Blog · AI visibility 7 min read

What is LLM optimisation? A plain-English guide for UK business owners

LLM optimisation is the work of making your business clear, consistent and quotable enough that AI assistants recommend it by name. Here's what that means, why it's suddenly worth your attention, and how it differs from the SEO you already know.

The thing that's changed

For twenty-five years, getting found online meant one thing: ranking on Google. Someone typed a few words, Google returned ten blue links, and you wanted to be near the top of that list. The whole industry of SEO grew up around that single screen.

That screen is no longer the only one. A growing number of your customers now open ChatGPT, Perplexity, Gemini or Google's own AI answers and simply ask: "Who's a good branding photographer in Cambridge?" or "Which firm should I use for X?" The AI doesn't hand back ten links. It hands back an answer — often two or three named recommendations, sometimes just one.

If your business is the name in that answer, you've won the customer before they've compared anyone. If it isn't, you're invisible, and you'll never even know the conversation happened. That's the shift. LLM optimisation is the work of being in the answer.

What "LLM" actually means here

LLM stands for large language model — the technology behind ChatGPT, Claude, Gemini and the rest. You'll also see the same work called GEO (generative engine optimisation), AEO (answer engine optimisation) or simply AI visibility. The labels differ; the goal is identical. Make your business something these systems understand clearly and feel confident recommending.

And here's the part most people miss: these systems don't think the way a search engine does. Google asks "which page best matches these keywords?" An AI assistant asks something closer to "what do I actually know about this business, and can I trust it enough to put my name behind recommending it?" That's a higher bar, and it's met differently.

The four things that get you cited

In practice, being recommended by an AI comes down to four things. None of them is magic. All of them are doable.

1. A clear entity

An "entity" is just a thing an AI can recognise and reason about — a person, a business, a service. Entity clarity means the whole web describes you consistently: same name, same location, same services, no contradictions. If one page says you're a wedding photographer, another says you're a consultant, and a directory lists an old address, the AI gets confused and falls back on someone clearer. Consistency is unglamorous and it is the single biggest lever.

2. Structured data

This is a small block of code, invisible to visitors, that spells out in plain terms who you are and what you do — your name, your services, your prices, your reviews. It removes the guesswork. The AI doesn't have to infer anything; you've told it directly, in a format built for exactly that purpose.

3. Quotable content

AI assistants lift answers out of pages. If your website is full of vague marketing language — "passionate about delivering excellence" — there's nothing to lift. If instead you answer real questions in clear, self-contained paragraphs, the way this very article does, you give the AI something it can quote directly. Write the answer your customer would want, in the words they'd use to ask.

4. Third-party authority

AI systems trust what other credible sources say about you, not just what you say about yourself. Reviews, a Google Business Profile, a mention in the local press, a listing in a reputable directory — these corroborate your claims. When several independent sources agree on who you are, the model treats it as fact rather than marketing.

"Do I really need this yet?"

Fair question. Here's the honest answer. If your customers could plausibly ask an AI for a recommendation in your line of work — and for most local and professional services, they already are — then this is worth doing now rather than later. Not because it's urgent in a panicky way, but because these systems learn over time, and the businesses that set up clear signals early are the ones the AI learns to trust first. Being early is cheap. Catching up later is not.

And there's a quieter reason: almost nobody is doing it well yet. The bar to stand out is low today and will only rise. That's a window.

Where photography comes into it

This is the part that surprises people, and it's close to my heart because I'm a photographer as well as a consultant. Original imagery is an AI-visibility asset most businesses waste. Distinctive photos, with proper captions and descriptive alt text, give these systems extra, high-quality signals about who you are — signals a competitor leaning on stock images simply cannot produce. Consistent branding photography doesn't just make you look credible to humans. It helps the machines understand you too. I've written a separate piece on exactly that.

The honest caveat

One thing I won't do is promise you a guaranteed position in any AI tool. Nobody who's being straight with you will. These systems are owned by other companies and they change without notice. What I can promise is to apply current best practice, measure where you stand before and after, and show you the change. Most clients see meaningful improvement within a quarter. That's the realistic frame.