I have been watching PR pros make a claim that keeps spreading: earn media coverage, and AI will recommend your brand. The pitch sounds clean, but the reality is messier. Landing a placement and getting AI to cite or recommend you are two different things, and treating them as the same thing leads to decisions built on shaky ground.
Here is what is worth understanding, whether you are a CEO evaluating your team’s PR investments or trying to make sense of why AI visibility keeps coming up in strategy conversations.
PR can help, but the pathway matters
The BuzzStream team studied 12,000 AI responses covering about 200 brands and found that only 23.1% of brand mentions in those responses were accompanied by a citation to that brand’s domain in the same response. That gap alone should shift how you think about what you are actually buying when you invest in PR with a GEO angle.
PR can support AI visibility in four ways, and the degree of directness varies significantly.
1. Your research can become a source AI cites.
This is the most direct pathway. If your team publishes original, data-driven research tied to a buyer problem, and that research earns coverage in relevant publications, AI systems can encounter it as a credible source. The key phrase is “tied to a buyer problem.” Data that is merely related to your industry is not the same as data that answers a question a prospective buyer is actively searching.
The BuzzStream example is instructive. A study on what link building costs earns AI citations on searches like “how much does link building cost?” because the person asking that question is deciding whether to build a process in-house or pay for links. That is buyer intent. A study on the number of journalists in the US is interesting, but it does not serve the same function.
If you are commissioning research for GEO purposes, ask your team: who is searching for this, and what decision are they trying to make?
2. Consistent brand descriptions can shape how AI defines you.
AI language models are probability-based. The way your brand gets described across third-party publications influences how AI is likely to describe you in response to a query. If every press release, media pitch, and boilerplate describes you the same way, journalists who cover you are more likely to use that same phrasing, and AI has more consistent signal to work with.
This is a low-cost discipline: pick a descriptor and use it everywhere. If you want AI to describe your company as “a fractional analytics firm for mid-market CEOs,” that phrase needs to appear in how you introduce your brand to journalists, in your press releases, and in your boilerplate. One mention is not enough. The goal is consistency across enough third-party references that AI systems start associating those words with your brand.
3. PR-driven backlinks can support search visibility, which feeds AI.
This pathway is real but indirect. Links from credible publications can support your search rankings. Search is still part of how some AI systems surface content. So links matter, but you are looking at a two-step relationship, not a direct one. Do not expect a single earned link to show up in AI citations. Think of it as part of a longer-term visibility stack.
The types of content that earn links through PR are not complicated: proprietary data tied to trends journalists are already covering, recurring reports that build familiarity over time, and survey data that tells a timely story tied to a buyer fear or aspiration. The Sole Supplier stat on Vans sales during the Squid Game trend is a clean example, a specific and surprising number connected to something already in the news. Journalists linked to it because it was useful to their stories.
4. Brand awareness campaigns change what people ask AI, not just what AI says.
This is the most indirect pathway, and it is easy to underestimate. When someone searches for your brand by name, AI is significantly more likely to mention and cite you than when someone searches a generic category query. The BuzzStream research confirms this: “What does BuzzStream do?” produces very different AI responses than “What is the best digital PR tool?”
PR campaigns that put your brand top of mind before someone reaches the buying stage can change the prompt people use when they go to search. A campaign that does not influence AI directly can still influence how your customers approach AI, and that shapes what they find.
What good measurement looks like
Before your team runs any PR campaign with a GEO goal attached, three things need to happen first.
Get a baseline. Use tools like Ahrefs, Semrush, Peec AI, Profound, or Xofu to measure where you currently stand on the specific prompts that matter to your business. You cannot measure progress from an unknown starting point.
Pick one thing to move. Are you trying to get your research cited on a specific factual question? Are you trying to change how AI describes your brand category? Are you trying to increase branded search volume? These require different tactics and different measurement approaches. Trying to claim credit for all three at once is how PR results get inflated.
Track the same prompts over time, per platform. One Profound study found that 40 to 60 percent of cited domains changed when comparing months on the same platform. That volatility means a single snapshot is not meaningful. You need to watch the same set of prompts across weeks or months, and you need to track platforms separately, because ChatGPT, Google AI Overviews, and others behave very differently.
Report earned placements separately from AI visibility changes. They are related, but they are not the same metric. Combining them obscures what is working.
The decision you need to make
PR can play a real role in how AI encounters, understands, and recommends your brand. But none of the four pathways is a guarantee, and treating a placement as proof of AI influence is the kind of faulty reasoning that leads to misallocated budget.
The right question is not “did we earn coverage?” It is “did we move the specific metric we were trying to move, and can we trace the path?”
That second question is the one every agency report eventually faces, usually from the most skeptical person in the room: where did that number come from? I call the answer a glass box. Name the inputs, show what each one counts for, say what you left out, and be honest about what you still cannot prove. A visibility number nobody can open does not survive the meeting, even when it happens to be right.
If your team cannot answer that question, you are investing in activity, not outcomes.
Book an Executive Data Assessment and find out what your visibility numbers are actually telling you.
Original source: PR for GEO: 4 Ways It Can Influence AI Visibility
