Anyone who promises you direct revenue from AI can’t make that true. And that is okay, as long as you measure the right things. AI visibility does not yet translate neatly into euros in your dashboard, and that feels uncomfortable for anyone used to the straight line from click to conversion. But that line is fading, and that does not mean AI delivers nothing. It means you have to measure differently.
Traffic becomes a poor KPI
First the bad news for anyone steering on traffic. Research by Pew Research (2025) shows that on a results page with an AI summary people click through to a search result only 8% of the time, versus 15% without. A click on a link within the summary itself: 1%. More and more answers are given directly, without anyone visiting your site. Steering on organic traffic alone then means using a shrinking yardstick for a growing channel. We wrote earlier about what an impression in AI Overviews really means.
AI visibility works with a delay
Here is the heart of it, and it is good news. AI visibility does pay off, just not at the moment and through the channel you expect. Clickstream research by Similarweb (2026, US desktop users in finance, travel and beauty) shows that people who get a brand recommended in ChatGPT visit that brand’s site 2.5 times more often on average within seven days than a direct competitor’s. But note where that traffic comes in: about 56% of it arrives through a search engine (versus around 40% for regular visits), according to Similarweb mostly as a search for the brand name, and not as a visible AI referral. Someone sees your brand in ChatGPT, remembers it, and later simply searches your name. In your analytics that looks like branded search or direct traffic. The AI touchpoint gets no credit, even though it was the starting point.
To stay honest: Rand Fishkin (SparkToro), who contributed to the research, points out that it looked at big brands you’ve probably heard of, such as Sephora, American Express and Capital One. He calls it an open question whether the effect also holds for smaller, unknown brands. Think of it as a flywheel you first have to get spinning.
Don’t judge AI on the click it never sends, but on whether your brand sticks.
Giacomo Perticara, founder and SEO strategist at GRP Digital
What you can measure
The platforms don’t give you revenue figures, but they do give useful signals. Google’s Search Console has shown, since 2026, impressions from AI features at page level (without clicks or queries). Bing Webmaster Tools shows you, in its AI Performance report (still in preview), your citations, grounding queries and, since June 2026, your citation share: the share of all citations for a grounding query that goes to your site. That last one is exactly what you want to track, because visibility relative to your competitors is the unit that counts. And your server logs tell you something too: Cloudflare measured in 2025 that Anthropic’s crawler fetched tens of thousands of pages for every visitor Claude sent back, OpenAI’s around a thousand and Perplexity’s a few hundred at most, versus a handful for Google. Being crawled by AI crawlers therefore brings relatively few direct visitors. The measurable unit is visibility and share, not direct revenue.
Being mentioned is not being recommended
Another nuance that’s easy to miss in a tracker dashboard. Research by Semrush and Kevin Indig (2026) found that 61.7% of AI citations do not lead to a visible brand mention. Your source is used, but your brand name is not dropped. And when you are mentioned, context counts: a premium brand called “expensive” is no problem, that fits the positioning. A tracker that only records yes or no misses that entirely, and a sentiment score doesn’t know what fits your positioning either. So measure not just whether you are mentioned, but how.
Chris Green, senior SEO consultant, put it in the panel by Aleyda Solis this way: “relative visibility is the best we have.”
How to set the expectation
Selling AI internally or to a client? Keep it simple and honest. Steer on relative visibility and your share against the competition, on the trend in your brand demand, and on several channels together instead of only the organic bucket. And at the end of the customer journey just ask: how did you find us? That one tick makes the AI influence visible your analytics hides away. At GRP Digital we promise no direct revenue from AI, because nobody can make that stick. We do promise that your visibility and brand demand move in the right direction, and that is go on to what feeds the revenue.
Frequently asked questions
So does AI visibility actually make money? Yes, but with a delay and indirectly. People see your brand in an AI answer and still reach your site later, through a search or a direct visit. The value is real, only the attribution is tricky.
Why do I see so little AI traffic in Google Analytics? Because a large part comes in through search or as direct traffic instead of as an AI referral. So you are measuring the effect in the wrong place.
What is the best KPI for AI visibility? Relative visibility and citation share against your competitors, plus the trend in your brand demand. Not direct revenue from the AI channel alone.
Sources: Pew Research (click behaviour with AI summaries), Similarweb and SparkToro / Rand Fishkin (downstream effect of AI mentions), Google Search Console and Bing Webmaster Tools (what is measurable), Cloudflare (crawl-to-click gap) and Semrush / Kevin Indig (ghost citations).
