A glass search lens focusing scattered points of light into one bright cited beam

How to Rank in AI Search: A GEO Strategy for UK SMEs

To rank in AI search, make each page easy to lift: answer the question in the first two sentences, use question headings, keep the page indexed and eligible for a snippet, let AI crawlers in through robots.txt, and earn mentions on sites AI systems already trust. Then check ChatGPT, Perplexity and Google AI Overviews monthly and fix the gaps.

Why is SEO alone not enough to rank in AI search?

Traditional SEO earns a place in a list of links. AI search answers the question itself, pulling passages from several pages and citing a handful of them. You still need the SEO basics, because Google's AI features draw on its own index, but you also need passages clear enough to quote on their own.

AspectTraditional SEOAI search
GoalA top position for a keywordA citation or mention inside the answer
What gets pickedThe whole pageA passage that answers one question
Content focusMatching the keywordA complete, direct answer
StructureKeyword headingsQuestion headings with the answer first
Success measureRankings and clicksCitations, mentions and clicks from AI answers

When an AI Overview appears, it sits above the organic results and can fill most of a phone screen. A page can show in more searches yet earn fewer clicks, because the answer is given on the results page. I explain what that does to your figures in why your Search Console CTR collapsed.

Who is eligible to be cited

Google says that to be shown as a supporting link in AI Overviews or AI Mode, a page must be indexed and eligible to be shown in Google Search with a snippet. Beyond that there is no extra technical requirement, only one Search Console setting: the site must be included in Search generative AI features, which is the default. Eligibility is only the floor, though.

Ahrefs found that 37.9% of URLs cited in AI Overviews also appeared in the first 10 results for the same query (March 2026), down from about 76% in July 2025. Citations are not limited to the top-ranking pages, but they go to pages that answer the question well.

If you want the definition and background of the discipline first, read my guide to generative engine optimisation. If you would rather talk it through for your own site, Book a Discovery Call.

Comparison table of traditional SEO and AI search across goal, what gets picked, content focus, structure and success measure
The goal moves from a position in a list to a passage quoted inside the answer.

How do you structure content so AI search engines find it?

Put the answer first and make every section stand on its own. Use one title, question-style H2s, and open each section with a short paragraph that answers its heading without needing the rest of the page. AI systems lift passages rather than whole pages, so a passage that makes sense alone is the one that gets quoted.

A pattern that works

  1. A heading phrased the way a buyer would ask it.
  2. A 40 to 60 word answer straight underneath.
  3. The detail: steps, an example, a short table for any comparison.
  4. A source for every figure, named and dated.

AI Overviews often present answers as lists, so give them lists to work with. Use bullet points for features and options, numbered lists for processes, and tables wherever you compare prices, plans or methods.

What changed with schema markup

Schema markup still helps search engines understand a page, but the rich results it earns have changed. In August 2023 Google limited FAQ rich results to authoritative government and health websites and stopped showing HowTo rich results. Since 7 May 2026, FAQ rich results no longer appear in Google Search at all.

Most businesses should keep Article, Organization and LocalBusiness markup, and keep FAQs as visible text on the page, where AI answers can still read them.

Action step: check your three most important pages for one H1, logical H2s and H3s, and an answer in the first two sentences under each heading.

Pro tip: educational content that answers real customer questions gives AI Overviews something worth citing. For a service business, that is usually the best place to start.

How long should content be to rank in AI search results?

There is no word count that ranks. A page should be as long as it takes to answer the question completely, with sources, real examples and a named person behind it. A focused 800-word answer to one question can serve a reader, and an AI system, better than a 3,000-word guide that buries the answer halfway down.

What the research suggests

The most-cited study here is the GEO paper by Aggarwal and colleagues, first posted to arXiv in November 2023. Its authors report that "GEO can boost visibility by up to 40% in generative engine responses". The caveat matters: that was a lab benchmark on 2023-era models, so treat it as a direction, not a forecast for your site.

The methods that helped most in that study were adding quotations, adding statistics and citing sources. All three are things a careful writer does anyway.

Show who wrote it

Add an author bio with real credentials, a publication date and an updated date. Cite primary sources, and for UK topics prefer GOV.UK, the ICO, HMRC, Companies House, ONS or the relevant trade body.

Action step: take your three most important service pages and make each one fully answer the questions buyers ask, with five to seven FAQs and examples you can evidence.

Common mistake: rewording a competitor's page. It adds nothing an AI system cannot already find elsewhere, so there is no reason for it to cite you.

What technical changes help a website rank in AI search?

Make sure crawlers can reach your pages, the pages are indexed and eligible for a snippet, and they load quickly. For Google's AI features the technical bar is the same as ordinary Search, plus inclusion in Search generative AI features in Search Console, which is on by default. For ChatGPT search, your robots.txt must not block OpenAI's search crawler. Once access is sorted, fix Core Web Vitals on the pages buyers use most.

Check who your robots.txt lets in

CrawlerRun byWhat it is for
GooglebotGoogleCrawling for Google Search, including AI Overviews and AI Mode
OAI-SearchBotOpenAISurfacing websites in ChatGPT search results
GPTBotOpenAICollecting content that may be used to train OpenAI's models

OpenAI controls each of its crawlers with a separate robots.txt setting. That means you can block GPTBot from training on your content and still allow OAI-SearchBot, so your pages can appear in ChatGPT search. Perplexity and Microsoft publish their own crawler names, so check their documentation before you block anything unfamiliar.

Hit the Core Web Vitals thresholds

Google's Core Web Vitals set three targets for a good experience, measured at the 75th percentile of page loads:

  • Largest Contentful Paint (loading): 2.5 seconds or less
  • Interaction to Next Paint (responsiveness): 200 milliseconds or less
  • Cumulative Layout Shift (visual stability): 0.1 or less

Compress images with a tool such as Squoosh before uploading, minify CSS and JavaScript, and use a CDN if your visitors are spread out. Create an XML sitemap and submit it in Google Search Console.

Add descriptive alt text to every meaningful image and use link text that says where the link goes. Keep text contrast readable on every background.

Action step: run your key pages through PageSpeed Insights. Fix the worst metric first, then re-test after each change so you know what moved it.

Pro tip: read the page aloud before publishing. If it sounds like you answering a customer, it is usually clear enough for AI search too.

Checklist of seven technical checks for AI search: crawler access, indexing, three Core Web Vitals thresholds, sitemap and alt text
Crawler access and indexing come first; the speed and accessibility checks help the people who click through.

How do you find the right keywords for AI search?

Start with the questions your customers already ask, in their own words. People put longer, conversational questions to AI assistants, so a page built around "How do I stop missing calls after 6pm?" fits better than one built around "call answering". Your Search Console queries, sales calls and inbox are the best sources you have.

In my experience, AI answers show up most on longer, question-style searches. Short head terms still matter for ordinary results, but the questions are where a direct answer gets lifted.

Turn questions into headings

Instead of "Email marketing benefits", write "How does email marketing keep customers coming back?". Then answer it in the first two or three sentences. AI Overviews are short, so a tight, direct answer is easier to lift.

Use short paragraphs, bullet lists or numbered steps for the detail, and a table for any comparison.

Action step: in Google Search Console, open the Performance report and filter queries with a custom regular expression such as ^(how|what|why|can|does|is) . Pick ten question queries and give each a clear, dedicated answer, either as a section on an existing page or a page of its own.

Pro tip: add a short FAQ to each service page with five to seven questions you hear on sales calls.

Build a cluster, not one-off pages

Google says AI Mode uses a query fan-out technique, breaking a question into subtopics and issuing many related searches at once. One buyer question therefore turns into several sub-questions, and each of them is a chance to be cited.

Map those sub-questions to one hub page that answers the main question and a few supporting pages that each answer one sub-question well. Link the hub and the supporting pages to each other, and cover the questions people really ask rather than every possible wording.

How do you earn citations in ChatGPT and other AI answers?

No AI company publishes exactly how it picks sources. In my own checks, the pages cited tend to answer the question cleanly and come from businesses that trusted sites already mention, such as trade bodies, local press and established directories. So you need two things: a passage worth quoting, and mentions in those places.

The writing side is covered in detail in my guide on how to write content that AI search engines actually cite. This section is about the reputation side.

Building mentions in the UK

  • Join a trade association or your local Chamber of Commerce and make sure the member listing links to you.
  • Claim and complete your listings on Yell and Thomson Local.
  • Write genuinely useful guest articles for regional business press and industry sites.
  • Keep your name, address and phone number (NAP) identical everywhere.

ChatGPT search relies on what OAI-SearchBot can reach, so the robots.txt check in the technical section applies here too. Perplexity shows its sources beside each answer, which makes it easy to see which sites it already trusts in your sector.

Action step: search your three main buyer questions in Perplexity and note every site it cites. Those are the places worth earning a mention.

Common mistake: chasing link volume. I would put effort into a handful of relevant mentions from respected UK sites before dozens of low-quality directory links.

How do you optimise for Google AI Overviews?

To rank a website in Google AI Overviews, make it eligible for ordinary Search, then make the answer easy to lift. Google's guidance is plain: "There are no additional technical requirements." It adds that specific optimisation isn't required, but all existing SEO fundamentals continue to be worthwhile. So the work is indexable pages, a short answer under each heading, then lists, tables and steps.

Format for the Overview

Open each section with a two or three sentence definition or answer. Follow with a bullet list or numbered steps. Add a comparison table where the reader is choosing between options.

Check the Overview exists first

Before you reformat anything, check whether your query shows an AI Overview at all. Search it signed out in a private window, on mobile and desktop, and note what appears and who is cited.

Read Search Console carefully

Google counts clicks and impressions from AI features within the overall web search figures in the Performance report. Queries with rising impressions and falling clicks can be a sign an Overview is answering on the results page.

Action step: reformat your best-performing post with definition paragraphs, lists and a table. Then compare its Search Console figures for the next 30 days with the 30 days before.

Why do video and audio content matter for AI search?

AI answers are mostly written text, so text is the easiest material for them to quote. A video with a full transcript, a podcast with proper show notes and an infographic with descriptive alt text all give them words to work with. A video with no words around it leaves a text answer very little to quote, however good the video is.

Make every format readable

  • Publish a full transcript with each video, on your page and on YouTube.
  • Write detailed podcast show notes that cover each point discussed.
  • Give infographics alt text that states the point the graphic makes.
  • Use descriptive file names: "employee-onboarding-checklist.jpg" beats "IMG_1234.jpg".

On YouTube, phrase titles as the question the video answers. Write a proper description and add timestamps so the video is split into chapters.

Build one piece, then reuse it

Start with one strong article. Turn its key points into a short video, cut the audio into a podcast episode, and make one infographic. Link all of them back to the article so every format points at the same answer.

Action step: record a five to ten minute video from your most popular post and upload it with a full transcript.

How do you track AI search performance?

Track three things: whether AI answers name or cite you for your key buyer questions, how many visits arrive from AI assistants, and how Search Console impressions and clicks move. No single tool shows all three. A simple monthly routine with a spreadsheet beats a dashboard nobody opens.

A monthly check

  1. List 10 to 20 questions a buyer would ask before choosing a business like yours.
  2. Ask each one in ChatGPT, Perplexity and Gemini, and search it in Google signed out.
  3. Log whether you were named, whether your site was cited, and which competitors appeared.
  4. Run each question twice, because answers vary from one run to the next, and count it as named if you appear in either run.
  5. Compare with last month and note which pages you changed.

Separate AI referrals in GA4

In Google Analytics 4, create a custom channel group with a rule that matches referral sources such as chatgpt.com, perplexity.ai, gemini.google.com and copilot.microsoft.com. It only counts people who click through, never citations without a click, which is why the manual check still matters.

My guide to AI visibility tracking covers the paid tools, what they measure and when they are worth it.

Worked example (illustrative)

Say a firm tracks 10 questions across three assistants: 30 checks a month. In month one it is named in 6 checks, which is 20%. After rewriting four service pages answer-first, it is named in 8 checks in month two and 12 in month three, which is 40%. These figures only show the arithmetic; they are not a client result.

Common mistake: setting it up once and forgetting it. Answers change as models and indexes update, so the sheet is only useful if someone reads it every month.

Line chart, illustrative: a firm named in 6 of 30 AI answer checks in month one, 8 in month two and 12 in month three
Illustrative example of a monthly tracking sheet: 10 questions across three assistants, not a client result.

How do UK local businesses rank higher in AI search?

Nobody outside the AI companies knows exactly what local AI answers weigh, but the sensible bet is the signals local search already uses: a complete Google Business Profile, a consistent name, address and phone number across directories, recent reviews, and pages that say clearly which towns you serve. Your profile, your Yell and Thomson Local listings and your own site should all say the same thing.

Google Business Profile first

Complete every section, choose accurate categories, add photos regularly and post updates. Ask every happy customer for a review, and reply to the ones you get, good and bad.

Back it up on your site

Add LocalBusiness schema to your homepage with your address, service area and opening hours. Build a page for each town you genuinely serve, with detail a visitor would find useful: who you work with there, typical jobs and how fast you can get out. A page that only swaps the town name helps nobody.

Write about local issues your customers face. "How London's ULEZ affects delivery businesses" reaches a specific reader with a specific question, which is the kind of query AI answers handle well.

Action step: spend an hour on your Google Business Profile this week. Add ten photos and ask five recent customers for a review.

What trust signals do AI systems look for in a website?

No AI company publishes a trust checklist, so I work on the evidence any careful reader looks for: a real, accountable business behind the page. That means HTTPS, a clear privacy policy, full contact details, named authors with real credentials, and reviews and mentions elsewhere that match what your site says about you. Most of these are cheap to fix.

The basics on your own site

  • HTTPS on every page, with no mixed-content warnings.
  • A privacy policy written for UK GDPR, saying who you are and how to contact you.
  • Full contact details and an About page: physical address, phone, email and real people.
  • Named authors with credentials on every article.

Proof that lives elsewhere

Keep your name, address and phone number consistent across Google Business Profile and directories, and collect recent, genuine reviews steadily. Show certifications and memberships only where you hold them, such as Cyber Essentials, ISO 27001 or a trade body. For B2B, keep your LinkedIn company page and founders' profiles current.

Action step: if you process customer data, check on the ICO website whether you need to pay the data protection fee, and update your privacy policy to match how you actually use the data.

Common mistake: letting reviews dry up. A steady flow of recent reviews says more than a burst from three years ago.

Checklist of seven website trust signals for AI search, from HTTPS and a UK GDPR privacy policy to consistent NAP and genuine reviews
Each one shows a real, accountable business stands behind the page.

How to rank in AI search: what should you fix first?

Start with three jobs: rewrite your top pages so each section answers first, check that robots.txt lets Googlebot and OAI-SearchBot in, and start the monthly tracking sheet. Mentions, local listings and video come next. None of this needs a special tool or file, and every step also helps ordinary search.

Where I can help

AI Agency Plus is an automation agency, so the part I build is the routine around this work. For example, I can build an n8n workflow that runs your buyer questions on a schedule, logs each answer to a Google Sheet and flags when a competitor is named. It saves the manual checking, and because answers from an API can differ from what people see in the ChatGPT app, I pair it with a regular hand spot-check.

On a Discovery Call we look at your site, agree which pages to fix first, and work out whether automating the tracking is worth it for you. If a monthly manual check covers it, I will tell you so.

Book a Discovery Call

Frequently asked questions

Does AI-generated content rank well in search?

It can, if it is genuinely helpful. Google's published guidance from 08/02/2023 says it rewards high-quality content however it is produced, and that using automation mainly to manipulate rankings breaks its spam policies. AI drafts that someone with real expertise checks, corrects and adds examples to can do well. Unedited bulk content rarely gives anyone a reason to cite it.

How do I get my website to appear in AI search results?

Make sure your pages are indexed and crawlable, including by OAI-SearchBot for ChatGPT search, then answer real buyer questions directly under question headings. Earn mentions on UK sites your sector trusts and keep your business details consistent everywhere. Check monthly which questions you appear for and rewrite the pages that are missing.

What is AI search optimisation?

AI search optimisation is the work of making your content easy for AI assistants and Google's AI features to find, understand and cite. It builds on SEO and adds answer-first structure, crawler access for AI search bots, and tracking of citations alongside rankings. Many people call it generative engine optimisation, or GEO, and my GEO guide covers the definition in full.

How long does it take to see results from AI search work?

Expect weeks rather than days. Google says crawling can take from a few days to a few weeks after you ask it to recrawl a page, and its AI features draw on the same index. You can confirm the new version is indexed with URL Inspection in Search Console. I judge changes over a full quarter of monthly checks, because single answers vary too much to read anything into one week.

Can small businesses compete in Google AI Overviews?

Yes. Google's bar for being eligible as a supporting link is being indexed and eligible for a snippet, and Ahrefs found in March 2026 that only 37.9% of cited URLs also appeared in the first 10 results. A small firm that answers a narrow question better than anyone, with real local and practical detail, gives an AI system a good reason to cite it. In my view, depth on your niche is a better bet than thin coverage of everything.

Book a Discovery Call