AI assistants still send a sliver of Google's traffic

Search Engine Roundtable grouped the three studies on 22 October under a headline suggesting AI visibility and AI search are not worth investing in. Haynes and The SEO Works read Google Analytics 4 data from about 100 sites each.

Haynes, who published on 21 October, looked at more than 145 million sessions on 100 randomly sampled sites of varying sizes. With every AI assistant counted, the share rose only to 0.73% of Google organic sessions. Of the 55 sites that tracked conversions, ChatGPT visitors converted at 13.8% against 3.02% for Google organic, yet Google organic still drove 27 times more conversions than all the AI assistants together. She calls her dataset small, and her advice is that Google Search remains by far the most likely way people will find you.

The SEO Works, an agency that sells the services its study discusses, counted 2,825,594 organic sessions and 72,095 AI referral sessions on 95 sites from September 2025 to August 2026. On the 63 sites with enough traffic to compare conversions, AI visitors converted at 7.18% against 6.08% for organic, 1.18 times the rate, but less well than paid search visitors, at 0.67 times. For every 100 conversions from organic search, AI referrals delivered 3.2. The gap is narrowing slowly: AI traffic was 2.0% to 2.8% of organic until April 2026 and 3.86% in August.

Kevin Indig, writing in Growth Memo on 19 October, looked for AI influence hiding in Direct traffic on 12 sites, most of them B2B and aimed at the US, and in branded search for four brands. He could not detect it, and could not establish a reliable link between a brand's AI visibility and its later branded search or Direct traffic. "An effect could still exist," he wrote. "I just couldn't isolate it."

Some AI traffic is counted as something else. The SEO Works left out Google AI Overviews because their clicks arrive as ordinary organic sessions, so it calls its AI figure a floor, and it notes that in-app browsers, stripped referrers and people who search for a brand by name after reading an AI answer are probably missed too. For a week this month ChatGPT's own links lost their tracking tag and its visits were filed as Direct, which we covered in ChatGPT traffic showed up as Direct for a week.

The bigger gap is the answer itself. A buyer who asks ChatGPT for a shortlist reads the names before deciding whether to click anything, and an analytics tool only sees the click. Our UK AI Shortlist Study asked UK buying questions on 6 and 7 September 2026 and compared each answer with Google's first organic result for the same question. ChatGPT did not name the business ranked first on Google in 1,183 of 1,632 eligible answers. Google AI Overviews, whose clicks show up in analytics as organic search, did not name it in 36.7% of the Overviews we scored (539 of 1,470).

So a healthy organic traffic line can sit alongside AI answers that never name you, and a small AI referral count can hide answers that name a rival instead.

Do not take budget away from search engine optimisation on the strength of AI traffic. Haynes and The SEO Works both found that organic search still drives most visits and most conversions, and The SEO Works found AI referrals converting worse than organic for its group of B2B product sites, though there were only 10 of them.

Make sure your analytics records conversions at all. Haynes found that 45 of the 100 sites she sampled were not tracking any key events, which makes any comparison between channels guesswork.

Then measure AI answers as answers. Ask the questions your buyers ask and record whether each answer names you, which rivals it names and which pages it cites. To start, check if your business is on ChatGPT: our free checker puts 3 buyer questions about your category to ChatGPT in a logged-out session and shows whether each answer names you. Our ChatGPT visibility tracker page explains how we measure it over time.

The figures on AI answers come from the UK AI Shortlist Study, which put 1,000 UK buying questions about software, services and products to six consumer AI services on 6 and 7 September 2026 and compared each answer with Google's first organic result for the same question. A comparison counted only when Google's first result was classified as a business's own website, and Google AI Overviews were scored only where an Overview appeared. The counts above are from the study's published CSV, which is free to download from the study page. The traffic figures are the three studies' own and we have not re-run them.