Semrush says published AI traffic benchmarks disagree, measure your own site

Most marketing teams still quote a single AI traffic benchmark as if it applied to their own site. Semrush published a piece this week that lists four of the studies in use today. The conversion numbers in them disagree too much for any reader to apply them. The author is Sergei Bezdorozhev. His answer is to stop picking a number from the press. Measure your own site instead.

The studies Semrush names:

• Siege Media's 2026 analysis of first-party GA4 data reports a median AI-to-organic conversion rate of 1.26 times. The window was January to May. AI traffic matched or beat organic on 72 percent of the sites in the sample.
• Shopify's Q2 2026 figures say AI-referred shoppers are about 80 percent more likely to convert than organic visitors. Watches convert at 2.4 times organic. Apparel sits at a 60 percent advantage.
• Adobe's retail data puts AI referrals 60 percent ahead of other visits, with travel 1 percent behind. A year earlier Adobe's non-AI retail visits were worth 128 percent more than AI ones. Over 12 months the figure flipped.
• Semrush's own Impact of AI Search study values AI visitors at 4.4 times an organic visitor.

The numbers do not line up because the sites do not

Semrush says the studies differ because they measure different audiences, conversion events, baselines and time periods. The piece quotes a Marketing Science study of 973 ecommerce sites. In the first cut, ChatGPT referrals converted better than paid social but worse than other established channels. In a follow-up, the gap between ChatGPT and organic search was no longer statistically meaningful.

The piece also quotes Microsoft Clarity data on 30 billion sessions across 2,000 projects that get the most LLM traffic. AI sessions had 21.6 percent fewer quick bounces than organic search. They lasted longer and covered fewer pages. Semrush's own channel mix figures say AI traffic was 0.14 percent of visits in 2025. It grew 66 percent that year.

What Semrush asks the reader to do

The piece is a four-step plan, run on the reader's own analytics. First, mark AI referrals as their own channel in GA4. Use GA4's AI Assistant channel group and add custom rules for the AI domains that are missing. Second, compare session key event rate and revenue per session against organic search, paid search and direct. Third, audit the landing pages the AI picks against the pages organic traffic lands on. Fourth, tie the visibility figure, which pages the AI quotes, back to post-click revenue. Semrush recommends a 90-day window for sample size.

The piece ends on a case study of one small ecommerce site. ChatGPT drove 471 sessions and 12 purchases in a month. That is a 2.5 percent conversion rate and about $2,000 in revenue. Per visit, that was better than the site's Google organic traffic.

The gap between the 1.26 and the 4.4 numbers is the point. A budget decision about AI visibility has to rest on your own 90 days of data. The published benchmarks do not agree on what AI traffic is worth.

Source

AI traffic quality: look past the industry benchmarks and measure ROI, by Sergei Bezdorozhev at Semrush.