More Press Isn’t “More”: How Digital PR Drives Entity Formation in AI Search

George Zlatin

By George Zlatin

Last week we earned two placements for a client in a high-authority publication in their industry, positioning them as a credible voice on a specific concept they want to own. Those placements will still deliver the referral traffic and authority that good coverage has always delivered, but now, they’re doing a little something extra: They’re shaping how AI describes the client.

When a buyer asks ChatGPT, Gemini, or Perplexity to recommend a vendor or a partner, the model draws on a picture it has already assembled of each brand: who they are, what they are known for, which ideas their name shows up next to, and how consistently. That “picture” has a name, it’s your “entity”, and shaping it is called entity formation, or “entity SEO”. This is the variable that matters most in AI search right now, and shaping it is becoming the core job of digital PR.

What predicts an AI citation

Here’s the pitch you’ve heard from every agency with a fresh “GEO” page on their site: earn more coverage, build more links, and the AI will start citing you. More is more. But the data points in a different direction.

Ahrefs studied 75,000 brands and measured which signals best predicted showing up in Google’s AI Overviews. Branded web mentions, meaning references to a brand anywhere on the web, correlated at 0.664. Traditional backlinks correlated at 0.218. Branded search volume came in at 0.392. The three strongest signals were all off-site, and brand mentions predicted AI brand visibility roughly three times as strongly as backlinks.

Familiarity is what moves the needle. A brand whose name comes up often, in credible places, in connection with the same ideas, becomes an entity the model trusts enough to recommend. The head of content at Screaming Frog made the point in a recent Distinctly study: earned media fuels AI, and there is a widespread misconception that piling up press mentions automatically produces more AI visibility. Volume builds a fuzzy entity. Relevance and consistency build a sharp one.

What a good placement is actually doing (and when it counts)

Looking back at our placements we landed last week, the reason they matter isn’t the domain rating. It’s that a highly relevant publication, in our category, is now on record describing us as a credible voice on a specific concept we want to own. That’s a brick in the entity wall. Get enough of those bricks, all reinforcing the same idea, and the models start to treat the association as fact.

A placement earns its keep when an AI engine can read it, when it is relevant, and when it is recent.

Start with whether the engine can read it, because plenty of publishers block AI crawlers. The Distinctly study found that roughly 45% of AI crawlers could reach national publications, and about 37% could reach regional ones, with wide variation by sector. A feature in an outlet that blocks the bots builds your reputation with human readers but stays invisible to the models (and most PR target lists never check this). Because of this, it’s important to build crawler access into how you value a publication.

And… Freshness compounds it. AI systems lean toward recent content, so a placement from last month carries more weight than one from three years ago. Entity formation decays without new coverage feeding it.

You can’t put all your ducks in the DPR basket

Now the part we feel strongest about, because it’s where budgets get burned: Digital PR amplifies a strong foundation, so you have to build the foundation first.

Digital PR is not a magic wand you wave at a dead rabbit in a top hat. If your site confuses crawlers, if your content doesn’t actually answer the questions your buyers ask AI, then it does not matter how many eMarketer placements you stack on top. You will have built thought leadership on quicksand.

Think of it as a stack. The bottom is technical structure: a site AI can crawl, parse, and trust. The middle is content that genuinely resolves the questions people are asking, written to be lifted and quoted, not buried under a 400-word throat-clearing intro. (Models pull disproportionately from the top of a page, so put the answer where it can be found.) Digital PR is the top layer. It’s the authority and thought-leadership coat that turns “a company that exists” into “the company AI names when this topic comes up.”

Get the base and the middle right, then stack intentional PR on top. That is where you start cooking.

Running digital PR for entity formation

The word doing the heavy lifting in that last section is intentional. Random, scattered coverage builds a blurry entity. Deliberate coverage is how we build a sharp one.

Choose the concept before you pitch the story. Decide the one or two ideas you want the brand known for, then write down the prompts buyers actually type into AI in that category. Those prompts are your targets now. The client placements above landed because the concept came first and the pitch served it.

Position your proprietary data with intent. Original data is the strongest asset you have here. A Search Engine Land analysis of AI-cited content found 52.2% of it contained original or owned data. A good dataset gets requoted across the web, and every requote reinforces the entity. The data has to point at the concept you are building. A study that supports your chosen angle compounds your authority. A grab-bag of unrelated stats scatters it.

Choose relevance over raw authority, and stay consistent. A tightly relevant trade publication the AI can crawl does more for a niche entity than a bigger outlet that buried the brand in a roundup. Describe the brand the same way across every placement so the association repeats and sticks.

Cover the sources the models lean on. AI engines pull from review platforms, communities, and reference sites alongside editorial coverage. A real presence on the review sites and forums your buyers use feeds the same entity your PR is building.

Measure the entity. Baseline how often each engine cites the brand for category questions, which placements the models actually use as sources, your share of voice against named competitors, and how the brand gets described. Run the baseline, do the work, watch the share move. That is the measurable difference we hold ourselves to.

The takeaway

AI search raised the stakes for digital PR. Coverage now teaches the models who a brand is and what to recommend it for, on top of the link and the traffic it has always delivered. But it only works as the top layer. Solid structure, content that actually answers the question, and then intentional, relevant, data-led PR stacked on top, all pointing at the concepts you want to own. That’s the recipe. Anyone selling you the PR layer alone is selling you a roof with no house under it. 

When you build on that foundation, digital PR is how a brand makes it into the answer itself.

When you’re ready, we can show you the entity you have already built. Tell us your category and the questions your buyers ask AI, and we will run a baseline of how the models describe you today and where competitors are taking your share of the answer.

Read more from our blog