Most tree surgeons can tell you roughly how many calls they got last month, but almost none can tell you that two of those quotes started with someone asking ChatGPT “who’s a reliable tree surgeon near me?”. Measuring AI search traffic closes that gap. It is the practice of spotting the visitors who reached your site from an AI answer — Google’s AI Overviews, ChatGPT, Gemini or Perplexity — and following them through to a quote request or a phone call. This guide is the analytics companion to our overview of GEO and AI SEO for tree surgeons: that page covers getting cited, this one covers proving it pays.
What counts as “AI search traffic”?
AI search traffic is the set of real human visitors who land on your website after clicking a link inside an AI assistant’s answer. Someone asks an assistant a question — “how much does it cost to remove a large oak in Guildford?” — the assistant cites or links a source, and the reader clicks through to your page.
It is worth separating two things that often get muddled:
- AI referral traffic — a measurable click from an AI tool’s domain (such as
chatgpt.comorperplexity.ai) to your site. This is what GA4 can actually record. - AI influence — the wider effect of being mentioned in an answer, which may not produce a tracked click at all but still drives a later branded search or a direct visit.
You can measure the first reliably. The second you can only infer. Being honest about that line is what separates a trustworthy report from a wishful one.
Does GA4 track AI traffic on its own now?
Partly — and this changed recently. On 13 May 2026, Google added a native AI Assistant channel to GA4’s default channel group. When an incoming session has a referrer matching a recognised AI domain, GA4 now tags it with the medium ai-assistant, assigns the reserved campaign label (ai-assistant), and files it under the AI Assistant channel with no setup required.
The catch is in the recognised list. At launch Google’s documentation covered ChatGPT, Gemini, Copilot, Deepseek and Grok — but Perplexity, one of the highest-intent AI sources for local research, is not included and still lands in your Referral channel. The native channel is also not retroactive, so it only reflects sessions since it went live.
So the answer is: GA4 gives you a free head start, but not the full story. The fix is to run the native channel alongside a custom one that catches everything Google misses.
How do I set up complete AI tracking in GA4?
The reliable approach is a custom channel group that you order above Referral, so high-value sources like Perplexity get caught instead of being swallowed. The full step-by-step is in the how-to box on this page; here is the logic behind it.
GA4 evaluates channel rules top to bottom. If “Referral” sits above your AI rule, any AI source that passes a normal referrer (Perplexity, Claude and others) matches Referral first and never reaches your AI channel. Placing the AI rule above Referral fixes the ordering. Your AI channel uses a single case-insensitive regex on Source. A practical pattern for mid-2026 is:
chatgpt\.com|chat\.openai\.com|perplexity\.ai|gemini\.google\.com|claude\.ai|copilot\.microsoft\.com|deepseek\.com|grok\.com|you\.com|meta\.aiThis routes every major assistant into one row. Crucially, this only affects new sessions — GA4 does not reclassify historical data — so the sooner it is in place, the cleaner your trend line.
| Source | How GA4 sees it by default | What you need to do |
|---|---|---|
| ChatGPT | AI Assistant channel (native) | Nothing — but verify the utm_source=chatgpt.com tag is landing |
| Gemini, Copilot, Deepseek, Grok | AI Assistant channel (native) | Nothing |
| Perplexity | Referral | Add to custom AI channel, ordered above Referral |
| Claude | Referral / AI Assistant (varies) | Add to custom AI channel to be safe |
| In-app mobile clicks | Often Direct (no referrer) | Cannot fully recover — monitor supporting signals |
Why is so much AI traffic hiding in “Direct”?
This is the most important caveat, and the one cheap “AI traffic” dashboards quietly ignore. Many assistants — especially inside their phone apps — strip or never send a referrer header when a user clicks out. With no source to read, GA4 defaults that session to Direct.
Measurements through early 2026 put the share of AI-referred sessions that arrive without referrer data anywhere from roughly a third to the majority, depending on device and platform. The practical consequence: your reported AI figure is a confident floor, not the ceiling. A tree surgeon seeing forty AI-channel sessions a month is almost certainly being influenced by more.
Rather than pretend the gap away, measure around it. Watch for branded searches for your business name rising in Google Search Console after you start getting cited, and note unexplained climbs in Direct traffic that line up with AI mentions. These corroborating signals let you estimate the influence your channel report cannot see.
Which numbers actually matter for a tree surgeon?
Traffic counts are vanity. A homeowner reading your storm-damage page does nothing for your diary; a homeowner requesting a quote does. The metrics worth reporting are the ones tied to revenue:
- Quote-form submissions segmented by AI channel.
- Phone clicks (
tel:taps) segmented by AI channel — vital, because most tree-work enquiries come by phone. - Enquiries by job type where you can capture it — emergency call-out, removal, crown reduction, stump grinding — since their value differs hugely.
- Cost per lead by channel, once you compare AI against your paid and organic sources.
To get there you mark your form submission and phone clicks as key events in GA4, then build a Free-form exploration with your AI channels as rows and those key events as values. The result is a single table that answers the only question a busy owner cares about: did AI search send me work this month? That answer-first, lead-focused thinking is the same principle behind content that earns AI citations — there is no point being the answer in an AI Overview if you never find out it sent you a job.
Don’t forget the phone: call tracking
Web forms are only half the picture for a trade. A panicked homeowner whose tree just came down in a storm will tap the number, not fill in a form. If you only measure form fills, every phone enquiry that started from an AI answer disappears.
Connecting a call-tracking number or a robust tel: click event into GA4 lets a call show up as a conversion against the AI channel that produced it. This is exactly the kind of end-to-end measurement that underpins effective tree surgeon marketing: every lead, whatever channel it came from, attributed back to its source instead of guessed at.
What does this look like in practice?
This is the part of our work we are proudest of, because it is genuinely uncommon among tree-surgery marketers. Our background is in data and analytics — GA4, Google Ads and proper reporting — so we instrument first and decorate second. When we rebuilt the Jax Tree Removal website, the priority was not just a smarter-looking site; it was lead tracking that tells the owner which enquiries came from which clicks, so marketing spend stops being a leap of faith. The same setup that surfaces AI-sourced leads also surfaces paid and organic ones, in one honest view.
We do not promise inflated AI numbers, because the measurement gaps above are real. What we promise is a clear, conservative picture: the AI traffic we can prove, attributed to actual quotes and calls, alongside the supporting signals for the influence we cannot.
A simple checklist to get started
Run your own GA4 property against this list before your next reporting cycle:
- AI Assistant channel visible under Session default channel group (native, since 13 May 2026).
- Custom AI channel created, with Perplexity and Claude included.
- Custom AI channel ordered above Referral.
- Quote form marked as a key event.
- Phone (
tel:) clicks tracked and marked as a key event. - Call tracking connected for off-site phone enquiries.
- A saved exploration showing AI channels against leads.
- A monthly note to refresh the regex as new assistants appear.
If that list feels like one job too many on top of running the saws and the van, that is exactly the work we take off trade owners’ hands. Request a free website and analytics audit and we will show you, in your own GA4, where your AI search traffic is going, what it is converting, and where the leaks are.
Sources
- Semrush: GA4 adds AI Assistant channel for referral tracking — on the native channel launched 13 May 2026 and the assistants it recognises.
- Google Analytics Help: Default channel group definitions — the official reference for how GA4 channels are assigned.