Overview
On its most recent earnings call, Google’s CEO Sundar Pichai said search queries hit an all-time high, and pinned it directly on AI: “People love our AI experiences like AI Mode and AI Overviews, and they’re coming back to Search more.” AI Mode alone crossed one billion monthly users in its first year, and AI Overviews now reach 2.5 billion.
So here’s the question that should be nagging at you: if AI search is so popular, why is the AI traffic in your analytics not telling the same story? Why do the big industry reports benchmark AI at well under 1% of web traffic – Ahrefs put it at just 0.17% – when it’s supposedly the biggest shift in search since Google launched?
Because the numbers aren’t what they seem.
The channel sending you some of your highest-intent visitors is the one your data is least equipped to see. Until now. We’re going to help you find the real data behind your AI traffic.
We asked 171 marketers how they’re tracking AI traffic
Before we show how to track AI traffic with ease, let’s get into some research.
We asked 171 marketers, founders and SEO leads how they’re dealing with AI search traffic. Here are some of the standout results.
54% struggle to find their AI traffic
Without prompting, more than half described the same frustration: they know AI is sending visitors, they just can’t find them in the data. The visits scatter across Direct, Referral and “unassigned” buckets depending on the platform, the device and whether the referrer header survived the journey.
“A huge chunk of AI-referred traffic just shows up as ‘direct’ or gets lumped into regular organic. I’d estimate we’re probably missing 30-40% of actual AI-driven visits because of how broken the attribution is right now.”
44% said branded search lift is the signal they trust most
People rarely click the citation in an AI answer – they open a new tab and Google the brand name instead. That verification habit shows up in Search Console even when the recommendation never will, which is why nearly half our respondents named branded search as their most reliable AI signal.
“When AI tools recommend a business by name, users often follow up with a direct Google search. Watching that branded query trend tells me more about AI-driven awareness than any referral report does.”
Roughly one in three said AI visitors convert better
The pattern was remarkably consistent: smaller volume, sharper intent. By the time an AI platform recommends you, it’s already done the comparison shopping – so visitors arrive pre-qualified, asking better questions and moving faster.
If AI traffic were low volume and low value, you could safely ignore it. Low volume and high value is a different story – it means the visitors most likely to become customers are the ones your reporting is crediting to nothing.
“AI accounts for just 1.46% of my traffic, yet it’s my highest-quality converting channel.”
There’s no such thing as a “normal” AI traffic number
The self-reported numbers were all over the place. Visible AI traffic estimates range depending on industry, content type and how hard the team had worked to surface it. Travel, B2B and SaaS sites with strong informational content sat at the top end; publishing, healthcare and ecommerce towards the bottom.
Estimated AI traffic by industry
That spread tells you something important: there is no “normal”. Anyone quoting you a single universal benchmark is selling you a simplification – your honest comparison point isn’t an industry average, it’s your own trend line.
“Anyone quoting you a precise percentage is guessing with confidence.”
Why AI attribution is so broken
It’s safe to say that a lot of existing AI data is pretty broken – where are the problems stemming from?
Referrer stripping – the direct traffic problem
When most AI platforms send someone to your website, they strip the referrer header on the way out. GA4 sees a visitor with no origin and files them under Direct, exactly as if they’d typed your URL into the browser bar.
The scale of this is hard to overstate. Digital Bloom and Loamly data shows that 70.6% of AI traffic is invisible “dark traffic” – and the marketers we surveyed back it up with their own audits.
Dark traffic is any visit to your site that arrives without a trackable referral source, so your analytics can’t tell where it came from. It typically gets bucketed as “direct” traffic or missed entirely.
I’ve watched this happen in real time – check a session in GA4, see “Direct,” then pull the server logs for the same timestamp a nd find a chatgpt.com referrer sitting right there, stripped out before it ever reached the dashboard.
“60-70% of AI-driven visits to my B2B clients landed as Direct because referrer data gets stripped.”
Want to estimate what this is hiding on your own site? Run the numbers:
AI traffic and conversions calculator
Enter your numbers to see the potential impact of AI traffic.
hidden from your attribution
every month
It's a rough estimate, not gospel - but even at half that rate, most businesses are sitting on a meaningful blind spot.
Brand search lift - the attribution theft problem
Here's the journey your analytics can't see…
Someone asks ChatGPT for a recommendation. ChatGPT mentions you. The user doesn't click the citation - they open a new tab and Google your brand name instead. GA4 logs a branded organic visit, your SEO report gets recognition, and the AI recommendation that started everything gets no credit.
AI does the selling - “organic” takes the credit
The same applies to those who jump between apps or even between AI platforms. While an exact global head count does not exist, recent market research confirms that the vast majority of active AI users rely on multiple AI tools, averaging 3 tools per person.
“We saw a retail client's branded queries jump 31% in two months with no extra ad spend, purely from showing up more consistently in AI-generated answers.”
I've tracked this pattern on a handful of client accounts now - an AI referral bump precedes a branded search bump almost every time, usually within a week or two.
And it has a cost attached. If those verification searches come through branded paid search, you're literally paying to capture a customer AI already won for you. Worse, if a competitor is bidding on your brand name - and they almost certainly are - AI sends someone your way, and a rival intercepts them at the last second.
Benchmark reports - the data pool problem
If the data on your own site is murky, the industry benchmarks are arguably worse - because they look authoritative while measuring completely different things.
Ahrefs AI traffic study: 0.17%
The Ahrefs AI traffic study is large (3,000 sites) and well run, but it can only count traffic where the referral source is trackable - by their own admission, the visible minimum.
Their headline 0.17% average is also from early 2025, which in AI terms, is ancient history.
SE Ranking’s analysis of AI: 0.32%
SE Ranking's AI traffic research is built on a much larger base - 101,574 websites across 250 countries and territories. But it has the same issue. It only counts AI visits that arrive with a trackable referrer.
ChatGPT vs Google dashboard: 0.28%
The ChatGPT vs Google live dashboard currently shows AI assistants at 0.39% of traffic across 101,000+ sites - but those are just sites that have opted into Ahrefs Web Analytics.
Search Engine Land: 1.08%
Search Engine Land draws on a large dataset - 13,770 domains and more than 3.3 billion sessions across 10 industries between May and September 2025 - so it's a solid benchmark, but it still only measures trackable AI referrals.
None of this makes the reports useless; it just means you should take them with a pinch of salt. Your industry, your content mix and your audience's AI habits will swamp any global average.
Google Analytics 4 - the complicated problem
GA4 can be configured to separate AI referrals, but it’s far from simple:
- Go to Admin > Data display > Channel groups.
- Click "create new channel group".
- Add a new channel and name it AI referrals.
- Click "add condition group"
- Add a new condition > source > matched regex > copy the regex below > apply
- Reorder > Drag your new AI referrals channel to the top.
- Revisit the regex quarterly - new platforms launch all the time.
I've set this up more times than I can count at this point, and the regex list is the part that bites people - miss one platform and your "AI referrals" channel quietly underreports from day one.
Doable? Yes. Fiddly? Also yes. Even a perfectly configured channel group only captures traffic where the referrer header survives the journey. The 70%+ that arrives without one will still land in Direct.
GA4 alone cannot give you the full picture. Which is exactly why the marketers we surveyed have stopped relying on any single number.
“We stopped relying only on referral source attribution and started correlating spikes in branded queries, assisted conversions, and sudden engagement increases on pages frequently cited in AI summaries.”
The fix? Three metrics that actually tell the truth
There are three metrics that work hand in hand to give an accurate portrayal of AI traffic. Conveniently, they're the same three signals our respondents kept circling back to.
Metric 1: AI referrals
This is the traffic you can directly attribute to AI platforms via referral headers in GA4 - the channel group you just built above.
Trending up means AI platforms are citing you more and sending more people. Just remember what you're looking at: roughly the visible 30% of the story, not the whole of it.
ChatGPT still leads - but its grip is loosening fast
Metric 2: Brand search volume
These are your branded impressions and clicks in Google Search Console, watched as a trend.
AI sends awareness more often than it sends clicks. People increasingly use AI tools to do the research - to shortlist, compare and decide - then switch to Google, Safari or Bing to actually navigate to the brand they've settled on. That hand-off is invisible in your referral reports, but it's measurable in your branded search trend.
Something to keep in mind
Branded search rises as a business grows anyway. New customers, PR, a strong quarter, a social post that lands - all of it nudges the number up, and none of it is AI. So a gently rising line on its own proves nothing.
The trick is correlation. Don't read branded search in isolation - line it up against your AI referral traffic. If you start ranking highly in AI answers one week, you'll often see both move together: referrals spike, and branded impressions and clicks spike alongside them.
That combination - two independent signals lifting at the same time, with no campaign to explain it - is far harder to dismiss than either number alone. Watch for it as a sustained pattern over 60-90 days rather than a one-off blip.
Metric 3: Dark traffic trends
Strip the homepage out and look at direct traffic to interior pages - blog posts, service pages, product pages - and the picture cleans up dramatically. It’s rare for someone to type /blog/ai-traffic-attribution-guide from memory, unless it’s cached in your browser.
Someone landing directly on a deep page almost certainly clicked something - and if there isn’t already an obvious referral, it likely came from AI.
Tracked as a trend over 60-90 days, non-homepage direct traffic is where dark AI traffic leaves a detectable footprint.
Just like brand search, dark traffic is at its most convincing when you line it up against the other two metrics mentioned.
Why track all three metrics together?
Each metric alone is incomplete. AI referrals miss 70% of sessions. Brand search is noisy. Direct traffic catches everything and explains nothing.
But when all three correlate together and show a different story from normal trend lines, AI search becomes much more reliable for tracking.
The solution? Our free AI traffic dashboard
Everything above is doable manually. Channel groups, regex maintenance, Search Console exports, homepage-excluded segments, cross-referencing the lot - it works, and now you know how.
It's also several hours of setup plus ongoing upkeep, which is precisely why most teams never do it. So, we built a free DataStudio dashboard that does it for you!
What you’ll find on the AI dashboard
The dashboard focuses on the important metrics from this article, pre-built in one view:
- AI referral traffic: GA4 visits from AI platforms, with the custom channel work already done for you.
- Brand search trends: Branded clicks and impressions from Search Console, so you can spot the AI lifts that never show up as referrals.
- Dark traffic: Homepage-excluded direct traffic, tracked as a trend - a pinch of salt as a number, gospel as a direction.
The dashboard sits in DataStudio on top of data you already own. No new subscriptions, no code, no agency retainer - and a view your boss can read in 30 seconds rather than a GA4 maze that takes 20 minutes to interpret.
How to set it up
- Grab the dashboard using the download banner at the end of this guide.
- Open it in DataStudio.
- Select your GA4 property from the data dropdown at the top of the report.
- Connect your Google Search Console profile the same way.
- Enter your brand terms in the Brand search trends section - then you're live.
What you'll see
Once connected, you get the full three-metric trend view:
- Which AI platforms send you traffic, and what those visitors do.
- Whether your brand searches are climbing independently of campaigns.
- Whether your interior-page direct traffic is trending up.
New update! As of 3rd June 2026, Search Console has dedicated generative AI performance reports - isolating, for the first time, how often your pages appear inside AI Overviews and AI Mode. You can break it down by page, country, device and date, so you can finally see which of your URLs are surfacing in Google's AI answers. It's impressions only for now - no click or query data yet, so you get visibility without its traffic value.
It’s time to stop ignoring AI search traffic
Every time I've built one of these dashboards for a client, the same thing happens - the moment all three metrics line up, the debate about whether "AI traffic is real" for that business just ends. The data does the talking.
If you only remember five things from this guide:
- Standard GA4 undercounts AI traffic badly. Roughly 70% of it lands in Direct with no flag - your dashboard is showing you the floor, not the ceiling.
- Treat industry benchmarks with scepticism. Most have selection bias, and many conflate three different metrics under one name. Context, not targets.
- Three metrics beat one. AI referrals, brand search volume and homepage-excluded direct traffic each cover the others' blind spots.
- This is worth fixing now. Across multiple studies and our own research, AI traffic converts at multiples of standard organic. Small volume, outsized value.
- You don't have to build it yourself. Our free dashboard puts all three metrics in one Looker Studio view, on top of the GA4 and Search Console data you already have.
Download our AI traffic dashboard
Get more reliable insights into your AI data.
Download here
