The next decade of PR will be won by brands that show up inside AI answers – and that means rethinking how research, storytelling and earned media work together.
Neville Doyle, Strategic Client Partner, Ideally
***
Ask ChatGPT or Claude a question about your category. The answer it gives back isn’t random. It’s stitched together from sources the model has decided to trust.
Recent Muck Rack analysis found 94% of citations in generative AI tools come from non-paid media. Earned media alone accounts for 82%. Paid is a rounding error in the systems that increasingly decide who shows up and who doesn’t.
Gartner now predicts PR and earned media budgets will double by 2027 as AI replaces traditional search. That’s a structural shift in where attention lives and how brands get found.
This is the territory of GEO – generative engine optimisation, sometimes called AEO. The practice of showing up inside AI-generated answers. The mechanics aren’t complicated: answer engines reward sources that are credible, citable, and referenced consistently across the web. A brand running a steady programme of proprietary research and thought leadership in quality outlets is building exactly the signal GEO rewards.
Every piece of credible coverage now does two jobs: it reaches today’s reader. And it trains tomorrow’s answer engine.
The research bottleneck has to go
PR has lived with the same problem for years. The research that gives a story real weight takes too long, costs too much, and can’t be run often enough to sustain a programme. A bespoke study runs six weeks and a multiple of what most comms budgets can absorb on repeat. So research gets used once or twice a year, or not at all. Stories that could have landed a quarter ago stay in someone’s drafts.
That model is finished. When 94% of AI citations come from earned media, the speed at which a brand can produce credible, attributable data becomes the new battleground for consumer attention.
The tools to close that gap now exist. Platforms like Ideally let teams run nationally representative consumer research overnight, at a fraction of traditional cost – fast enough to catch a cultural moment, repeatable enough to sustain a programme across a year. The constraint has moved from cost and timeline to imagination and editorial judgement.
From single study to storytelling engine
Most brands still run one big study a year and call it a thought leadership programme. The brands that will win the GEO era are already moving to something different: a storytelling engine built on three layers.
Planned tentpoles. Quarterly or bi-monthly research-led stories that anchor brand narratives. Same rhythm as a content calendar, same authority as a report.
Reactive quick dips. Fast-turn surveys tied to live moments in culture, media or the category. The stat that lets a brand step into a conversation already happening, with something to say.
Repeatable output. A consistent flow of evidence-backed stories. The compounding part of the system.
Each study produces a new citeable asset. Each asset expands the brand’s footprint across the web. Over twelve months, a brand running three to four research-led stories per quarter builds a source library AI systems will draw from for years.
What it looks like in practice
You can see the model working in two recent examples:
The first is Betty’s Burgers, the Australian burger chain, preparing the launch of its new Madman Burger, hot enough that customers had to sign a waiver before eating it. The product was built to be a talking point, and the launch was built around it: a full integrated campaign designed to land from day one. Working with Ideally, the team layered in a research moment to evolve the story and extend cut-through beyond the initial launch beat. A three-question survey to 1,000 Australians overnight, not about the burger itself, but about the appetite for extreme food challenges. The insight came back clean and pitchable: one in three Australians say they’re eager to take on an extreme food challenge. That stat opened a second wave. The story moved from product news into a behavioural conversation about how people want to experience challenge right now, and the campaign landed a five-minute prime-time segment on the Today Show, hosts taking on the challenge live, branded Betty’s products in frame.
The second is Clemenger BBDO’s Fast Feels, built on the same thinking but at a different cadence. The premise was simple: capture real-time consumer reaction to cultural moments and turn quick-turnaround insight into sharp public commentary. From the Tesla backlash to election day to surprise interest rate calls, Fast Feels let’s the agency respond to what is happening in the market with evidence, not just opinion. It has become a recognisable way for them to share thinking consistently, and it kept the agency in the conversation week after week.
Elsewhere, we’re working with enterprise service clients in slow-moving, low-interest categories to build data tentpole earned media moments across the year. We’re early in the GEO era, but in categories where consumers are now as likely to ask ChatGPT as Google when they need a new telco, insurer or bank, first-mover advantage matters. Retraining how AI agents consider, trust and rank your brand is the work, and the brands doing it now will set the defaults everyone else fights against.
The choice in front of comms teams
The brands that win the GEO era will look more like newsrooms than campaign teams. A story every few weeks. A point of view backed by data. A growing library of evidence that earns coverage today and feeds the answer engines buyers will rely on tomorrow.
Every survey not run is a citation a competitor will claim instead. Every quarter spent waiting is six potential stories that never enter the source pool AI systems are learning from right now. The window to build that footprint is open, but it won’t be for long. The brands setting up their storytelling engines this year will be the default answers in their category by the next.
***
Ideally is a proud supporter of the 2026 Australian Effie Awards.
Entries close June 1, more here.