OpenAI Just Put the GEO Industry on Notice. That's Good News for PR.

ChatGPT now tells companies how often it mentions them. Counting mentions just got cheap. Earning them didn't.

For two years, a whole software category has been built on one question. What does AI say about your brand?

GEO platforms raised real money answering it. They run thousands of prompts, log which brands show up, and sell the result as a visibility score. They could do that because the AI companies themselves shared almost nothing.

That changed on November 10. OpenAI launched Businesses in ChatGPT in beta, with a Business Hub where a verified company can claim its profile, correct its information, and see how often ChatGPT mentions it, how many people click through to its website, and which conversation topics surface it. It's free.

When the platform starts handing out the data your category was built to estimate, your business model needs a second look. For those of us in PR, it's the best news in a while.

A sample was never the real thing

Prompt tracking is useful. I say that as someone who publishes a quarterly AI Visibility Index built on it. It shows who gets cited in your category and where you're absent.

But a tool that says you appear in 35% of its test prompts is describing its test prompts. It can't say how many actual buyers saw your name or what they did next. OpenAI can, because it's counting real conversations among the 1.2 billion people who use ChatGPT each week.

We've seen this before. When Google made web analytics free in 2005, counting visits stopped being much of a business. The vendors that survived sold what to do about the numbers.

Don't write the obituary yet

Business Hub is a beta, and a narrow one. The numbers are aggregated and only appear above a minimum volume. You don't see prompts. You don't see competitors. You see nothing about Gemini, Claude, Perplexity or Google's AI Overviews. And claiming a profile doesn't move you up in ChatGPT's recommendations.

So third-party tools keep a job in competitive benchmarking and cross-engine comparison. What they lose is the ability to charge a premium for counting mentions.

Every GEO vendor now has to answer the question their dashboards have always dodged. You've shown me where I'm missing. What are you going to do about it?

A dashboard can find the gap. It can't close it.

This is where PR comes in.

AI assistants don't just take a company's word for it. Ask one to shortlist vendors and it leans on what other people have said in reporting, analyst notes, research and expert commentary. A dashboard can tell you a competitor made that shortlist and which publications put it there. Somebody still has to earn the coverage and give the market a clear reason to believe you.

That has been the PR job description for decades. Our industry spent those decades apologizing for impressions and reach, because nobody could connect credibility to demand. Now there's a discovery channel that runs on third-party credibility, and its owner has started publishing the numbers.

One caution for my own industry. Earned media doesn't guarantee an AI recommendation, and a visibility bump after a placement doesn't prove the placement caused it. Agencies that paste "GEO" onto the capabilities page and change nothing else will get caught, and they should.

What I'd do this week

1. Claim your Business Hub profile. It costs nothing and gives you a first-party baseline no vendor can.
2. Keep one independent tool for what OpenAI won't show you, meaning competitors and the other engines.
3. Write down the ten questions a serious buyer asks before building a shortlist. Track those and ignore the rest. A mention in a generic explainer is worth far less than a place in "who should I evaluate?"
4. Before you sign with any agency or vendor, ask which gap the work closes and how you'll both know it closed.

Knowing whether AI recommends you is now close to free. Being a company it has good reason to recommend is the expensive part, and it always was.

Source: OpenAI, "Introducing Businesses in ChatGPT," November 10, 2026. Launch details also reported by Search Engine Roundtable and Trending Topics.