Opinions expressed by Entrepreneur contributors are their own.
Key Takeaways
- The playbook that worked for Google doesn’t move AI engines, and understanding why is the strongest business case for PR in twenty years.
ChatGPT, Gemini and Perplexity build their answers from earned media — the one channel you can’t buy, automate or publish yourself. That quietly changes what PR is for, and who needs it.
Try a quick experiment. Open ChatGPT or Perplexity and ask it to recommend the best companies in your category — the exact question a prospect might ask. Then ask what your company does and who it serves. Now compare those answers to the positioning your team spent months refining.
For most businesses, the two don’t match. The engine’s description runs a few years out of date, borrows a competitor’s framing, or flattens a carefully built value proposition into a generic one-liner. And the mismatch matters more every quarter: a Prosper Insights & Analytics survey found that 53% of Americans who use generative AI turn to it to search the internet. Those users don’t get ten blue links and a chance to click yours. They get a summary, and increasingly they treat it as settled.
So what actually moves those answers? Companies have spent two decades learning to influence Google with content and SEO. That playbook doesn’t transfer, and understanding why is the strongest business case for public relations I’ve seen in twenty years of doing it.
The engines have an editorial hierarchy, and your website is at the bottom
AI engines weight sources by credibility, and they’re refreshingly blunt about the pecking order. At the top sits third-party editorial coverage: reporting that has passed through journalists, editors and fact-checkers. Your own website, blog and social channels sit well below it, for an obvious reason — on your own properties, you can claim anything. When my firm surveyed the engines directly for a report on the attention economy, ChatGPT put a number on it, saying as much as 63% of its answers derive from traditional media sources.
In budget terms, that means the majority of what the most-used AI engine says about your company is drawn from a channel most marketing plans treat as a line item somewhere below paid search. Gartner has done the math and expects PR and earned media budgets to double by 2027, driven largely by AI’s preference for vetted editorial content.
Why this is the one channel you can’t self-serve
Here’s the uncomfortable part for the do-it-yourself instinct. Every other marketing channel is self-serve: you can buy the ads, publish the content, post the thread. Earned media is the exception by definition. It exists only when an independent journalist decides your story is worth telling. That gate is precisely what makes it valuable to the engines, and precisely what makes it hard.
Getting through the gate consistently is harder than it looks from inside a company, for two reasons. The first is perspective. Internal teams sit too close to their own story to see it the way a reporter will. The craft of PR lives at the intersection of what a company wants to announce and what journalists actually want to write about, and finding that intersection requires a third-party vantage point that an in-house team, by definition, doesn’t have.
The second is cost. A functioning in-house PR operation is not one hire. You need a senior strategist to set direction, a VP-level team member to oversee messaging and writing, and junior staff to handle the daily work of pitching media. Salaried out, that structure costs considerably more than retaining a proven PR team, and standing it up is a distraction from the work marketing departments are actually built to do. The honest division of labor is to keep strategy, content, paid and email in-house, where those channels are easily managed and proximity to the business is an advantage. PR is not one of those channels.
It’s also worth being clear about what a good agency actually sells. It was never really “press releases.” It’s external messaging, positioning, media relations strategy, editorial judgment and access — and in the AI era that combination compounds, because engines don’t just index the coverage — they adopt its language. Across our client work, I’ve watched a precise, memorable phrase from a single article in a modest but authoritative outlet surface in AI answers within days. When coverage is loose and generic, the engines improvise, and you inherit whatever they invent.
The other pattern I see, aside from building internal teams, is companies skipping specialist talent altogether and putting their own press release straight on the wire. The result is almost always the same. With no media relations done in advance, the release gets no coverage, and because it was written without an editorial eye, it reads as poorly constructed and overly promotional. That’s the worst of both worlds in the AI era: a release sitting uncovered on a wire service is exactly the kind of unvetted, self-authored material the engines discount, so it costs money and teaches them nothing.
What to actually ask of a PR partner now
None of this means hiring the first agency that mentions AI in its deck. It means holding any PR partner to a new standard. If you’re evaluating one, two requirements matter.
First, demand a baseline. Before any pitching starts, they should audit what each major engine currently says about you — same questions, clean sessions, answers logged — and define success as movement in those answers, not a clip count. “Coverage secured” is the old metric; “the engines now describe us the way we describe ourselves” is the new one.
Second, evaluate strategy, not contacts. Prospects almost always lead with the wrong questions: what publications do you work with, who are your connections? Access matters, but for any established agency it’s table stakes. Last year my agency placed stories with over 10,000 different journalists, and one recent story alone was picked up by 400 outlets. Numbers like those tell you a pipe exists; they don’t tell you what flows through it. The evaluation that actually predicts results is looking at the work: ask an agency to walk you through the strategies they built for different clients, and how they turned those strategies into narratives that shifted a market’s perception — both in AI engines and in the minds of the people reading the articles. That’s the skill you’re buying.
I’ve written before that marketing goals should translate directly from business goals. Earned media is the mechanism that carries your story into narratives that shift the market’s perception in support of those goals — both in the minds of human readers and in the answers AI engines give about you.
Key Takeaways
- The playbook that worked for Google doesn’t move AI engines, and understanding why is the strongest business case for PR in twenty years.
ChatGPT, Gemini and Perplexity build their answers from earned media — the one channel you can’t buy, automate or publish yourself. That quietly changes what PR is for, and who needs it.
Try a quick experiment. Open ChatGPT or Perplexity and ask it to recommend the best companies in your category — the exact question a prospect might ask. Then ask what your company does and who it serves. Now compare those answers to the positioning your team spent months refining.
For most businesses, the two don’t match. The engine’s description runs a few years out of date, borrows a competitor’s framing, or flattens a carefully built value proposition into a generic one-liner. And the mismatch matters more every quarter: a Prosper Insights & Analytics survey found that 53% of Americans who use generative AI turn to it to search the internet. Those users don’t get ten blue links and a chance to click yours. They get a summary, and increasingly they treat it as settled.
