This piece is part of Mode’s research desk: we trace the statistics this industry repeats back to their primary sources and publish what we find, including when the trail is unflattering to our own trade. Method at the bottom.
The stat you will be quoted
If you talk to anyone selling AI search services this year, you will hear a version of this sentence: “AI visitors convert 4.4 times higher than normal search visitors.” It appears in sales decks, agency pages and social posts, usually with no source attached. We traced it.
Where it actually comes from
The primary source is a genuine study by Semrush, the SEO software company, which analysed traffic from AI search tools and estimated the value of the average LLM visitor at 4.4 times the average organic search visitor. The number is real and the study is serious. What gets left out when the number travels:
1. The dataset skew. The study leaned heavily on marketing and SEO topic websites, the single industry with the highest AI adoption in the world. Applying its multiplier to a joinery firm or a cafe is like measuring umbrella sales in Manchester and forecasting for Marrakech.
2. The real range. Across the studies we collected, the figure runs from 2x to 27x depending on the industry and how “conversion” is defined. A 973-site e-commerce study found ordinary Google traffic actually beat ChatGPT referrals by around 13% for completed purchases. Both things are true: AI visitors are often more engaged, and there is no universal multiplier.
3. The inflation is compounding. The number has already escalated in the retelling: “AI search visitors convert 23x higher” is now a live headline in the industry. That is how statistics launder: a narrow, honest finding gets stripped of its context, rounded up, and repeated until it becomes furniture.
The three questions that deflate any inflated stat
You do not need to be a researcher to defend yourself. Ask the seller: What dataset is that from? Which industry did it measure? How was “convert” defined? A serious provider answers all three without blinking, because they know their sources. A decorator changes the subject.
Apply the same test to us. Mode publishes original research with the method on the page precisely so you can check the numbers we quote, and this desk exists because we would rather trace a statistic than repeat it.
Method, so you can check it
We followed the 4.4x claim from its common uses back to the primary source (Semrush’s AI search traffic study), read the study’s own description of its dataset, and collected the surrounding evidence: industry coverage of the figure, the range across comparable studies (2x to 27x), the 973-site e-commerce comparison in which organic search outperformed ChatGPT referrals for transactions, and the escalated variants now circulating. All figures are reported as published by their sources; where sources conflict we state the range rather than picking a side. This page is date-stamped July 2026 and will be updated if the underlying evidence changes. Corrections welcome at hello@modemarketing.co.uk.
Where Mode fits
Mode sells AI visibility work with published prices and without borrowed multipliers: we measure what AI systems currently say about your business, fix what is measurable, and report movement honestly. If a number on our pages ever fails the three-question test, tell us and we will trace it in public.
