Blog - The prompt-to-purchase pipeline: What millions of search events told us about how AI influences buyer behavior

Open original

Analysis of anonymized AI chat data shows that when platforms like ChatGPT or Gemini actively recommend your brand, new users become roughly twice as likely to search you, visit your site, or view your products compared to mere mentions. The lift varies by platform, placement, and framing, with “best” or first-listed recommendations driving the highest follow-up intent.

  • When AI chatbots actively recommend a brand (vs. just mentioning it), users become roughly twice as likely to search for it, visit its site, or view its products—effects platforms usually misattribute to "organic" or "direct" traffic.
  • ChatGPT recommendations drive far stronger downstream action (up to 267% lift in retailer page views) than Gemini's, which barely moves site visits and doesn't boost retailer views at all—likely because ChatGPT's user base skews more purchase-ready.
  • The biggest gains come from converting new audiences with no prior brand interest, not from re-engaging existing customers.

When AI platforms like ChatGPT or Gemini actively recommend your brand to someone who hasn’t shown recent interest, that person is roughly 182% more likely to Google your name, 117% more likely to visit your site, and 185% more likely to land on a retailer’s product page within a week. Scrunch’s analysis of millions of anonymized AI chats and corresponding web behaviors (February–May 2026, opt-in panel) shows that these “recommendations” carry about twice the impact of mere mentions. Because most AI search results don’t generate a direct click, brands often misattribute the traffic boost to “organic” or “direct” channels, underestimating the role generative engines play in feeding the funnel. Not all AI assistants perform the same. ChatGPT recommendations drive a 116% lift in Google searches, a 171% lift in site visits, and a 267% lift in retailer page views. Gemini, by contrast, delivers a 109% lift in searches, only a 39% lift in site visits, and no meaningful boost to retailer views. Claude’s smaller sample shows even bigger swings—up to 436% more searches—but follows the same pattern. The key difference isn’t the code; it’s who uses each tool. Frequent ChatGPT users tend to be more purchase-ready than typical Gemini users. The biggest gains come from reaching new audiences—people with no prior brand searches or site visits—rather than nudging existing customers.

Ask about this article