“Great. So, what did it make us?”
Most AI visibility reports can’t answer this question, which is undoubtedly a phrase you’ve heard from a bottom-line-focused executive following a marketing report presentation.
If you’re a CMO or founder, you might be tempted to drop it into the “interesting, but what did we actually get from it?” bucket.
The reality is that AI traffic is hard to see if you don’t know how to use the AI traffic analytics that prove it, and this can lead to problems, including:
- No clear way to tell whether ChatGPT, Perplexity, Gemini, or Claude are actually sending visitors to your site
- AI clicks being buried as “direct traffic” in Google Analytics
- A visibility report for leadership that you can’t connect to leads or revenue.
All three are blind spots that could be costing your company customers.
Read on to find out how you can use AI traffic analytics to improve your marketing ROI.
We’ll look at how you can find AI referral traffic in GA4 in five minutes, tag your placements so you can ensure every citation is traceable, and connect those sessions to closed deals in your CRM.
With this knowledge, you’ll be able to answer that opening question with a business-building number.
We’ll cover:
- What is AI traffic?
- How to track AI referral traffic in Google Analytics with no developer
- How to tag your placements with UTMs to prove which one worked
- How to connect an AI citation to a closed Deal in your CRM
- Case study: Tracing Duckfund and Lula’s PR from AI answer to revenue
- How to use AI traffic analytics to prove what AI search is worth
Want to stay up to speed with the latest in AI traffic analytics? Sign up for our monthly newsletter and join our growing community of 2,000+ marketers and business leaders.
What is AI Traffic?
AI traffic is any visit to your website that an AI tool directed there.
When someone asks a large language model (LLM) like ChatGPT, Gemini, or Claude a question, the AI synthesizes an answer from a vast online dataset. If you’re one of the most eligible sources, it will mention you and link to your site. The searcher can then click on it and land on your page.
This is what we call AI referral traffic. They’re real people sent to you by an AI answer, much like how they always have done via Google or a newsletter link.
Why most AI traffic hides in “direct”
Google Analytics (GA) records these visits, but not always clearly. They might show up under the AI tool’s name, but many also get mixed up with unidentified or “direct” traffic, which GA can’t trace back to.
This happens partly because AI platforms don’t always pass referrer data (more on that below), and partly because of how people use AI assistants. Users read an answer on their phone, then open your site on a laptop later. Or they copy the URL out of the answer instead of clicking it. Either way, the referrer tag is gone by the time they arrive, and GA logs them as if they typed your address from memory.
“The honest truth is that most AI-sourced traffic still gets lumped into ‘direct’ because users copy a URL rather than click,” says Siim Kostabi, Co-Founder of Pageloot, a QR business code generator that regularly sources traffic from LLMs. “Your form data often tells you more than your analytics does.”
In other words, ChatGPT and other LLM traffic is often already on your site. You just can’t see it yet.
How to track AI referral traffic in Google Analytics with no developer
It only takes around five minutes in Google Analytics 4 (GA4) to see whether AI tools are sending you visitors, and you don’t need to consult a developer.
Here’s a simple step-by-step guide.
You can also save this data as a comparison or custom report in your GA4 library, so you can monitor ChatGPT or other LLMs traffic without a manual check each month. For a shareable dashboard, send the data to Data Studio (formerly Looker Studio).
Note: This data catches visitors who clicked directly through to your page, but not those who pasted your URL into a separate window. These still land in “direct” traffic, and this is why the next section matters.
How to tag your placements with UTMs to prove which one worked
Here is the next challenge: Even if you do the above set up, you may still miss out on recording clicks from AI.
AI traffic is hard to track in GA4 because LLMs don’t always pass referrer data and because AI apps hand links off to the browser in a way that can strip it. So many visits will still show up as direct – even if a user did click on your URL. Special tags can fill the gap, but only when they are present on the link.
Once you know which AI tools sent you visitors, this next stage, AI traffic analytics will tell you which placement earned them: UTM tags.
A UTM tag appears at the end of a URL as extra pieces of text that tell GA4 where the click came from. Every link from your domain should have a tag before it goes live, including guest posts, press releases, or a partner mention.
Your citations won’t all come from your own site. Some will come from a guest post, some from a press release, and some from a third-party review, which is exactly why each one needs its own tag.
Here’s the format you should use:
[page URL]?utm_source=[where the link sits]&utm_medium=[type of placement]&utm_campaign=[the specific push]
A Copy-Paste Example:
Here’s the URL broken down into its parts:
How UTMs work
UTMs on your own links often survive when an LLM passes a link on as-is, and GA4 can capture those parameters even when the AI tool sends no clear referrer.
Nothing guarantees it, though. ChatGPT added its own UTM parameters to some links in 2025, and one practitioner’s October 2026 data suggests that after the GPT-6 rollout it may now be dropping both the tag and the referrer, pushing those clicks to Direct. OpenAI hasn’t yet confirmed this.
GA4 admin access isn’t necessary here, but you do need to tag every outbound link before it’s published.
Connecting a session to a lead to a closed deal
Finding out which AI source sent you visitors is only half the battle: you also need to know whether that person converted.
To find this out, you’ll need to see the original lead record, which captures where that person arrived from, not just where they came from when they converted. This way, you’ll capture buyers who:
- Asked an LLM about your category, clicked through on the citation, and left without converting
- Then returned three weeks later by typing your name into Google and booking a demo.
Your CRM logs the lead as “Google search,” because that’s the last thing that happened. The AI answer that actually created the lead is invisible, which creates a blind spot for many companies.
“The step most businesses skip is preserving the original source in their CRM. Once it disappears at form submission, they cannot connect AI traffic to pipeline or revenue,” says Tamara Jovanovic, founder and senior SEO strategist at phraseit.agency.
Your web developer or CRM admin can set this first-touch capture up once. Most CRMs either handle it natively or need a small script added to the site. From then on, every lead arrives with its original source attached, as long as you store it server-side.
One caveat here: never rely on a browser cookie alone. Safari caps script-set cookies at seven days, so a buyer returning three weeks later (exactly the scenario above) can look like a brand new visitor. Write the source server-side, or into the CRM on the first form submission.
This data will help you fine-tune your marketing approach. When a contract closes months later, sales can open the record, see the first visit came from ChatGPT, and trace the deal back to the AI response that started it.
Why conversion rates from AI traffic matter more than volume
AI-driven traffic works differently from organic search. It arrives in smaller numbers, but an AI tool has already pre-qualified it, so we tend to see high conversion rates as a result.
As any leads-driven marketer would agree, a hundred sessions that close two contracts beat a thousand that yield nothing.
AI tool Ahrefs found this out firsthand when they discovered that 12.1% of sign-ups were coming from AI search, which accounted for just 0.5% of total traffic. AI visitors converted at a rate 23x higher than organic visitors.
The next section looks at how Mint Position applied its own methodology, Consensus Optimization, to earn AI search results for two of our customers.
Case study: Tracing Duckfund and Lula’s PR from AI answer to revenue
Sometimes, all it takes is one AI recommendation to make an AI traffic strategy worthwhile.
This is what happened to Duckfund, a leading US real estate financing platform, after partnering with Mint Position. In December 2025, they closed a $500,000 CRE financing deal that they could trace back to one ChatGPT answer – the largest deal to date at that time. It was a sign of more to come.
We achieved this by following a simple yet effective strategy:
- We studied which content formats LLMs were drawing from in the CRE financing space. These were mainly listicles, comparison pages, and community discussions.
- We created content to publish on all three fronts, including listicle content on their own domain, earned placements on a high-authority third-party sites, and a deliberate presence in CRE investor communities on Reddit.
The time from publication to first AI citation? Just under a month.
This was part of a wider improvement. Duckfund’s average ChatGPT visibility rose from 22% to 38% over the quarter to January 2026, while its average position improved from 2.7 to 1.6. In other words, it was within the top two answers whenever ChatGPT named a CRE financing provider.
That one deal wasn’t an outlier, either. AI search alone is up 840% year-over-year, and registrations on Duckfund’s platform from AI search tripled over the same period, outpacing organic search as an inbound channel for the first time.
Together, AI search and organic have grown from 8% to 17% of inbound calls since H2 2025, at a fraction of the cost of paid channels.
South African business funding platform Lula saw a similar effect at aggregate scale. They appear in 59.5% of LLM results in their category, ahead of much larger legacy banks Standard Bank (54.1%) and FNB (51.7%), with an average citation position of 3.2. They are consistently among the first brands named in AI answers, despite having a fraction of the big banks’ marketing budget.
That visibility converted. Over the same quarter, organic leads rose 26.2% while Lula’s total site-wide leads fell 11.8%, making organic’s share of all leads climb from 13.35% to 19.1%. This gained ground as other channels weakened. A single use-case page on small business funding in South Africa generated 96 signups in the quarter.
That visibility also extends into Google’s own AI. Over a trailing three months, Lula’s top 10 pages generated more than 413,000 impressions in AI Overviews and AI Mode, with a single article on business ideas in South Africa accounting for 149,297 of them.
Impressions are not sessions. Most of those readers saw Lula’s content inside an AI answer without clicking through, so they sit entirely outside GA4, on top of the traffic and leads Lula was already getting. Tracking AI traffic means counting both.
“In a zero-click channel, counting referrals will tell you AI isn’t working when the truth is you’re being read but not linked,” says John Talasi, Founder and CEO of Financer.com, a financial comparison platform operating in more than 20 markets.
This is why Search Console should be in your tracking setup alongside GA4. Without it, Lula’s half a million impressions would be invisible, and the picture would look smaller than it is.
These two case studies prove one important point: visitors arriving from an AI answer have probably already shortlisted you, and they land directly on your site to confirm their decision.
How to use AI traffic analytics to prove what AI search is worth
To find out the true value of AI search, you must connect visibility to revenue, and you’ll need to track three numbers each month to do this.
- The number of AI-sourced sessions from GA4 and Search Console
- Your citation share
- The leads and closed deals that came from an AI answer.
It’s only by viewing this together that you can see whether your customers came from an AI answer or not. If not, you’ll need to find out where the gap is: whether you’re not being cited at all, or being cited and not clicked.
As someone responsible for financial marketing, this is nothing new: you’ve always had to justify spend against closed business rather than traffic by measuring what matters, as we cover in our SEO for financial services guide.
AI search doesn’t change the standard, only the channels you’re measuring against it.
This is the thinking behind Consensus Optimization, Mint Position’s framework that unites the GEO vs SEO debate and influences how AI describes your brand.
It’s only by influencing the whole AI picture that you’ll get to the revenue that comes with it.
Want to get more customers from AI search? Book a no-cost consultation with the Mint Position team, and we’ll show you how our Consensus Optimization approach will get you ranked and cited.
FAQs
How do you track ChatGPT mentions of your brand?
ChatGPT source tracking tools that directly query AI chatbots are the best way to track mentions. Peec AI, Semrush’s AI toolkit, and Ahrefs Brand Radar all run your chosen prompts on a schedule and record whether your brand appears, and in what position, across ChatGPT, Claude AI, Gemini, and Perplexity.
Does GA4 track Google’s AI Overviews?
No. GA4’s AI Assistant channel covers LLMs including ChatGPT, Gemini, Microsoft Copilot, and DeepSeek, but Google excludes AI Overviews and AI Mode, which count as organic search.
Two further limits are worth knowing. The channel only launched on 13 May 2026 and isn’t retroactive, so any AI traffic before that date stays filed under Referral or Direct. Perplexity also isn’t named in Google’s definition, so those visits may sit under Referral.
For AI Overviews and AI Mode, Google’s own Search Console reports launched on 3 June 2026 for a subset of sites, and they show impressions rather than clicks, which is why Lula’s 413,000 would otherwise be invisible.
What’s the difference between AI visibility and AI-driven revenue?
Visibility metrics tell you how often AI assistants name your brand. Revenue tells you what those mentions produced. These two metrics might be connected, but they’re not interchangeable. A brand can hold a high citation share, for example, and earn nothing if no lead record captures where buyers came from. You should report both side by side.
Do AI crawlers show up in my traffic data?
No. Bots like GPTBot and ClaudeBot from OpenAI and Anthropic appear in server logs rather than as visitors in sessions. This is mainly because crawlers don’t run GA4’s JavaScript, so no session is recorded in the first place. GA4 also excludes bots on a known list, but a bot using a normal browser user agent can still get through.
