In short. For twenty years PR helped search by earning links from news sites. That still works, but it’s no longer the main event. Google’s AI Overviews, ChatGPT and Perplexity build answers from sources they trust, and Muck Rack’s analysis of more than a million AI citations found 89% of them point to earned media. Gartner expects PR and earned media budgets to double by 2027 because of it. A press release on its own rarely gets cited, but the coverage it starts does. If a parent asks an AI which stroller or STEM toy to buy and your brand isn’t in the answer, the fix is a PR problem, not a keyword problem.
The case for PR in an SEO plan used to be simple. News sites have authority, a story earns a link, the link lifts rankings. That’s still true, and it’s still part of what we do.
What changed is where people get answers. A growing share of product questions now go to an AI assistant or get answered in Google’s AI Overview before anyone clicks a result. Gartner reports ChatGPT’s traffic grew 608% and Perplexity’s 262% between the first half of 2024 and the first half of 2025, while Google and Bing each slipped about 1% (Gartner, Top Predictions to Inform 2026 Comms Strategies). Search didn’t shrink. The answer moved.
An answer engine doesn’t rank ten pages. It reads a handful of sources it trusts and writes a reply, often with a citation. So the question for a brand is no longer “where do we rank” but “are we one of the sources.”
Muck Rack looked at more than one million links cited by AI systems and found 95% were non-paid and 89% were earned media. Journalism alone was 27% of citations, rising to 49% for questions about recent events (Muck Rack, What Is AI Reading?, 2025).
Read that against what most brands spend on. Paid placements don’t get cited. A sponsored post doesn’t get cited. A trade story that quotes your VP, a review in a parenting outlet, a news item about your retail launch: those do. The same work that earned you a backlink in 2019 now earns you a seat in the answer.
For a toy, baby or family brand that’s an advantage, because the category has a real press. Toy trade media, parenting outlets, holiday gift guides and the morning shows that run hot-toy segments every November all produce exactly the kind of dated, attributed coverage a model reaches for.
I made the case on Substack in April: the press release, written off for a decade as a corporate reflex, has become one of the more useful formats for AI discovery. Releases carry the things a model extracts well, clear facts, dates, names, numbers and a quote from someone with a title, and they’re the starting point for the coverage a model ends up citing (The Press Release Is Back From the Dead, April 2026).
The condition is the part people skip. “Send three or four releases a month” is not a strategy, and PR does not stand for press release. A wire serves a purpose when there is news: a new client, a product launch, a result worth reporting. Pushing a release with no story behind it produces a wire listing and nothing else, and a wire listing is not earned media.
So the release is infrastructure. The story and the relationships are the product.
Four things we do on every program, because they serve reporters and models at the same time.
We build pages worth citing. A case study with a named client, a date, a result and a quotable paragraph gets picked up. A page of adjectives doesn’t. When we tested buyer prompts in September, the pages that got Swerve named were case studies, an industry page and a dated news post (why PR drives AI visibility).
We time news to the retail calendar. Toy and baby coverage runs on a schedule: Toy Fair, back to school, the holiday hot-toy lists that get written in October. A story that lands inside that window gets covered and then gets cited for the rest of the season.
We put a person behind the quote. A model, like a reporter, prefers a named expert with a title over “a spokesperson said.” Executive profiles and trade commentary are search work now, not just reputation work.
We measure presence, not just rank. Alongside rankings we track which prompts name the brand and which sources carried it. The sources tell you where the next placement should come from.
Audit the answers before the plan. Ask ChatGPT, Perplexity and Google what a parent would ask about your category, and write down who gets named and which pages get cited. That list is your brief.
Fix the pages the coverage will point to. Reporters link to product pages and press rooms; make sure yours carry the facts a model can lift, with dates.
Earn in the window. One strong placement in a holiday gift guide does more for citation than a quarter of wire releases.
Run PR and SEO as one program. The teams that still treat them as separate vendors end up with coverage nobody links and pages nobody covers.
Yes. Links from news and trade sites still carry authority in Google, and they're a by-product of the same coverage that gets cited by AI. You don't have to choose.
On its own, rarely. The coverage it starts will. Write the release for the reporter, with a real story, and the citation follows the story.
Recent coverage carries the most weight for recent questions, so a steady cadence beats one big push. Evergreen pieces, a strong case study or a trade feature, can keep getting cited for a year or more.
The mechanics are the same; the outlets differ. We run both from one plan so a US launch and its Canadian counterpart land in the same window and cite each other.
About the author
Andrew Wagar founded Swerve Strategic Marketing in 2010 and runs it as President and CEO. He’s spent more than 30 years in PR and marketing, including PR Manager roles at Spin Master and the Canadian Standards Association.
He’s an Accredited Public Relations Professional (APR) with the Canadian Public Relations Society and has served on the Canadian Toy Association board since 2018, where he chairs marketing communications. Reporters covering the toy business call him when they need a source.
