Pillar · AI search optimisation (GEO / AI SEO)
The answer now sits above the links. The job is to be the cited source.
For a growing share of searches, Google writes the answer at the top and your customer never scrolls to the blue links. ChatGPT, Gemini and Perplexity do the same, reading across many sources and naming only a few. AI search optimisation, also called generative engine optimisation or AI SEO, is the work of being one of those named sources. Here is what it is, why it matters now, and how we build it into the managed work for UK small businesses.
Plain English. Grounded in sourced numbers and Google's own guidance, not folklore.
Why it matters now
Being the cited source is the new page one.
The old goal was to rank first in the list of links. That is no longer enough on its own, because for many questions the searcher reads the AI summary and never reaches the list at all.
~60%
of searches now end without a click to any website
Source: SparkToro, zero-click search research
~900M
weekly active users on ChatGPT by early 2026
Source: OpenAI announcement
Put those two numbers together and the shift is plain. A large share of search demand is being answered on the results page itself, by an AI summary stitched from a handful of sources, and a second, fast-growing channel of demand has moved into chat assistants that never showed a list of links in the first place. In both, the question is answered and only a few businesses are named. If yours is one of them, you are present at the exact moment of the question. If it is not, you are invisible for that search, no matter how well you would have ranked in the old blue links.
This is not the death of SEO. It is the same craft with a higher bar: alongside ranking, you now write to be quoted, mark up your facts to be extracted, and build the brand presence that makes a model confident naming you. The honest caveat, which we will keep repeating, is that nobody outside Google or OpenAI can promise you a citation. The mechanics are not fully disclosed. What we can do is close the gaps that reliably keep a small-business page out of the answer, which is the bulk of the problem. The full read on those gaps is in why your business is invisible in Google's AI Overviews.
The disciplines
Five things that decide whether a machine quotes you.
None of these are tricks. They are the same fundamentals that make a page good for a human reader, which is precisely Google's stated position on what makes a page eligible for AI features. Here is each one, and where to go deeper.
Be quotable and answer-first
State the answer plainly, near the top, in short self-contained sentences a machine can lift. Question-shaped headings, a real FAQ, no waffle to wade through.
Entities and schema
Declare who you are and what your page asserts in machine-readable form: Organization, LocalBusiness, FAQPage and Article markup, so an engine can read your facts without guessing.
llms.txt and crawlability
Make your site easy for AI crawlers to fetch and parse, from clean indexable pages to an llms.txt that points the models at the content that matters.
Brand presence and third-party citations
Answer engines lean on the wider web: best-of roundups, directories, forums and reviews. Being named in the sources the models already trust gets you into the answer by proxy.
Measuring share of voice in answers
Track how often, and how accurately, you are named for your key questions across the main answer engines, because you cannot improve a citation rate you never measure.
In depth
What each discipline actually asks of your site.
-
Be quotable and answer-first
A machine that has to wade through marketing copy to find your point will quote a clearer rival instead. Structure pages around the questions your customers actually ask, put each question in a heading, and answer it in the first sentence beneath, in one or two tidy, self-contained statements. A genuine FAQ section, each answer a sentence or two, is one of the most reliable ways to feed an answer engine exactly what it needs. The discipline is simple: read each heading, read the sentence under it, and if that sentence does not answer the heading on its own, rewrite it until it does.
-
Entities and schema
Structured data is code in the background of your page that labels your facts, your business, your address, your hours, your FAQs, your reviews, in a form a machine can read without guessing. Google is careful here, and so are we: it states that schema does not directly improve ranking and that there is no special markup that buys a citation. What it does is remove ambiguity, declaring "this entity is a plumber in Leeds, here is the answer to this question", and it earns rich results in classic search, so it pays its way regardless. The full markup detail is in schema for AI answer engines.
-
llms.txt and crawlability
An answer engine cannot quote a page it cannot fetch, render or parse. The foundations are unglamorous: clean, indexable, fast pages with the content in the HTML rather than buried behind heavy code. On top of that sits llms.txt, an emerging convention modelled on robots.txt that points AI crawlers at the content that matters. It is cheap to add and worth doing, but it is not magic, and we say so plainly in the llms.txt guide for UK small businesses.
-
Brand presence and third-party citations
Models do not only read your site. They lean on the wider web: best-of roundups, directories, trade bodies, forums and reviews. Being named in a respected "best plumbers in Leeds" style article gets you into the AI answer by proxy, because the engine draws its recommendation from that list. So the work runs off-site too: earn your place in the roundups, keep your name, address and details consistent everywhere, and build the recognisable, stable brand entity a model learns to associate with your field. This is the slowest discipline and the hardest for a rival to copy, which is exactly why it compounds.
-
Measuring share of voice in answers
You cannot improve what you do not measure. The discipline is to check, regularly, how often and how accurately you are named for your key questions across the main answer engines, where you are cited, where a rival has taken your place, and whether the facts the models repeat about you are even correct. That tells you where to concentrate effort next, the same way rank tracking did in classic SEO. The how-to is in how to get cited by ChatGPT and AI Overviews.
How we do it
It is built into the managed work, not sold as a bolt-on.
There is no separate "AI SEO" package to buy, because there is no separate stack. A well-built, well-structured, fast, attributable site is what makes a page citable, and that is the standard every site we run is already held to. In practice, making a UK small-business site citable looks like this:
- Restructure key pages around real customer questions, each as a heading with a direct one-to-two-sentence answer in the first line, and add a genuine FAQ to the service pages.
- Ship the structured data: Organization or LocalBusiness site-wide, FAQPage on the new FAQ content, Article with a named author on guides, validated against Google's Rich Results Test so the markup is clean.
- Make the entity legible: name, address and phone rendered as text, consistent with your Google Business Profile, a real About page, and author bylines with a one-line credential on every guide.
- Keep the pages fast and fresh: sub-second loads on mobile, no layout shift, and a light review rhythm on the pages that genuinely change, so nothing useful is hidden or stale.
- Track where you are named, so the work is measured against actual citations across the answer engines, not guessed at.
This is the seventh of the nine disciplines in our SEO playbook, run joined-up with the other eight as part of the managed website service, on one monthly bill with no lock-in. The same fundamentals that earn an AI citation are the ones that win a human searcher, which is why we do not treat them as two separate jobs.
Go deeper
The supporting guides.
This page is the hub. Each discipline has a full write-up, with the honest caveats and the practical steps. Start wherever your gap is.
llms.txt for UK small businesses
What the file is, how to write one, and an honest read on how much it actually moves the needle.
Read the guide →Schema for AI answer engines
The Organization, LocalBusiness, FAQPage and Article markup that lets a machine read your facts cleanly.
Read the guide →Get cited by ChatGPT and AI Overviews
The practical playbook for being quoted, plus how to measure your share of voice across the answer engines.
Read the guide →For the foundations this all sits on, see the full 9 types of SEO playbook and why most small-business sites are invisible in AI Overviews.
FAQ
AI search optimisation, answered.
What is AI search optimisation?
AI search optimisation is the practice of getting your business cited inside AI-written answers: the AI Overviews that now appear at the top of Google, and the direct answers from tools such as ChatGPT, Gemini and Perplexity. Instead of showing a list of links and letting the searcher choose, these systems read across many sources and write one answer, naming a few of them as references. The goal is to be one of those named references. It works in two ways: making your facts easy for a machine to extract cleanly, and making your brand familiar enough that the models reach for you when they answer.
Is generative engine optimisation (GEO) the same as AI SEO?
Yes, in practice they describe the same work under different names. Generative engine optimisation (GEO), AI SEO, and answer-engine optimisation all mean optimising so that AI systems quote and cite your business when they generate an answer. The labels are new and still settling; the discipline is consistent. We use AI search optimisation as the plain-English term because it says what it does.
Why does AI search matter for a UK small business now?
Because the answer now sits above the traditional results, and for many searches the customer reads it and never scrolls to the blue links. Around 60% of searches now end without a click to any website (SparkToro), and ChatGPT alone reached roughly 900 million weekly active users by early 2026 (OpenAI). If your business is named inside that answer you are visible at the moment of the question; if you are not, you are effectively invisible for that search, however well you would have ranked in the old list of links. Being the cited source has become the new page one.
How do I get my business cited by ChatGPT and AI Overviews?
Make your answers easy to lift and your brand easy to trust. In practice that means answering real customer questions in plain English with the answer in the first sentence, adding a genuine FAQ, marking up your facts with Organization, LocalBusiness and FAQPage schema, keeping your details consistent everywhere they appear, and earning a presence in the third-party sources the models lean on, the best-of roundups, directories and forums. Google's own guidance is that there is no secret schema that buys a citation: the same fundamentals that make a page good for human searchers make it eligible for AI features.
Does structured data guarantee a citation in AI Overviews?
No. Google states that structured data does not directly improve your ranking and that there is no special markup or technical requirement to appear in AI Overviews. What schema does is help a machine understand and qualify what your page asserts and who is behind it, and it earns eligibility for rich results in classic search, so it pays its way regardless. The honest framing is that schema removes the guesswork; it does not buy the answer. Anyone promising a guaranteed citation is selling folklore.
What is llms.txt and do I need one?
llms.txt is a simple text file, modelled on robots.txt, that points AI crawlers and models at the content on your site that matters most. It is an emerging convention rather than a settled standard, and it will not buy you a citation on its own, but it is cheap to add and it sits alongside the more important fundamentals: clean, crawlable, fast pages an engine can fetch and parse. The detail, and an honest read on how much it actually moves, is in the supporting guide.
How is this different from normal SEO?
It is built on the same foundations, not a separate stack you buy. A page that is well-structured, genuinely helpful, fast, and clearly attributable to a real source is what wins both classic rankings and AI citations. The shift is one of emphasis: alongside ranking in the list of links, you now also write to be quoted, mark up your facts for extraction, and build the brand presence that makes a model confident naming you. AI search optimisation is the seventh of the nine disciplines in our SEO playbook, not a replacement for the other eight.
Does UK Web Marketing offer AI search optimisation as a separate service?
No, it is built into the managed SEO and content work, not sold as a bolt-on. There is no separate AI SEO package to buy because there is no separate stack: a well-built, well-structured, fast, attributable site is what makes a page citable, and that is the standard every site we run is held to. We restructure key pages around real questions, ship the schema, keep the entity legible and the pages fresh, and track where you are being named. The free Site Score is the quickest way to see where your current site stands.
Ready to be the cited source?
See where your site stands, then we close the gaps.
The fastest first step is a free Site Score: it shows where your current site stands on structure, speed and the basics that decide whether a machine can quote you. After that, we build the AI search work into the managed plan, on one bill from £49 a month, with no lock-in and one named person to talk to.