Right now, usually nobody, and that is the actual problem. Muck Rack's 2026 State of PR report, a survey of 1,115 PR professionals fielded between May 14 and June 12, 2026, found that nearly three-quarters of them call generative engine optimization, getting cited by ChatGPT, Gemini and Perplexity rather than just ranked by Google, at least somewhat important to their work. At the same company, 29% say no single person or team is responsible for it. That gap between caring about something and assigning it to someone is the reason most companies' AI visibility stays an accident rather than a program.
Source: Muck Rack's 2026 State of PR report, surveying 1,115 PR professionals May 14 to June 12, 2026, linked in the article.
What did the 2026 State of PR study actually find?
A three-way split that leaves most of the work unclaimed. Among the 971 qualified responses Muck Rack analyzed after removing duplicates and low-effort answers, only 24% said PR or communications owns GEO at their organization, and only 11% said marketing leadership does. Add those two together and you still land under half. The remainder is split between other functions, shared informally, or, in that 29% figure, owned by no one at all. On top of that, 39% said they are not measuring GEO success in any form, which means even companies that have assigned the work often cannot tell if it is going anywhere. PRSA's own coverage of the report confirms the same methodology and figures.
Those numbers describe a function in the same position SEO was in roughly fifteen years ago: widely agreed to matter, inconsistently staffed, and reported on by whoever happened to notice it first.
Why doesn't PR or marketing just absorb it the way SEO did?
Because GEO's actual inputs sit on the PR side of the business, not the marketing side, which is the opposite of how SEO shook out.
Search engine optimization settled into marketing's territory because its core work, site structure, keyword-targeted content, technical crawlability, lives inside the same systems marketing already runs. AI visibility runs on a different input almost entirely: what independent publications say about you, written by journalists marketing teams do not pitch and do not control. Earned media, coverage a reporter decided on their own was worth writing, is specifically what large language models are trained to weigh more heavily than anything a brand publishes about itself. That is a PR function by definition, which is part of why the ownership question is genuinely unsettled rather than just under-resourced.
The technical half of GEO, structured data, crawlable content, how a site presents facts to a bot, does belong with the people who already own a company's web presence. So the honest answer is that GEO straddles two departments that rarely share a dashboard, and the 29% with no owner are the ones who never resolved which half goes where.
Why does it matter who is actually on the hook for this?
Because the work GEO depends on is exactly the work that goes undone when nobody is accountable for it.
Earned coverage is not a byproduct that shows up on its own. It requires a pitching program: a media list built around the right journalists, a steady supply of newsworthy angles, and someone tracking which placements are actually getting picked up by AI tools afterward. None of that happens through osmosis, and 29% of companies reporting no owner is a reasonable match for how many of them are likely getting coverage that never gets pitched with that outcome in mind. A press mention that nobody checked for AI pickup is still a press mention, but it is not a GEO program, it is a coincidence that occasionally pays off.
The measurement gap compounds this. If 39% of PR professionals are not tracking GEO results at all, a meaningful share of the 73% who called it important cannot actually show whether their efforts are working, which makes it hard to argue for more budget or headcount the next time priorities get set.
So who should own it: PR, marketing, or someone new?
Put the accountability for the outcome, whether your brand gets cited, inside PR or communications, and treat the technical implementation as a shared dependency marketing or engineering executes against a brief PR writes.
That split follows the evidence rather than org-chart habit. The raw material AI assistants cite most is independently reported coverage, which is PR's job to earn, not marketing's job to optimize. Our guide to generative engine optimization covers the technical side in detail, structured data, crawlability, llms.txt, and that work still needs doing. But technical correctness on a site nobody is writing about has little to cite. Assign the outcome to the team that produces the coverage, and have them pull in technical help the same way they already pull in a designer or a web developer for anything else that needs specialist execution.
What should that owner actually be doing day to day?
Four things, none of which requires a new department.
- Pitching coverage with AI pickup in mind, not instead of the usual newsworthiness test, but as an added filter: a story in a publication large language models already treat as a trusted source does double duty.
- Checking a citation tracker on a regular cadence, whether that is Google's own Search Console AI performance report, live since August 2026, or one of the paid cross-platform tools that track mentions across ChatGPT, Gemini and Perplexity.
- Making sure the technical basics are in place, confirming schema markup is implemented correctly and that an llms.txt file exists, even if someone else on the web team does the actual implementation.
- Reporting the result as its own line item, separate from traditional coverage counts, so the 39% who currently measure nothing stop treating AI visibility as an assumption rather than a number anyone can check.
None of this is exotic. It is the same discipline a stated PR strategy already requires, applied to a newer outcome that most reporting still leaves out.
What if your company has no dedicated PR hire at all?
Then the founder or the most senior marketing person on the team is the de facto owner, whether or not anyone ever said so out loud, and the fix is to say so out loud.
Most of the companies in this audience, startups in AI, fintech, crypto and tech more broadly, do not have a standalone PR department large enough to formally assign GEO to a sub-team. That does not make the ownership question disappear, it just means the answer defaults to whoever is already fielding press inquiries or writing the company blog, usually without the bandwidth to also track AI citations. The practical move at that size is smaller than a hire: pick one person, even part time, who checks a citation tracker monthly and owns the relationship with whichever agency or contractor is doing the actual pitching. An external PR partner can run the pitching and placement work, but someone internal still has to be the one who notices whether it is translating into actual AI pickup, because an agency reporting placements and a founder confirming those placements show up when a prospect asks ChatGPT about the category are two different checks, and only one of them catches the gap Muck Rack's data describes.
The direct answer, for anyone who read this far
Muck Rack's 2026 data shows a real gap between how many PR professionals say AI visibility matters, 73%, and how many companies have actually assigned someone to it, with 29% naming no owner and 39% measuring nothing. The fix is not a new department. It is naming PR or communications as the accountable owner of the outcome, since the coverage AI assistants cite is earned media PR already produces, while treating schema markup and technical crawlability as a dependency that team commissions rather than something left for whoever gets to it.
If AI visibility currently belongs to no one at your company, request a quotation and we will help you build the program, not just the report.
