different jobs with three different outcomes. GA4 treats them all as noise.
The measurement gap: cookies, scripts, and invisible requests
GA4 tracks human visitors by loading a JavaScript snippet that drops a cookie or uses localStorage. That works for people with browsers. AI crawlers do not run JavaScript. They fetch your HTML, read it, and leave. No cookie is set, no pageview event fires, no session is recorded.
The visitor identifier in GA4 is a client-side construct. It needs a browser to execute code. A crawler is a server-side HTTP client. It makes a request, receives a response, and moves on. Nothing in that exchange touches GA4's tracking.
This is also why Kymo is a server-side receiver. Your pages contain a small script (or you send events server-side), but the site's visitor counter, and especially the AI crawler classification, happens by reading the request directly. It keeps no raw IP, it keeps no human user-agent, and it hashes identifiers per day. But a crawler's user agent is recorded so you can audit the classification.
The result is that GA4 undercounts human visitors who block scripts and misses every AI crawler entirely. You are flying blind.
What a real answer looks like: classification, prompts, and click-throughs
When an AI assistant fetches your site, there are three things you need to know:
- What kind of request it was. Live answer for one person, search indexing, or bulk training.
- Which pages were fetched. That tells you what content AI engines value.
- Whether the fetch led anywhere. Did someone click through from an AI answer to your site?
GA4 gives you none of that. Kymo gives you all of it. The AI Visibility report shows which pages AI assistants fetch, the likely prompts behind those fetches, and confirmed click-throughs from AI referrals. This is measurement, not magic. It is the difference between guessing and knowing.
Consider the comparison in practical terms:
| What you need to know | GA4 | Kymo |
|---|---|---|
| How many human visitors? | Approximate, script-dependent | Counted server-side, no cookies |
| Which AI engines crawled you? | No data | Classified into answer, index, training |
| Which pages did AI fetch? | No data | Page-level report |
| What prompts drove the fetch? | No data | Likely prompts shown |
| Did AI referrals send visitors? | Aggregated as organic or direct | Confirmed click-throughs from AI referrals |
| Does AI tracking count against limits? | N/A | No, it is separate from your event volume |
Kymo is built so you can see the whole board. One dashboard shows human visitors, their referrers, their paths, and their bounce rates, then flips to show you exactly what ChatGPT, Claude, and Perplexity are reading.
Why this is a business problem, not a tech problem
You don't need to care about crawler anatomy. You need to care about revenue. The AI search for small business category is growing because real customers ask real questions in chatbots. When a customer asks ChatGPT which plumber, which lawyer, which bakery is worth visiting, the answer depends on which site got fetched, and what that site said.
If your site is being cited, you want to know so you can create more of what works. If your site is being ignored, you want to know that too. Either way, the first step is the same: measure it.
We have written before about what an AI visibility checker is and whether you need one. The short answer is that you need a way to see this traffic. The post on AI SEO tools for small business goes deeper into what is worth paying for.
The crawler directory and the free check
If you want to see the full list of which engines are tracking you, the AI crawler directory documents them all. It is a public reference. You can look up any agent, see what it does, and decide how to treat it.
You can also run the free AI visibility check on your own site. It reads a URL from public signals only, so your site does not need Kymo installed. The report arrives by a magic link over email. It tells you whether AI engines can even see you in the first place.
That free check is a useful starting point. It is like a health screening. For continuous observation, you need something installed. The Kymo documentation walks through the setup, and it takes minutes, not days.
Bounce rate, budget, and the real cost of bad data
There is a secondary cost to GA4's blind spot. Because it miscategorizes AI referrals, your human analytics are skewed too. A visitor who clicks through from an AI answer might be labeled as "direct" traffic. You lose the referral credit, and you lose the ability to track what that visit is worth.
Bounce rate is a good example. A visitor from a chatbot who reads your pricing page and leaves might look like a bounced direct visitor. In reality, that is a qualified lead. Understanding what normal bounce looks like for your type of site matters, as we covered in Bounce Rate for Small Business Sites: What's Normal, but you cannot even assess it if the traffic source is mislabeled.
You also need to think about the cost of tools. GA4 is free, but the hidden cost is the time you spend creating convoluted reports to extract data it was never designed to hold. Platforms that were built for AI visibility are often narrow, tracking chatbots but ignoring your human traffic. You end up paying for two tools to see half the picture.
Kymo is priced simply. Solo is $9 a month or $90 a year, for up to 10,000 events a month. Studio is $29 a month or $290 a year, for up to 100,000 events a month. Every feature is included on both plans. The plans differ only by event volume. Both start with 14 days free, no card required. AI crawler tracking does not count against your event limit. If you exceed your limit, the dashboard pauses and your data is not deleted.
That is the entire pricing story. No tiers of features, no "contact sales" for the basic features you need.
What to do about it now
The broader context is that AI answers are rerouting discovery away from traditional search result pages. You can read The Small Business Guide to AEO for a full strategy, but the measurement piece comes first. You need to know where you show up today before you can improve it tomorrow.
GA4 cannot show you that. It was built for a different generation of search, and it is still good at that job. But it is not good at this one.
The fix is straightforward
Run the free check on your own URL first. You will get a report by email that shows what AI engines can see. Then decide whether you need the continuous view.
You have two choices. You can keep your GA4 setup for human analytics and accept the blind spot. Or you can add a tool that measures both. The latter is the only way to see the full picture.
Start tracking your AI visibility
Run the free AI visibility check on your site to see what AI engines already know about you, or start tracking it continuously with a dashboard that shows both your human visitors and every AI crawler alongside them. The free check reads a URL from public signals, so your site does not need Kymo installed, and the report arrives by email.