Ask ChatGPT about a brand these days, and here's something wild: it's probably not reading that brand's own website to answer you.
It's reading what journalists wrote about it.
That's the finding from Muck Rack's May 2026 study, which looked at more than 25 million links cited by ChatGPT, Claude and Gemini. The result? 84% of those citations trace back to earned media — actual news coverage, academic research, government sites, third-party reporting. Paid content and advertorials? A measly 0.3%. You could spend a fortune on sponsored placements and barely register.
That number isn't a fluke, either. Across three editions of the Muck Rack study, earned media's share has held between 82% and 89%. The AI models seem to have picked up a very human habit: they trust outside voices over a company talking about itself. Makes sense when you think about it. Would you trust a used car salesman's pitch more than a friend's review? The machines learned from our writing, so they learned our instincts too.
Buyers Are Asking the Robot First
Here's the part that should really grab any marketer's attention. Forrester's 2026 State of Business Buying report found that B2B buyers now start their research with generative AI — ahead of a vendor's website, ahead of a sales call.
Think about that. Your prospect may already be building a shortlist inside ChatGPT before anyone on your sales team even knows they exist. And what's shaping that shortlist? Mostly earned coverage.
I've spent years building a life around curiosity — asking "why does this work the way it does?" and "what happens if I try something different?" This whole shift is a perfect example of why that mindset matters. The old rulebook said: buy ads, build a flashy website, hope people find you. The new rulebook says the story someone else writes about you might matter more than anything you write about yourself.
One Story, Many Payoffs
What really caught my eye is how earned media keeps working long after it's published. A paid ad campaign dies the moment the budget runs out. A news story doesn't. It sits there, gets indexed, gets cited by AI tools for months, sometimes years.
One good placement becomes a blog post, a sales asset, a stack of social posts, a proof point in your next pitch — and a source an AI model keeps pulling from every time someone asks a relevant question. One input, a dozen outputs. That's the kind of leverage that makes a flywheel spin on its own.
The Problem: Nobody's Measuring It Yet
Here's where it gets tricky. A Semrush study from June found that 45% of marketing leaders can't accurately measure their brand's presence in AI answers, and only 9% have real tools to track it. No referral links, no last-click attribution. It's a mess, honestly — the same mess SEO was in the early days, when nobody quite trusted it but everyone who ignored it got left behind.
So what do you actually do with this? A few practical moves:
Run the prompts your buyers are running. Ask ChatGPT, Claude and Gemini the real questions your prospects ask, and see if your brand shows up — and who gets credited when it does. That's your starting line.
Track your earned-to-owned ratio. How much of your AI visibility comes from outside coverage versus your own site?
Build a checklist for every placement. What does this story become next? An email? A LinkedIn post? Don't let one win sit idle.
And maybe the biggest shift: stop filing earned media under "PR update" and start putting it on the CMO scorecard, with its own metrics, its own budget, its own seat at the table.
None of this means abandoning your website or your paid search. You still need a solid home base. But the game has changed. The most trusted advisor your buyer talks to now might be an AI model — and it's reading the news, not your homepage.
So here's the real question worth sitting with: if a reporter wrote about your brand tomorrow, would it be a story worth repeating a hundred times over?
Source: CMSWire
Missed the last one? Building a Website Used to Take Weeks. Now It Takes an Afternoon.
