ChatGPT leans on vendor sites over Gartner and G2 for software picks

A new DerivateX study says ChatGPT’s software recommendations rely mostly on vendors’ own sites and obscure blogs, not on Gartner, G2 or the business press. The findings raise new questions for buyers and software marketers about which sources AI tools actually trust. Why it matters: - AI software recommendations may be steering buyers toward sources that are self-interested, lightly vetted or hard to identify. - Software vendors that invest in analyst relations and review badges may be optimizing for channels that do not drive visibility inside ChatGPT. - The findings suggest AI search is reshaping software discovery at the moment buyers are closest to a purchase decision. What happened: - DerivateX, a B2B SaaS search agency, studied where ChatGPT gets software recommendations across 40 software categories. - The company submitted one buyer-intent query per category with web search enabled. - DerivateX classified 188 cited sources across 105 domains. - Vendors writing about their own products accounted for 51% of ChatGPT’s cited sources. - Small, often anonymous websites made up another 23% of citations. - Analyst firms, review platforms and the business press combined for 16% of citations. - G2 and Capterra received zero citations across all 40 categories. - Gartner appeared twice, and both citations came from its user-review pages rather than analyst research. The details: - In one product-management category, the blog ideaplan was cited six times, more often than Forbes, Reuters or Gartner. - In marketing automation, a single blog post from a small consulting firm supplied four of the recommended tools. - DerivateX calls the pattern the “Authority Inversion,” arguing that the traditional middle layer of software research has been hollowed out. - Apoorv Sharma, co-founder of DerivateX, said software companies have spent years building trust through analysts, review platforms and the press, but AI has changed which sources matter. - DerivateX says the study is a snapshot of one platform and one query type. - The company says the work captures the point when a buyer is closest to a purchase decision. Between the lines: - The results suggest AI search may be rewarding content that is easy for models to surface, not necessarily content that buyers already recognize as authoritative. - For buyers, that means ChatGPT recommendations should be treated as a starting point for verification, not as an independent verdict. - For software companies, the study suggests visibility now depends on surfaces many teams are not actively tracking. - DerivateX argues that analyst coverage and review-site badges still matter for human trust, but may not be enough to influence AI-generated recommendations. What’s next: - DerivateX is positioning its own services around helping B2B SaaS brands get cited in ChatGPT, Perplexity, Gemini and Claude. - The company also publishes buyer guidance, including a guide to hiring an AEO agency for B2B SaaS teams evaluating outside help. - DerivateX says the broader shift could push software marketers to rethink where they earn citations and how they measure search visibility. The bottom line: - ChatGPT’s software picks appear to come more from vendor content and obscure blogs than from the trusted research ecosystem buyers have relied on for years.