You can’t throw a stone today without hitting someone talking in broad terms about how AI is going to change everything about the way we live. This is surely true. But we thought it would be more useful to look at something specific: how AI is already changing the way we look for, receive and process information. For those of us in communications, that’s massively important. It changes how we reach the right people, deliver the right messages and build real authority with our target audiences. That’s pretty much what we do every day, and AI is already substantially changing how we do it. 

The numbers below illustrate that fact, and they offer some clear guidance on how we all need to adjust. 

Here’s a practical question for communications people: When someone asks an AI system a question in your field, does the AI system know enough about you (or the brand you’re trying to build visibility for) to potentially include you in the answer? If so, does it trust what it knows enough to actually include you? That’s quickly becoming what matters most. Not where you rank on a page of links, but whether you’re part of the answer at all.

That’s what the data below is really about.

Across ChatGPT, Claude and Gemini, 84 percent of all AI citations come from earned media. Paid and advertorial content accounts for less than one percent. That’s a pretty clear signal about what these systems consider trustworthy enough to repeat and share.

Journalism alone makes up 27 percent of those citations, with other forms of earned media (such as expert commentary, bylined articles, trade press coverage, etc.) making up another 57 percent. Brand-owned content, everything a company publishes on its own site, accounts for only 16 percent of what is included in AI. Research out of the University of Toronto found that AI engines cite earned media roughly five times more often than brand-owned websites. And a controlled study from Stacker and Scrunch found that content distributed through third-party outlets sees a 325 percent lift in AI citation rate compared with the same content published only on a brand’s own site.

None of this is random. AI systems are built to look for credible, independent, well-corroborated information. This, of course, is what good editors have always required in a piece before publishing it. We can write anything we want about ourselves (and so many of us do). It carries a different kind of weight when someone else, a journalist, a media outlet, an independent voice, decides we have a story worth telling. AI systems seem to have picked up on that same distinction.

It also helps explain why the traffic that does come through AI search converts so well. One analysis found a 14.2 percent conversion rate for AI search traffic, compared with 2.8 percent for traditional organic search. Separate data from Ahrefs found that AI-referred visitors converted at 23 times the rate of standard organic search. It seems that people who arrive by way of an AI citation arrive with a higher level of trust baked in than they did before.

We’ve talked to our clients a lot about the PESO Model®, developed by Gini Dietrich, as a framework for integrated communications for several years now. This data doesn’t change that (to some of their chagrin). In fact, it makes the case for it more clearly.

None of this means paid, shared and owned media matter less. They matter differently, and, in some ways, more now. Owned content still has to reflect real expertise for anyone, journalist or AI system, to find it worth citing in the first place. Paid still plays a vital role in building visibility and demand. Shared platforms increasingly function as a place where AI systems pick up on how people and communities talk about a brand, not just how the brand talks about itself. Earned media has taken on a bigger role in that mix, but it works in tandem and in partnership with what’s happening across the other three.

This is where consistency comes in. At their core, LLMs are super powerful pattern recognition systems. To them, reliable, consistent patterns equal reliable trends and trustworthiness. This is critically important to understanding how they work.

AI systems build a picture of a brand from every touchpoint where that brand’s name shows up in the digital ecosystem. This includes the website, social posts, press coverage, what leadership says publicly, even visual identity across platforms. They treat consistency across those touchpoints as a sign of trustworthiness and credibility. They treat inconsistency as a reason not to trust what they’re seeing. This has always mattered for building credibility with an audience. It matters just as much now for building credibility with the AI gatekeepers standing between a brand and its audience.

Behind all of this is a bigger shift in how people find things at all. More of us are turning to AI for research we once would have done through a search engine, a friend’s recommendation or our own reading. As that keeps happening, being the kind of source AI recognizes and repeats is becoming as important as being findable on Google was fifteen years ago. We’re adjusting our own approach with clients accordingly, and we’re happy to talk through what that looks like for anyone navigating these rapidly changing currents as well. 

We put together a short research briefing with the data behind this post, along with sourcing and a few figures we didn’t have room to cover here. You can download it below.