How to get your business recommended by ChatGPT.

Start with two things: being trusted over the internet and making sure your site structure is machine readable. Look at where your competitors are being cited across the industry, then look at your own website's messaging, service pages and case studies. Your content needs clear, context-rich data about your business.

Thomas Howcroft, interview, 26 October 2026 · edited for length

I look at the first image and the first sentence. Does it answer the questions of people searching for those services straight away? Do people know exactly what they're looking at and what kind of website they've landed on? If not, we've got a problem there and we need to fix it. People quickly scan rather than read, so the offer needs to be easy to understand.

Thomas Howcroft, interview, 26 October 2026 · edited for length

I start with websites relevant to the company's industry. For a creative consultancy or agency, I would look at creative awards, creative software websites and useful creative forums. For a builder or painter, I would look at relevant trusted-trader or supplier websites and any accreditation they genuinely qualify for. I would also work on getting genuine, detailed customer reviews on relevant platforms, such as Google, Trustpilot or trusted-trader websites.

These are suggestions, not accreditations TomHow or a client is claimed to hold. Relevance, a listing or a review does not guarantee an AI citation.

Thomas Howcroft, interview, 27 October 2026 · edited for clarity; editorial clarification follows

Make the areas you serve clear on your website and keep your business details consistent across your website, Google Business Profile and other trusted listings. NAP means name, address and phone number: wherever you publish those details, they should be accurate and consistent. I would also describe the business and its service areas in schema markup that matches the visible website.

A business address and a service area are different. Serving Stevenage does not mean having premises there. Do not invent an address or publish a private home address to create consistency. Keep a service-area business profile configured appropriately. Schema describes the facts; it does not ensure inclusion in AI answers.

Thomas Howcroft, interview, 27 October 2026 · edited for clarity; editorial clarification follows

OpenAI's help page on ChatGPT search describes no submission form. It says a website becomes eligible for inclusion when it allows OAI-SearchBot to crawl it and its host or content delivery network allows traffic from OpenAI's published search-bot IP addresses. It also says placement is not guaranteed.

Source: OpenAI Help Center [4]

OpenAI does not list a Google ranking as a requirement. It says ChatGPT search ranks results using multiple factors and sometimes sends rewritten queries to partner search providers, such as Microsoft. The groundwork overlaps with SEO: pages search systems can crawl, content in plain text and credible references elsewhere.

Source: OpenAI Help Center [4], Google Search Central [6]

A mention and a citation are separate outcomes. A mention names your business; a citation links to a source, which may be a directory, a list article or a competitor's page. Semrush found that only 21% of the most-cited domains in a ChatGPT category were also its most-mentioned brand, so record both rather than treating them as one score.

Source: Semrush [3]

ChatGPT can rewrite your question into different searches, use your approximate location and, with memory switched on, details you have shared before. Record the date, tool, mode and location with every check, ask the same questions in fresh conversations, and compare like with like.

Source: OpenAI Help Center [4]

OpenAI says it can take about 24 hours from a robots.txt update for its search systems to adjust. Allowing OAI-SearchBot makes a site eligible; it does not guarantee that a page will be shown or recommended.

Source: OpenAI [5], OpenAI Help Center [4]

Not by itself. Google's systems decide which pages answer related questions, and Google stopped showing FAQ rich results on 7 May 2026. Write each answer so it makes sense on its own, put the direct answer first, and keep any FAQ markup word for word with the visible text.

Source: Google Search Central [8], Google Search Central [7]

SEO focuses on search visibility. AEO focuses on answer extraction. GEO focuses on generative engine mentions and citations. LLM setup makes pages clearer for language models. AI search setup measures the whole surface with prompts, citations, competitors, and search data.

From the earlier setup guide on this page

Start with measurement and crawl access: GA4, Search Console, robots.txt, sitemap, indexation, and manual page testing. Then fix content, schema, llms.txt, internal links, and prompt tracking.

From the earlier setup guide on this page

Yes. The first setup is mostly clarity, crawl access, page structure, schema basics, and measurement. The hard part is doing it carefully and keeping the evidence honest.

From the earlier setup guide on this page

No. Google ignores it for Search visibility and rankings, including its generative features. Other systems may ignore it too. It is optional public documentation, not a crawler access control, citation guarantee or substitute for useful linked pages.

From the earlier setup guide on this page

Check measurement weekly while recovering from a traffic drop. Check the full setup monthly, especially after new pages, redesigns, CMS changes, or major search movement.

From the earlier setup guide on this page

Google says there is no special schema.org markup required for its generative search features. Relevant structured data can describe visible business, author and page facts and support eligible search features, but it must match the page. Choose types for the actual content rather than adding every available type. Passing a validator shows a markup check, not a citation or ranking improvement. Any claimed movement needs separate dated observations.

From the earlier setup guide on this page

The setup is not five separate tricks. It is one clear system: make the site crawlable, make the offer understandable, make the answers easy to extract, make the entity consistent, and measure what happens.

A clean setup does not guarantee rankings or AI citations. It removes the obvious blockers and gives search engines, answer engines, and LLMs better evidence to work with.

Primary references

These are platform requirements, not measured TomHow results. Google eligibility does not guarantee inclusion. OpenAI documents search access separately from training access. Check the current guidance before changing crawler controls or buying an optimisation claim.

• Google AI features: eligibility, access and measurement Normal SEO, indexed pages and snippet eligibility; no special AI schema is required.
• Google generative AI optimisation guidance Useful original content and the limits of AI-specific files and formatting.
• OpenAI crawler roles and controls Search, training and user-request access have distinct roles.

Simple map

SEO helps search engines find, understand, and rank pages. AEO helps answer engines extract direct answers. GEO helps generative engines understand and mention the brand. LLM setup makes pages clearer for language models. AI search setup ties all of that to measurement.

• SEO: crawl, index, rank, click
• AEO: answer the question clearly
• GEO: earn mention and citation inside generated answers
• LLM setup: make the page easy to summarise
• AI search setup: track prompts, citations, and competitors

Step 1

Start with the boring checks. If important pages are blocked, missing from the sitemap, slow, broken on mobile, or hidden behind scripts, the rest of the work sits on bad ground.

• Clean robots.txt
• Readable sitemap
• Important pages indexable
• Mobile pages usable
• Important content visible as text

Step 2

A strong page has one purpose. It tells the buyer what the service is, who it is for, what problem it solves, what proof exists, and what to do next.

• One main topic per page
• Clear H1 and title
• Short plain introduction
• Specific service detail
• Strong next step

Step 3

Answer the reader question directly, then explain the conditions, example and limits. Question headings are useful when they match the topic. Google does not require special AI formatting or a fixed chunk size. A short paragraph should earn its length by helping the reader, rather than imitate an extraction formula.

• Plain question headings
• Direct first sentence answers
• Short explanatory paragraphs
• FAQ where useful
• Definitions for buyer terms

Step 4

Generative engines need to know who the business is and why it belongs in the answer. Repeat the same name, service category, founder or company identity, location, audience, and offer across trusted pages.

• Consistent business name
• Consistent service labels
• About page with real operator detail
• SameAs links where relevant
• Third-party profiles using the same wording

Step 5

Use relevant structured data for visible facts and applicable search features. Google requires no special schema for AI Overviews or AI Mode. Do not invent reviews, authorship, locations or services. Markup validation is separate from indexing and citation measurement.

• Organization facts when relevant
• Person details with accurate authorship
• Service facts matching the offer
• Article details matching the content
• Visible breadcrumb relationships

Step 6

Google says it ignores llms.txt for Search visibility and rankings. Other services may choose to read it, but adoption should not be assumed. Maintain it only if a public-page summary is useful; prioritise the pages and normal internal links first. An avoid list does not control crawler access or protect private material.

• Optional public business summary
• Working links to useful public pages
• No private data in the file
• Separate robots and indexing controls
• No promised ranking or citation effect

Step 7

AI search measurement starts with buyer prompts. These are the questions a buyer might ask ChatGPT, Claude, Gemini, Perplexity, or Google before shortlisting a provider.

• Shortlist prompts
• Comparison prompts
• Problem prompts
• Audit prompts
• Pricing and report prompts

Step 8

A good setup has a before-and-after loop. Record GA4, Search Console, indexed pages, prompt checks, competitor mentions, citations, and what changed on the site.

• GA4 sessions
• Search Console clicks and impressions
• Indexed key pages
• AI citation checks
• Shipped work log

Good setup

A clean setup is easy to inspect. The site has crawlable pages, clear service explanations, answer blocks, schema, llms.txt, internal links, external corroboration, and a measurement loop.

• Search engines can crawl and index the important pages
• Buyers can understand the offer quickly
• Answer engines can extract direct answers
• LLMs can summarise the entity without guessing
• Prompt and search data are checked monthly

Reality check

No setup guarantees rankings, traffic, leads, AI Overviews, or ChatGPT citations. The point is to remove weak foundations and create better evidence for search systems to use.

• No guaranteed rankings
• No guaranteed AI citations
• No instant traffic promise
• No fake visibility score
• No one-file fix

If this page describes the problem you have, the next useful step is a dated audit with prompts, competitors, technical checks, and a signed action plan.

• Paid audit
• AI search services
• GEO services
• AEO services
• Sample report
• Recovery plan
• Resource library
• Tools
• Zero traffic recovery plan
• How to measure AI search visibility
• Schema generator
• llms.txt builder
• What llms.txt does and does not do