To track AI traffic in GA4 properly you need two numbers on the same slide: the share of sessions, which is around 1% for most B2B sites, and the conversion rate, which is higher for B2B and statistically fragile everywhere else. GA4 got a native AI Assistant channel on 13 May 2026, and it fixes less than the coverage suggests: it doesn’t backfill, it may not catch Perplexity, and somewhere between a third and two thirds of AI sessions arrive with no referrer at all and land in Direct regardless. Here’s the setup that closes most of the gap, the conversion premium, and how to present a 1% channel to leadership without overselling or burying it.
Where AI traffic goes in GA4
Before you can track AI traffic in GA4 you need to know which of four buckets it’s sitting in. Google confirmed on 13 May 2026 that a native “AI Assistant” channel was added to the Default Channel Group, requiring no configuration, with the medium set to ai-assistant automatically when the referrer matches a recognised assistant (TechWyse). Before that, AI sessions went to Referral or, without a referrer header, Direct.
| Where it lands | What triggers it | What it misses |
|---|---|---|
| AI Assistant (native, since 13 May 2026) | Referrer matches Google’s recognised list. | Not retroactive, so your history stays buried. Sources Google hasn’t listed. |
| Referral | AI referrer not on Google’s list. | Coverage of Perplexity is disputed between sources, which is itself the problem. |
| Direct | No referrer header sent at all. | The largest leak by volume. |
| Organic Search | Clicks from Google AI Overviews and AI Mode. | Indistinguishable from a standard blue-link click. |
The channel is forward-only, so every AI visit before 13 May stays filed as Referral or Direct and the trend that got you here isn’t visible in it (AIVO). On Perplexity the sources contradict each other outright, and when the trade press can’t agree on what a channel contains you don’t put it in a board deck without checking your own property first.
The referrer-less leak is the big one, with estimates putting it at 35 to 70% of AI referral sessions arriving with no referrer header and defaulting to Direct regardless of your setup (Authority Tech).
How to track AI traffic in GA4 with a custom channel group
Ten minutes in Admin, still worth doing now the native channel exists, because a custom group applies to historical sessions in reporting and covers whatever Google hasn’t listed.
- Admin → Data display → Channel groups → Create new channel group. Put the year in the name, because you’ll be editing the regex.
- Add a channel called AI Search, condition Source matches regex.
- Paste a hostname-anchored pattern. An unanchored
perplexitywill match any domain containing the string.
^(chatgpt\.com|chat\.openai\.com|claude\.ai|perplexity\.ai|www\.perplexity\.ai|gemini\.google\.com|copilot\.microsoft\.com|deepseek\.com|grok\.com|meta\.ai|you\.com|poe\.com)$
- Drag the AI channel above Referral. GA4 evaluates rules top-down, and this is the single most common reason a freshly built group sits empty while the traffic keeps showing up under Referral (Terminus).
- Diary a quarterly review. New assistants appear and existing ones change how they pass referrers, so a regex written in August is stale by spring.
What this doesn’t catch, and you should write it down next to the report so nobody discovers it in a meeting: referrer-less sessions, Google AI Overview clicks, and any assistant released since you last edited the pattern. The native channel misses those too.
The conversion premium, and why the numbers everyone quotes are shaky
The standard version of this article stops here. Ahrefs found AI search visitors converting at 23x the rate of traditional organic on their own site, with 0.5% of visitors driving 12.1% of signups (Ahrefs), and Semrush’s June 2025 study put AI visitors at 4.4x the value of organic ones across 500+ digital marketing and SEO topics.
Neither is what it’s being used as: Ahrefs is one site selling SEO tools to people who ask AI assistants which SEO tool to buy, close to the best possible case, and Semrush analysed a topic set inside its own category.
The study that ran the actual test reached a different conclusion: Amsive analysed six months of GA4 data across 54 websites with validated macro conversions and, in their words:
A paired t-test confirmed that the difference was not statistically significant (p = 0.794). These results show that LLM traffic did not convert significantly differently from Organic.
Organic came in at 4.6% and LLM referrals at 4.87%, with 56% of sites seeing LLM traffic outperform their average and 41% seeing it underperform (Amsive, Stan Ventures summary). Search Engine Land’s coverage also noted LLM traffic made up under 1% of sessions against roughly 32% from organic (Search Engine Land).
| Source | Sample | Finding | The caveat |
|---|---|---|---|
| Ahrefs, 2025 | Own site, 30 days. | 0.5% of visitors, 12.1% of signups (23x). | Single vendor site, ideal category fit. |
| Semrush, June 2025 | 500+ digital marketing topics. | AI visitors worth 4.4x organic. | Own category, vendor-published. |
| Amsive, 2025 | 54 sites, 6 months, paired t-test. | No significant difference overall (p = 0.794). | The only one that tested significance. |
| Amsive, B2B subset | Within the same 54. | 2.17% LLM vs 1.16% organic. | Smaller subset, but the segment that matters here. |
The B2B row is the one to hold onto: the aggregate premium dissolves under testing while the B2B slice still shows LLM traffic converting at nearly double organic.
The number your own site can’t tell you yet
At 1% of sessions on a site doing 10K a month, that’s 100 AI sessions, and at a 3% conversion rate you’d expect three conversions, with one or six next month. Neither number tells you anything about the channel.
Rough rule: a segment needs around 100 conversions before its conversion rate stops swinging on noise, and about 30 before it’s worth looking at. Most B2B marketing sites hit that somewhere between next year and never.
Reporting a 400% conversion premium off six conversions is the analytics equivalent of a rounding error wearing a suit.
The fix is free: report the raw conversion count alongside the rate every time and let the denominator do the arguing. It’s the same discipline I applied to pipeline attribution in the 48% marketing-sourced pipeline number and how it was produced.
How I’d build the leadership slide
One slide, four lines, in this order:
- Line one: the share, stated as a floor. “AI referrals were 1.1% of sessions in July, and that’s a starting point, because AI sessions without a referrer header land in Direct.”
- Line two: the absolute numbers. Sessions and conversions as counts. “112 sessions, 4 conversions.” Leadership reads counts faster than rates and can’t be misled by them.
- Line three: the comparison, hedged correctly. “Converting above site average on a sample too small to test, and the only multi-site study with a significance test found no aggregate difference, though its B2B subset showed nearly double.” That sentence separates a content lead who read a headline from one who read the study.
- Line four: the decision. What this changes, or explicitly doesn’t. Usually, keep publishing citable material, don’t move budget on a 1% channel, revisit at 30 conversions.
What I wouldn’t do is lead with the multiplier. When I ran 17+ KPIs per quarter across content, social and organic at a B2B EdTech SaaS, the metrics that survived contact with leadership were the ones I’d stress-tested myself first, and a number that collapses the moment someone asks about sample size costs you the rest of the deck. More on which numbers earn their place in the metrics that tell you if your content strategy is working.
When the numbers are too small to be relevant yet
The reason to track AI traffic in GA4 right now is to establish a baseline. Citations show up long before referrals do, so a monthly prompt log across ChatGPT, Claude, Perplexity and Gemini gives you a trend while GA4 is still on single digits (how I track whether AI is citing my content).
Search Console works the same way, since impressions move before clicks do (impressions up, clicks down). Pair citation presence with the GA4 channel and you have a story with a direction long before you have one with statistical power. The content work doesn’t change either way: same substance-first approach as the AEO/LLMO content optimisation framework, and the same reason AEO won’t save a weak content strategy.
The limits
SparkToro’s June 2026 zero-click study notes in passing that AI tools send less than 1% of all traffic out, and the Datos and SparkToro Q1 2026 report put AI tools below 2% of desktop visits across the US, EU and UK. A research-heavy vertical will sit higher than a B2B SaaS site does, so check your own property before quoting anyone’s percentage, including mine.
I also can’t tell you what share of your Direct traffic is really AI, and neither can anyone else. The 35 to 70% range is wide enough that treating its midpoint as an estimate would be false precision.
The significance argument cuts both ways: Amsive’s study is the most rigorous available and it’s still 54 sites over six months in a channel that grew fast during and after the measurement window, so a finding of no significant difference in 2025 doesn’t establish there’s no difference now. What it establishes is that the 23x headline was never supported by the evidence people cite it with, which is why the slide needs the share and the conversion rate side by side instead of whichever one flatters the argument you walked in with.
Frequently asked questions
Two layers. GA4 has had a native AI Assistant channel since 13 May 2026 that needs no setup and classifies sessions whose referrer matches Google’s recognised list. On top of that, build a custom channel group under Admin → Data display → Channel groups with a hostname-anchored regex on Source covering ChatGPT, Claude, Perplexity, Gemini, Copilot and the smaller assistants, then drag it above the Referral channel so GA4 evaluates it first. The custom group covers sources Google hasn’t listed and applies to historical sessions in reporting, which the native channel doesn’t.
Because a large share of AI sessions arrive with no referrer header, and anything without a referrer defaults to Direct no matter how your channel groups are configured. Estimates put that at 35% to 70% of AI referral sessions. Clicks from Google’s AI Overviews are a separate leak, counted as Organic Search and not currently separable from ordinary search clicks, which is one reason the AEO and LLMO distinction matters for measurement rather than only for strategy. Either way, any AI traffic figure you report is a floor rather than a measurement.
For B2B, probably better than organic, though not by the multiples in circulation. The widely quoted 23x came from Ahrefs’ own site and the 4.4x from a Semrush study of its own category, both vendor-published and single-context. The one multi-site study that applied a significance test, across 54 sites over six months, found no statistically significant aggregate difference (p = 0.794), with 56% of sites better and 41% worse. Its B2B subset showed LLM traffic converting at 2.17% against 1.16% for organic, which is the honest claim to make.
Report the share and the conversion rate together, with the raw conversion count next to the rate so the sample size is visible, and state the share as a floor. If the segment has fewer than about 30 conversions, give the count and skip the rate, because a percentage off single digits swings wildly month to month and gets challenged the first time it moves. The point of setting up a way to track AI traffic in GA4 at this volume is the baseline rather than the decision. Attach a decision anyway, even if it’s to change nothing and revisit at a defined threshold.
