The New Search: 40% of People Now Start Product Research on ChatGPT, Not Google

In 2020, if you wanted to find the best project management software, you typed it into Google. You got ten links. You clicked through three of them, read some comparisons, maybe checked a G2 review or two, and made a decision.

That process — the one that the entire SEO industry was built to win — is increasingly not how buyers research products anymore.

In 2025, 40% of B2B buyers start product research on ChatGPT, not Google. They type the same question — "what's the best project management software for a 20-person agency?" — and instead of ten links, they get a synthesised paragraph naming three or four specific tools, explaining why each one matters, and often recommending one for their situation.

The decision happens in the AI answer. Not in a click. Not in a comparison article. Not after five pages of research. In one response. And if your brand isn't named in that response, you don't exist for that buyer.

This article breaks down what the research shows, why buyers are switching, which AI platforms are driving the most purchase intent, and — crucially — what to do about it.

The shift away from Google-first product research is not anecdotal. Multiple independent research streams are pointing to the same conclusion: AI chatbots have become a primary channel for purchase intent queries, and they're growing faster than any acquisition channel since social media.

The 40% figure from Gartner is the headline, but the nuance is important: it refers to B2B buyers starting their research on ChatGPT. Many of them then continue to other sources — but the AI response has already shaped their initial shortlist. The brands named in that first AI answer have a significant head start in the evaluation process.

The McKinsey finding — that AI-assisted purchase research leads to decisions 2.9x faster — is arguably even more important for marketers to understand. Shorter research cycles mean fewer touchpoints. If your brand isn't present in the AI stage of research, you may have no opportunity to enter the consideration set at all before a decision is made.

The generational split is stark. Among buyers under 40, AI chatbots are already the default research starting point. The 67% figure from HubSpot tracks Millennials and Gen Z — the same cohort that is now moving into senior procurement and budget-holder roles. This isn't a trend that peaks and reverses; it's the leading edge of a permanent behaviour change.

The platform breakdown matters because each AI system decides which brands to recommend using completely different mechanisms. Understanding each platform's decision logic — not just its market share — is what separates brands that win AI visibility from brands that don't. We cover this in detail later in this article.

The shift to AI-first research isn't happening because buyers suddenly became more tech-forward. It's happening because AI search is a better experience for product research than Google — for several structural reasons that aren't going away.

The underlying point is this: Google was optimised for the information retrieval era. You had a query; Google helped you find the right page. In the product research era, buyers don't want pages — they want answers. They want synthesis. They want recommendations that account for their specific context.

AI chatbots are built for exactly that. And until Google fully closes the gap with AI-native search — something that is still a work in progress despite Gemini integration — the structural advantages of chatbot-first research will continue to drive the shift.

Here is the uncomfortable truth for most marketing teams: the fact that 40% of buyers are starting research on AI platforms is not matched by 40% of brands having meaningful AI visibility.

Airo has run AI visibility audits across hundreds of brands. The consistent finding: the majority of brands that rank on the first page of Google for competitive commercial queries have low or zero AI visibility on ChatGPT and Claude for those same queries. The gap isn't slight — in many categories, only 2–4 brands get named consistently across AI platforms, even when 15–20 brands compete meaningfully for Google rankings.

The visibility gap is growing, not shrinking, because the market is still early. Most brands haven't started investing in AI visibility. The few that have are pulling significantly ahead — and because AI recommendations compound (more citations generate more trust, which generates more recommendations), early movers will be difficult to displace once the category slots fill.

In many categories, there are 2–3 brands that consistently appear in AI answers and 10–20 that don't appear at all. If you're not tracking your AI visibility today, you don't know which group you're in.

Not all AI search platforms work the same way — and the platform a buyer uses determines which brands get named. Understanding which buyers use which platforms, and why each platform makes the brand recommendations it does, is the foundation of any AI visibility strategy.

The platforms reinforce each other. A brand that invests in citation breadth (for ChatGPT), answer-first content (for Perplexity), authority press (for Claude), and Google ecosystem presence (for Gemini) will outrank competitors on all four platforms simultaneously — because many of these activities feed each other.

But the starting point is ChatGPT. With 63% of AI search sessions, it's where the most buyers are — and it's often the first place brands discover they have a visibility problem.

The pattern we see with brands that successfully build AI visibility follows a clear sequence. It's not random. It's not just "publish more content." It's a structured set of activities aimed at two underlying goals: build the citation network that AI systems use to verify your brand exists, and create the content structure that makes it easy for AI systems to extract and recommend you.

Use this checklist to move from reading to doing. Items are saved in your browser — come back and check them off as you complete each one.