AI search traffic grew 16 times from 2024 to 2026, according to a SE Ranking study of 101,574 websites, with ChatGPT commanding 74.78 percent of AI referral traffic. Claude surged 320 percent year over year, Gemini traffic grew 231 percent year over year, and Gemini overtook Perplexity as the number two AI search engine in January 2026, with 11.56 percent against 7.23 percent, according to the November 2026 statistics compilation.
ChatGPT dominance of referral traffic has survived two years of intense competition, which tells SEOs where the referral volume actually is. Optimising for AI referral traffic without a ChatGPT strategy is optimising for the minority of the channel, however interesting the smaller assistants may be as products.
Below the top, the movement is fast. Claude 320 percent growth and Gemini 231 percent growth show a channel still redistributing share, and Gemini passing Perplexity in January 2026 confirms that the number two spot is contestable. A strategy built only on today second-ranked assistant risks being outdated within a quarter.
Google AI Mode hit 75 million daily users in early 2026, according to the compilation, while ChatGPT reached about 1.2 billion monthly active users by June 2026. Google AI Overviews covered about 48 percent of United States queries in February 2026, according to BrightEdge, which puts the Google AI surfaces in a different league from standalone assistants on query volume, even where referral shares differ.
The distinction between users, queries, and referrals matters. A surface can have enormous query volume and still send few referrals, as the Reddit case shows for AI Mode, while a smaller surface can send a larger share of referrals if its answers link out more readily. Share of referrals, share of queries, and share of users are three different leaderboards.
First, cover ChatGPT retrieval: content that ChatGPT search surfaces and cites, with clear facts and source pages that survive the assistant summarisation. Second, track Gemini and Claude growth separately, since their growth rates suggest they will matter more next quarter than last, and early citation share there is cheaper to win.
Third, measure referrals by assistant, not as a single AI traffic bucket. A blended AI number hides whether ChatGPT, Gemini, or Claude is actually sending visits, and hides which one stopped. The SE Ranking data argues for a per-assistant dashboard as standard practice.
Referral share measures only the visits that AI assistants actually send, which is a shrinking fraction of AI usage if answers increasingly satisfy users without a click. A brand can therefore lose referral share while gaining answer presence, or gain referral share while losing the answer, and either story would be invisible in a referral-only dashboard. The SE Ranking numbers must be read alongside citation and mention data to describe AI visibility completely.
The practical consequence is a two-metric dashboard: referrals by assistant, showing who sends visits, and citations or mentions by assistant, showing who shapes answers. A brand that is mentioned often but rarely cited, or cited often but rarely clicked, has a different problem from a brand that is absent entirely, and each problem has a different fix. Referral share alone cannot distinguish them, which is why it should never be the only AI metric a team reports.
A final caution concerns the study itself: referral shares describe the websites in the study panel, and panels skew toward sites that already track AI referrals carefully. Smaller sites, non-English sites, and sites behind login walls may see different splits. Use the SE Ranking hierarchy as a prior, not as a verdict, and replace it with your own per-assistant referral data as soon as you have enough volume to split it reliably.
Sources: GeoAura GEO Statistics, November 2026, citing SE Ranking, 2026.
Linkbaits helps you create content that earns citations and backlinks. Start with the free tools and build your first linkable asset this week.
Start free trial