Here is a number that should make every digital agency sit up straight. AI-referred visitors sign up at eleven times the rate of Google search traffic. And most analytics tools cannot see a single one of them.
That finding comes from a new analysis published July 13, 2026, by Rankability, an SEO (Search Engine Optimization) and AI visibility software company based in St. Louis, Missouri. The company pulled together data from Microsoft Clarity’s 2025 study of more than 1,200 publisher sites and Adobe Digital Insights’ 2025 report on generative AI referral traffic. The takeaway is short and a little uncomfortable. Your best-performing traffic source is probably sitting in a report bucket where you will never spot it.
AI Traffic Beats Every Other Channel, and It Is Not Close
The gap here is not a rounding error. Microsoft Clarity’s 2025 look at 1,200-plus publisher and news sites found that LLM-referred visitors (people arriving from large language models like ChatGPT or Perplexity) converted to sign-ups at 1.66%. Traditional search? 0.15%. That is the eleven-to-one difference in one line.
The other channels do not fare much better against AI. Social media traffic converted at 0.46%, a 72% drop from the AI rate. Traditional search sat at 0.15%, a 91% drop. Direct traffic came in dead last at 0.13%, trailing AI referrals by 92%. Line them up and AI referral is the yardstick everything else gets measured against.
One thing worth naming plainly. AI referrals still make up less than 1% of total sessions across the sites studied. The volume is tiny. The quality is not. Microsoft Clarity also recorded 155.6% growth in AI-driven platform traffic over eight months. So this is a small channel getting bigger fast, and its quality signal is already loud enough to read.
The Attribution Gap That Keeps Agencies in the Dark
Here is the mechanical reason agencies miss it. When someone clicks a link that ChatGPT or Perplexity surfaced, the referrer string handed to the destination site is often blank, broken, or unrecognized. Google Analytics and most tag-based tools have a simple rule for referrers they do not recognize. They call it “direct.”
So a click from an AI engine lands in the exact same bucket as a person who typed the URL by hand. In the default report, the two look identical. There is no flag, no label, nothing that says “this one came from AI.”
The fallout is real, and it is quiet. A client’s monthly report might show organic search going flat or sliding down. Meanwhile a newer, higher-converting source is stacking up inside the direct channel, unseen. Budget decisions, content plans, channel bets — all of it gets made on data that leaves out the channel with the best return. Adobe Digital Insights found AI-referred visitors spend 41% more time on site and bounce 23% less than other traffic. Those are engagement numbers that would jump off a dashboard, if anyone were reporting them in the right column.
Retail Data Points to the Same Story
This is not a publisher-only quirk. Adobe Analytics’ 2025 research found that traffic to U.S. retail sites from generative AI sources grew 1,200% between July 2024 and February 2025. That is the steepest climb of any acquisition channel Adobe’s analysts tracked over that seven-month stretch.
Retail AI visitors followed the same pattern as the publisher data: higher intent, stronger engagement, more downstream action than the site average. Two very different kinds of sites, chasing two very different conversion goals, both pointing the same direction. When a pattern holds across setups that have almost nothing in common, it usually reflects something real about the visitor, not a fluke of one category.
The reason makes sense once you say it out loud. A visitor who arrives from an LLM recommendation has already gotten a curated answer to a specific question. The decision is half made before the click ever happens. That is a warmer visitor than a random search result can deliver.
Looking ahead, Gartner’s 2024 scenario model projects that traditional search volume could fall by as much as 25% by 2026 as AI chatbots and virtual agents take on more queries. Read that number next to the growth figures above, and the direction of travel is hard to argue with.
What Agencies Can Do Right Now
The good news is that this is a measurement problem. Measurement problems get solved.
Pull AI Traffic Out of the Direct Bucket
Step one is separating AI referrals from direct traffic. That takes one of two things. Either you enforce UTM parameters at the source, which is not always possible with AI-generated links, or you use an AI visibility tool built to read referrer strings from known LLM domains and file those sessions as their own segment. Skip that split, and the conversion advantage from the Clarity and Adobe data gets blended into an average where it vanishes.
Once the traffic is isolated, judge it the same way you judge everything else. Cost to earn the visibility, sessions attributed, conversion rate, and revenue or lead value per session. On a per-session basis, the Clarity data suggests AI visitors are already paying their way, if only the sessions were being counted.
Rethink What Content Earns the Citation
There is a content lesson buried in here too. LLMs tend to recommend content that answers a specific question with precision and authority. Sites that show up in AI answers usually share a few traits: structured, well-sourced, and deep on their topic. That is a different emphasis from classic search optimization, and agencies that instrument AI traffic now will start to see which of their pages actually earn the citations. That feedback loop does not exist if the traffic stays buried in the direct channel.
My Take on Where This Is Headed
I will add my own read after years of watching new channels arrive. The quality signal in this data is convincing. The volume is not there yet, and that is exactly why so few teams are paying attention. That is also the mistake. Every channel that ended up mattering — organic search, paid social, mobile — looked small and weird right before it did not. The agencies that start counting AI traffic today are the ones who will have a year of data when their clients finally ask the question. The ones who wait will be guessing.
Common Questions About AI Traffic
What makes AI traffic hard to track?
LLMs do not reliably pass a recognized referrer string when someone follows a link from a chatbot answer. Standard tools, Google Analytics included, dump unrecognized referrers into “direct.” Without a tool that reads known LLM referrer domains, the AI channel stays hidden in default reporting.
Can agencies still measure ROI from AI visibility work?
Yes, but the AI sessions have to be separated from direct first. After that, they can be scored on conversion rate, session value, and revenue like any other channel. The 1.66% sign-up rate from Clarity gives you a benchmark for what the channel produces when it is measured right.
If AI traffic is under 1% of sessions, why care now?
Because of what those visitors do once they land. Eleven times the conversion rate of search, 41% more time on site, 23% lower bounce. A channel with numbers like that earns attention at any volume, and Adobe’s 1,200% growth figure says the volume is coming.
Why do AI visibility tools matter for agencies?
The reporting built for old channels does not read AI referrals on its own. Agencies need tools that flag AI sessions, attribute them, and tie them to conversions. Without that layer, they cannot show the value of content earning AI citations, and clients cannot see a channel that may already be their best.
How do agencies split AI traffic from normal direct traffic?
They use analytics configurations or purpose-built tools that recognize referrer strings from platforms like ChatGPT and Perplexity, then route those hits into a dedicated segment. Where the referrer is missing, UTM tags on linked content give a partial read when the source is known.
About Rankability and the Data Behind This
Rankability is a St. Louis, Missouri SEO and AI visibility software company founded in 2024, built to help digital agencies measure and improve how their clients show up in both search results and AI-generated answers. For this analysis, the company synthesized public findings from three named sources: Microsoft Clarity’s 2025 study of AI-driven traffic across 1,200-plus publisher and news sites, Adobe Digital Insights’ 2025 generative AI referral report on U.S. retail sites from July 2024 through February 2025, and Gartner’s February 2024 search-volume forecast. No proprietary survey was run. The Gartner 25% figure is a scenario model, reported as such, consistent with the firm’s own note.
The Bottom Line for Agencies
The story from Rankability’s analysis holds together across every dataset it touches. AI-referred visitors convert at 1.66%, eleven times the 0.15% from search. They stay longer and bounce less. The channel is growing at triple- and quadruple-digit rates from a small base. And nearly all of that performance is being filed under “direct,” where no agency will ever act on it. The quality is proven. The volume is climbing. The only thing missing is the measurement.
Set up the tracking now, and the payoff is a year of clean data when clients start asking where their AI traffic comes from and what it is worth. Leave it alone, and the answer stays stuck in a bucket labeled with the wrong name, doing its best work in the dark.