Over the past 18 months search has changed more than in the previous decade. If you’ve noticed your Google Search Console showing impressions holding steady, whilst clicks quietly fall, you’re not imagining it, and it’s not a ranking problem. It’s a structural shift in how people find information, and it requires a new way of thinking about visibility.
This guide explains what LLMO is, why it matters right now, and most importantly what you can actually do about it.
First, what’s happening to search traffic?
When someone searches on Google today, there’s roughly a 60% chance they never click a single result. Google’s AI Overviews answer the question directly on the page, and the user moves on. In Google’s newer AI Mode, that figure climbs to 93%.
The old equation of rank well, get traffic, is breaking down. Organic click-through rates for position one results have dropped by as much as 58% when an AI Overview is present. One documented case saw impressions rise 27% year-on-year while clicks fell 36%, despite rankings actually improving.
This isn’t a crisis. It’s a recalibration. And it comes with a significant upside: the visitors who do click through from AI search convert at roughly 23 times the rate of traditional organic visitors. The volume game is giving way to a quality game, which means the businesses that show up correctly in AI responses will win disproportionately.
What is LLMO?
LLMO stands for Large Language Model Optimisation. You’ll also see it called GEO (Generative Engine Optimisation) or AEO (Answer Engine Optimisation). The industry hasn’t settled on a single term yet, and frankly the distinctions between them are minor. They all describe the same goal: getting your content cited, referenced, or recommended by AI systems, ChatGPT, Google’s AI Overviews, Perplexity, Claude, and the rest.
The big shift in mindset is this: traditional SEO was about rankings. LLMO is about relevance, trust, and clarity. You’re not trying to win a position on a list, you’re trying to become the source an AI is confident enough to quote.
The good news is that strong SEO fundamentals are strong LLMO fundamentals. These aren’t opposing disciplines. But there are some specific things you need to layer on top.
1. Content is still king, but the format has changed
The most important thing you can do for LLMO is produce genuinely useful, original content. Not content that repeats what everyone else has said, but content that contains something an AI can’t synthesise from thin air: your own data, your client results, your expert perspective, your direct experience.
AI systems are trained on the web. If every agency blog says the same things, AI will say the same things back. The way to get cited is to have something worth citing.
Beyond what you write, think about how you write it:
- Lead with the answer. Don’t bury the key point three paragraphs in. AI systems, and the humans using them, want the answer first, context second.
- Write in plain, information-dense language. This isn’t dumbing down; it’s precision. Clear, direct sentences are easier for AI to extract and quote accurately.
- Structure for scanning. Use clear headings, short paragraphs, and logical flow. If a section answers a specific question, make that obvious in the heading.
- Keep it current. There’s evidence that AI citation rates for content drop sharply once it’s more than three months old. Refresh your key pages regularly — new statistics, updated examples, current context.
2. FAQs aren’t just UX, they’re AI fuel
FAQ sections have always been good for user experience. In an AI-driven search environment, they become one of your most powerful visibility tools.
When a user asks ChatGPT or Google AI Mode a question, it’s looking for content that directly answers that question. A well-structured FAQ, with a clear question as the heading and a direct, complete answer beneath it, is exactly the format AI systems prefer to cite.
Reinforce your FAQs with FAQ schema markup. Schema is structured data you add to your page’s HTML that explicitly tells search engines and AI crawlers what type of content they’re reading. FAQPage schema is one of the most consistently cited signals across both traditional AI Overviews and conversational AI platforms.
Think about the questions your clients actually ask you. The ones that come up on every sales call, every onboarding call, every check-in. Write clear answers to those questions and put them on your site, marked up correctly.
3. Make sure your content actually exists on the page
This one sounds obvious, but it catches a lot of businesses out.
Many modern websites are built using JavaScript frameworks that build the page in the browser after it loads. From a user’s perspective, the page looks fine. But if you right-click and choose View Page Source, you might find that your key content — your headings, your body copy, your FAQs, isn’t actually there as plain HTML. It’s generated dynamically by a script.
AI crawlers frequently don’t execute JavaScript. They read HTML. If your content only appears after a script runs, it may be invisible to the bots that determine whether you get cited.
The fix: View Page Source on your most important pages and check that the content you’d want an AI to read is there as plain text in the source code. If it isn’t, that’s a conversation to have with your developer.
Beyond that, check your robots.txt file. This is the file that tells crawlers what they’re allowed to access. Many sites inadvertently block AI crawlers, and some CDN providers (notably Cloudflare) changed their default configuration to block AI bots automatically. If you’re on Cloudflare and haven’t checked your settings recently, it’s worth doing now.
4. The llms.txt file, a useful tool with honest caveats
You may have heard about llms.txt, a file you can place in your website’s root directory that acts as a curated map of your most important content for AI systems. It’s a simple Markdown file that lists your key pages with short descriptions of what each one covers.
The honest state of play in 2026 is that no major AI platform has publicly confirmed it uses llms.txt as a direct ranking or citation signal. The bots that drive AI search visibility don’t appear to fetch it significantly more often than any other page. Google has explicitly said it doesn’t use it.
That said, it still has real value, particularly for AI coding assistants like Cursor or GitHub Copilot, which use it to navigate your site in real time when developers are building integrations. For a digital agency or software product, that’s worth having.
It’s also a useful forcing function. Building an llms.txt file requires you to ask: what are the 10–20 pages on our site that best represent what we do and what we know? That question itself is worth answering.
One important caution: many llms.txt files are auto-generated by plugins or tools, and they often include 404 pages, irrelevant content, or pages you’d rather not spotlight. If you’re going to have an llms.txt file, audit it manually. Check every URL actually resolves. Check the pages it points to are the ones that best reflect your business. A curated list of 15 well-chosen pages is far more useful than an auto-generated list of 80 that includes your cookie policy and a blog post from 2019.
Keep it updated. If a URL changes, update the file. If you publish important new content, add it.
5. Good technical SEO is good GEO
There’s a temptation to treat LLMO as a separate discipline from SEO. It isn’t. The technical foundations are the same, with a few additions.
The things that make your site readable to Google make it readable to AI crawlers too:
- Page speed matters even more. AI crawlers often operate with very short timeouts. A slow page may be abandoned before the crawler reaches your key content. Aim for a fast time-to-first-byte, and make sure your most important content appears high in the page’s HTML — not after a long chain of scripts and styles.
- Schema markup. Beyond FAQPage, look at Article schema, Organisation schema, and BreadcrumbList. These give AI systems explicit, machine-readable signals about who you are, what you publish, and how your content is structured.
- Author credibility. AI systems give weight to content that has a clearly identifiable author with visible credentials. Author pages with a bio, role, and relevant experience help both traditional E-E-A-T signals and AI citation likelihood.
- Internal linking. Clean internal linking helps AI crawlers understand which pages are most important and how your content relates to each other. An orphan page — one with no internal links pointing to it — is easy for both search engines and AI crawlers to miss entirely.
- Clean, logical site structure. If a crawler has to guess what your site is about, it will sometimes guess wrong. Clear navigation, descriptive URLs, and a logical content hierarchy remove that ambiguity.
The bottom line
LLMO isn’t a replacement for SEO. It’s SEO evolved for an environment where the goal isn’t a click on a blue link, it’s being the source that AI trusts enough to quote when someone asks a question you should be answering.
The businesses that will win AI search visibility are the ones that:
- Produce genuinely useful, original content — not just volume
- Structure it so AI can read and extract it easily
- Answer the questions their customers are actually asking
- Keep their technical foundations clean and crawlable
- Stay consistent, updating content, maintaining their schema, checking their crawl accessibility regularly
There’s no secret algorithm to game here. The signal AI systems are looking for is the same signal good clients are looking for: clarity, expertise, and evidence you know what you’re talking about.
