Do people actually use AI to decide what to buy?

Yes, and at a scale that would have sounded made up two years ago. Half of B2B software buyers now start product research in an AI chatbot more often than in Google. A third have bought from a company they’d never heard of because an AI suggested it. If you’ve been treating AI answers as a novelty, your buyers stopped agreeing with you this year.

The numbers, and where they come from

In March 2026, G2 surveyed 1,076 B2B software buyers for its Answer Economy report. The findings stack up fast. 51 percent of buyers now begin their research in an AI chatbot more often than in Google. One year earlier, 29 percent did. 71 percent lean on chatbots somewhere in their research. Chatbots are now the number one source shaping buyer shortlists. And eight in ten buyers say AI sped up their decision.

This isn’t a niche behavior anymore. It nearly doubled in a year.

It’s not just business buyers

The consumer side is bigger. ChatGPT passed one billion monthly app users in May 2026, reaching this mark faster than TikTok, Instagram, or YouTube did. When a tool reaches a billion people, everyday buying questions flow through it: who to trust, what to buy, where to go. The habit forms at home and follows people to work. The buyer asking ChatGPT about running shoes on Saturday asks it about software vendors on Monday.

Does this apply outside software?

The G2 survey covers software buyers, so let’s be honest about its edges. It measures one market. But the habit it measures isn’t about software. It’s about how people ask questions now. A billion monthly users aren’t all buying SaaS. They’re asking about dentists, roofers, lawyers, hotels, and gear. The pattern shows up wherever a buying decision starts with a question. If your customers ask questions before they buy, this applies to you.

The stat that should worry you: 69 percent

Buried in the G2 data is the number that matters most. 69 percent of buyers say an AI chatbot surfaced information that led them to choose a different vendor than the one they expected to pick. And one in three bought from a vendor they had never heard of before the AI named it.

Sit with that. The shortlist isn’t yours to set anymore. A buyer can arrive expecting to choose you and leave with someone else, because the engine told a fuller story about a competitor. It cuts both ways, though. The same mechanics let an unknown business get named into a deal it never would have seen. We’ve written about why engines recommend one business over another. The short version: engines recommend who they understand, and understanding can be built.

What this means for your marketing

Google visibility still matters. But it measures the old front door, and ranking well doesn’t put you in AI answers. When half of your buyers start with a question to an engine, the first thing they see is an answer, not a results page. Whether you appear in this answer depends on whether engines can read you, quote you, and name you. This is the whole job of Answer Engine Optimization, and it runs on different signals than the ones your SEO reports track.

How to see it in your own numbers

You don’t have to take a survey’s word for it. Open your analytics and look for three things. First, referral visits from chatgpt.com, perplexity.ai, and their kin. These are people who read an answer about you and clicked through. Second, direct traffic drifting up while search clicks stay flat. People who hear your name in an answer often type it in later. Third, listen on sales calls. Buyers now say things like “ChatGPT suggested you.” When you hear it, write it down. This is your namability, measured for free.

What to do about it

Treat it like the measurable question it is. Ask an engine what your customers ask, and note who gets named. Run a free single-page scan to see whether an engine can read and cite your most important page. And if the answers in your market keep featuring other names, an AI Visibility Audit measures why and turns the gap into a prioritized list of fixes.

The question was never really whether buyers would use AI. A billion people already do. The question is whether the answer they get includes you.