ChatGPT referrals to business-to-business websites surged 303% year-over-year, climbing from approximately 645,000 monthly visits in June 2025 to 2.6 million in June 2026 across a dataset of 1,584 companies tracked by Demandbase, the marketing intelligence firm reported August 12. The acceleration, which covers more than 11 billion total website visits during a 14-month window, marks an inflection point: artificial intelligence search has become a measurable customer-acquisition channel, one that delivers fewer but better-qualified leads than Google ever did.
The inflection arrived suddenly. Demandbase's data shows a sharp uptick beginning in May 2026, when OpenAI rolled out a product tweak that now looks like a structural shift in how software buyers find vendors. On May 7, ChatGPT began embedding clickable brand links directly inside its answers. Before that date, most responses contained recommendations but no URLs. After, a prompt asking for marketing-automation tools or CRM comparisons would return a structured list with working hyperlinks.
Multiple analytics panels clocked an immediate jump. Outbound referrals from ChatGPT nearly tripled within a week of the change, according to traffic studies published by Botpresso and others. By June, ChatGPT had established itself as a new referral source in the B2B sales funnel, with Demandbase tracking the sustained growth across its dataset of 1,584 companies.
What makes the shift noteworthy isn't volume. AI referrals still represent a sliver of total web traffic—0.32% globally in 2026, up from 0.24% the prior year and 0.02% in 2024, according to SE Ranking's analysis. Organic search, by comparison, accounted for 42.8% of sessions in a First Page Sage study published in August, though that figure has declined 8.5 percentage points since early 2023.
The surprise lies in conversion rates. Tracking data from multiple sources suggests ChatGPT-referred visitors convert at significantly higher rates than organic search traffic, a phenomenon practitioners have started calling "intent compression." The term describes buyers who complete research, vendor comparison and qualification inside the chat interface before clicking through. By the time they land on a website, they've already narrowed the field.
Austin Heaton, tracking Lumanu's analytics through March 2026, logged 5,130 ChatGPT referrals and 101 conversions over 60 days. His calculations showed large-language-model visitors converting roughly 4.4 times more often than those arriving from Google. Kinstel Solutions reported a 54.55% key-event rate from ChatGPT sessions between late June and mid-July. BeVisibleIQ published a case study last month showing a $10 million B2B software company generated $250,000 in annual recurring revenue from AI referrals over eight months, even though those referrals represented just 4.2% of organic sessions but contributed 23% of total revenue from search channels.
Rachel Truair, chief marketing officer at Demandbase, framed the change cautiously. "AI assistants are becoming an increasingly important way B2B buyers discover and evaluate companies, but much of that activity happens before a buyer ever reaches a brand's website," she said in the August 12 announcement.
That pre-click activity matters because it alters what marketers optimize for. Traditional search-engine optimization chases traffic volume and keyword rankings. Answer-engine optimization—the emerging discipline around influencing AI recommendations—focuses on citation frequency and brand mentions inside conversational responses, even when users never click out. Academic research suggests most don't. A July 8 paper found ChatGPT produced outbound clicks in 5.2% of conversation sessions. An August 5 study reported that clicks to sources in Google AI Overviews occurred in roughly 1% of those experiences.
ChatGPT dominates the AI referral landscape by a wide margin. Previsible's 2026 AI Traffic Study, which analyzed 6.77 million sessions across 166 Google Analytics properties from November 2024 through May 2026, reported ChatGPT accounts for 92.4% of trackable large-language-model referral sessions, according to Search Engine Land coverage published in July. SE Ranking's global web-share study put the figure slightly lower at 74.78%, followed by Gemini at 11.56%, Perplexity at 7.23%, Microsoft Copilot at 3.51%, and Anthropic's Claude at 2.62%.
Buyer behavior surveys help explain why the channel emerged so quickly. A Forrester survey of business buyers published in February 2026 found 94% used generative AI or conversational search as a meaningful information source during the purchasing process. A Gartner survey of 645 B2B buyers conducted August through September 2025 found 69% rely on sales representatives to validate insights generated by AI tools, according to a May 20 press release. SEMrush's March-April 2026 survey of more than 600 U.S. B2B professionals showed buyers deploying AI across discovery and evaluation: 72% research topics, 63% use it for writing and editing, 60% for data analysis and summarization.
SEMrush clickstream data from October 2024 through February 2026 showed average prompts per session increased substantially late in the tracking window, suggesting users were conducting deeper research within the chat interface before clicking out to vendor sites. That pattern aligns with what vendors and practitioners describe as intent compression—the collapsing of traditional funnel stages into a single conversational session.
Kevin Indig, writing in his July 27 Growth Memo, argued the shift is fundamentally about brand rather than performance marketing. "AEO/GEO is a brand channel disguised as a performance channel," he wrote. "AI recommendations shape demand (not citations)."
The measurement infrastructure remains uneven. Google Analytics 4 added a native "AI Assistant" default channel group on May 13, automatically categorizing traffic from AI chatbots including ChatGPT, Gemini and Claude, Search Engine Journal reported the following day. The rollout simplifies tracking, but practitioner reports suggest AI referrals still sometimes land in Referral or Direct buckets depending on implementation. Similarweb now offers an "AI Traffic" module. Vendors including ProCloser, Petra Labs, DiscoveredLabs and Avante Visibility have built answer-engine optimization practices emphasizing citation engineering, structured assets and third-party reviews to influence AI-generated answers.
Case studies show wide variation by vertical and optimization method. Petra Labs reported that Fractional Jobs saw 65% of inbound leads come from AI search in July, according to a case study with Google Analytics 4 snapshots. DiscoveredLabs documented roughly 100% growth in AI referrals for a B2B financing platform after implementing answer-engine optimization tactics through the first three quarters of 2026. Catalyst reported in May that B2B conversion rates from answer-engine optimization run four to 20 times higher than organic search. ProCloser noted AI referrals outperformed all other sources on purchase rate for a home-services software company, citing Analytics, Search Console and internal logging data.
Several sources cautioned that AI referrals remain low-volume but high-intent, and companies should track Analytics' "AI Assistant" channel and referrer domains to segment performance by engine.
The regulatory landscape adds uncertainty. The U.K.'s Competition and Markets Authority imposed publisher conduct requirements on Google's AI search features in early June, including opt-outs, clearer source links and non-retaliation commitments in rankings. The New York Times v. OpenAI/Microsoft litigation advanced through September, with the U.S. Department of Justice filing statements supporting OpenAI and Microsoft on September 2.
New surfaces for assistant-driven discovery continue to emerge. Apple announced Siri AI across devices on June 8, potentially creating another referral channel once link-out behaviors mature. Cloudflare announced a research pilot with OpenAI on July 8 to explore attribution and business insights. Nick Ryder, vice president of research at OpenAI, said in the press release that "up-to-date information is important for delivering accurate answers to people using ChatGPT."
Forrester predicted in February that buyers will increasingly use agentic modes to synthesize internal artifacts—product demos, sales transcripts—and compare vendors, compressing traditional go-to-market timelines further. IDC reported in April that 74% of enterprises have deployed at least one AI agent, and nearly 90% commit more than $1 million to agentic AI spending in 2026.
For B2B companies, the playbook is straightforward if not simple: track AI referrals by engine, optimize for recommendations rather than citations alone, and prepare stakeholders for lower traffic volumes paired with higher intent. The 303% year-over-year growth in ChatGPT referrals suggests the channel has moved past proof-of-concept. Conversion rates four times higher than organic search mean even modest traffic volumes can deliver meaningful pipeline. Companies that build answer-engine optimization into their growth strategies now will control visibility as AI search scales, assuming the current dynamics hold. That's a larger assumption than it might have seemed a year ago.
