A year ago, if you wanted to research a product, a service, a brand, or had a problem to solve, you went to Google, typed something into the search bar, and picked through ten blue links until you found something useful. Today almost half of us (41% of consumers) start a search with an AI tool. Seven out of 10 searches now end without anyone clicking on a single result. ChatGPT alone has roughly 800 million weekly users. Google has rolled out AI Overviews to the majority of searches, and they sit above the traditional list rather than below it. The AI answer is becoming the search experience, not just an option inside it.

The shorthand for getting your brand cited and recommended inside those AI answers is GEO: Generative Engine Optimisation. Some people call it AEO, some call it AI Search, some still call it LLMO. All part of the same thing. 

For a brand that already takes PR seriously, this is, on balance, good news. Here’s why, and what to do about it.

What’s actually happening

Most brands are not in the AI answer. Yours probably isn’t either. This matters because the AI answer is now the first impression a buyer forms of a category. If your name doesn’t appear, you don’t exist to that buyer, and the rest of your funnel never gets the chance to do its job.

The brands that DO show up share a pattern. They have strong, accurate, consistent third-party coverage in the publications and platforms the AI engines treat as authoritative. A measurement study earlier this year found brands are 6.5 times more likely to be cited on someone else’s page than their own. None of that should sound unfamiliar to anyone who has spent a career in PR. 

The big one: for the first time, you can measure PR impact in an affordable way

Proving the downstream effect of PR on consideration and choice has been the industry’s quiet pain for as long as it’s existed. AVE was always nonsense and most people knew it. Share of voice was directionally useful but easy to game. Brand tracker surveys were expensive, slow, and never told you why anything changed. The honest answer for most PR programmes, most of the time, was “it’s working, trust me.”

GEO changes this, and not gradually. You can now ask the AI engines the questions your buyers are actually asking, see whose names come up alongside yours, see how your earned coverage is being absorbed, and see what changes when a new piece lands. The data is still noisy, the engines disagree, and AI visibility moves around. None of that makes it useless. It makes it a measurement problem rather than a faith problem.

For brands that have invested in PR seriously, this changes what’s possible. PR briefs sharpen. PR proposals stop selling activity and start selling impact. Renewals get easier. Boards get the story they’ve been asking for. The next twelve months of comms strategy is the first one in twenty years where the PR investment has a directly measurable effect on the answer the buyer is being given.

Three things you can sort out this week

Don’t wait for a strategy review or a procurement cycle. The first day of GEO work for any brand is housekeeping, and the housekeeping is almost always the same.

One. Get your own brand straight. Most brands have not written down, in one place, what they do, who they serve, and what makes them different. The website says one thing, the press release says another, the About page says a third, and the leadership bios say a fourth. AI engines cannot describe a brand if the brand cannot describe itself. Pick the plain-English version of who you are, what you do, and who it’s for. Put it on the homepage, the About page, the meta description, and the structured data block. Make sure it sounds like a human wrote it.

Two. Write the rest of the site as if you’re answering questions. This is the bit that catches PR-minded brands by surprise. AI engines favour content that is structured the way a buyer’s question is structured: clear question, clear answer, supporting evidence, then the next question. A good FAQ page does this naturally. A good service page can do it too, with a little restructuring. A typical brand brochure page, written as one long flowing claim, is much harder for an AI to lift cleanly into an answer. The rule of thumb is simple: if you can read a paragraph and not immediately tell what question it would answer, the AI will probably skip past it. Add headings that look like the questions a buyer would ask. Lead each section with the answer. Then explain.

Three. Read your robots.txt. A surprising number of brands have inadvertently blocked the AI crawlers that decide whether they appear in the answer at all. Make sure GPTBot, ClaudeBot, PerplexityBot and Google-Extended are allowed unless you have taken a deliberate decision to keep them out for licensing reasons. While you’re there, check that your sitemap.xml is current and listed. It takes ten minutes, and it’s the kind of thing that quietly undoes everything else.

That’s probably about a day of work. It won’t make you famous in AI, but it will stop you being invisible.

The bit you can’t sort out alone

There are plenty of tools and paid dashboards that let you type your brand in and get a number back. They look like the answer. In my opinion, they aren’t. 

AI visibility is not stable. Ask the same engine the same question twice in a week and you will frequently get different answers, different competitors named, and different sources cited. Recent measurement work is showing only around 30% of brands stay visible across runs of the same question. One snapshot misleads more than it informs. A pretty dashboard wrapped around an unreliable signal will give you a comforting graph and a strategy nobody can defend, which is worse than no strategy at all.

Real AI visibility work is a programme. It needs measurement across the engines that actually matter to your buyers, run on the questions they’re actually asking, with enough volume and frequency to see past the noise. It needs a PR partner who treats those measurements as a brief, not a scorecard, and who knows which third-party outlets, which experts, which comparison pages and which community conversations are driving the recommendations in your category. And it needs all that  housekeeping above held tight over time, so the work builds on itself rather than starting from zero every time. 

Where to start

If you’ve ChatGPT’d your own brand and didn’t like what came back, the next step isn’t to ChatGPT it again next week. It’s to ask your PR partner what they’re seeing across the category, and to commission real measurement so you know where you actually stand. The combination of real measurement and earned coverage briefed against the data is where the actual progress lives.

This is what my company, Known & Cited, does. We measure where brands stand across the AI engines that matter. We can focus on the discovery stage, consideration, or conversation. We work with you and your agency to build a search and process that is exclusive to your brand/business. We then crunch the data, provide a detailed report that tells you everything you need, from positioning through to recommendations, and then hand it back to your PR team.

Plenty of other people measure too. The point of this blog isn’t that you should buy from us. The point is that someone, with measurement to back them up and PR people to act on it, should be doing this work for you. Not an SEO business you’ve never heard of, of a fancy tech dashboard that requires a subscription fee, and a degree in IT to get the most out of it. 

A year from now, the brands showing up in AI answers will be the ones who took the housekeeping seriously, then built the proper programme on top of it, with their PR people in the room and measurement keeping everyone honest. That’s the prize, and it’s worth a great deal more than another quarter of guessing.