ChatGPT referrals to B2B websites are no longer a theory problem.
Demandbase’s own release on Aug. 12 shows ChatGPT-driven referrals to B2B sites grew roughly 4x over the past year. That is a big shift for one simple reason: it turns AI traffic into something you can measure instead of something you have to argue about.
For a long time, the objection was easy to make. “AI doesn’t send real customers.” “There is no traffic there.” “It is just hype.”
That argument is getting weaker.
What changed
The old model of discovery was built around search results, ads, and referrals you could track cleanly in analytics.
AI assistants are starting to create a second discovery layer. A buyer asks a question, gets a shortlist, and then clicks through from the answer surface. In that flow, the assistant is not just summarizing the web. It is acting like a recommendation engine with measurable downstream traffic.
That is what makes the Demandbase data important. It does not prove every business is getting meaningful volume from ChatGPT. It does prove the channel is real enough to show up in aggregate referral data.
That matters for anyone still treating AI visibility like a side conversation.
Why it matters
This is the part local businesses should pay attention to.
A dentist in DuPage County is not a B2B software company. A plumber, roofer, med spa, accountant, or law firm is not a Demandbase account. But the behavior underneath the channel is the same: people are asking an assistant who to trust, then acting on the answer.
That means the question is no longer, “Does ChatGPT send traffic?”
The real question is, “If someone asks ChatGPT for the kind of business I run, am I one of the names it can confidently send them to?”
That is a very different problem.
It is also why this pairs so cleanly with the question in How to tell if AI is actually sending you customers. The first step is knowing whether the channel exists for your business. The second is knowing whether the people it sends actually become customers.
What SMBs should do now
If ChatGPT is becoming a measurable referral source, small businesses should stop treating AI visibility like a brand exercise and start treating it like a traffic and trust problem.
That means three practical moves:
- Make your service pages easy to quote.
- Make your reviews sound like real customer outcomes.
- Make your business identity consistent across the web.
AI systems tend to reward clarity. They are better at pulling from pages that say exactly what the business does, who it serves, and where it operates.
If your website is vague, your reviews are thin, or your listings disagree with each other, you are making the assistant work harder than it wants to.
And when that happens, it will usually choose the easier option.
Where Proof Signal fits
This is the kind of shift Proof Signal should keep surfacing.
The story is not that ChatGPT is replacing search overnight.
The story is that it is becoming a measurable path into demand.
That is a much more useful conversation for business owners. It moves the debate away from opinion and toward evidence. It also gives local businesses a better lens for deciding where to invest: clearer service pages, stronger trust signals, better review language, and cleaner entity consistency.
Once a channel starts sending real referrals, the businesses that ignore it are not being skeptical. They are being late.
Do you know your AI footprint?
AI tools cite businesses that are Found, Understood, and Trusted (FUT). Get your free FUT score and see where your business stands.