AI visibility is not a new channel you can switch on
A founder searches their company category in Google, asks ChatGPT for suitable providers and sees competitors mentioned instead. The tempting response is to buy a “GEO package,” add more schema or publish hundreds of question pages.
That treats the symptom. Search and AI-answer systems first need access to the site. They then need enough clear, consistent evidence to understand the business and decide whether a page helps with a specific question. If the website uses vague category language, hides proof in images or makes claims that no other source supports, more content will not repair the underlying trust problem.
Google's current guidance says the standard foundations of SEO still apply to AI Overviews and AI Mode. There are no additional technical requirements, special AI schema or machine-readable AI files required. A page must be indexed, eligible for a search snippet and useful to the person asking the question. Google's AI features guidance.
OpenAI likewise says no business can guarantee top placement in ChatGPT Search. A public site can be considered when OAI-SearchBot is allowed to crawl it and the host or content-delivery network permits that traffic. OpenAI's ChatGPT Search guidance.
My verdict: build a verifiable business knowledge system before investing in an “AI optimisation” campaign.
The Business Discoverability Stack
I use five layers to separate a visibility problem from a positioning, evidence or measurement problem. Work from the bottom upward; later layers cannot compensate for a blocked or confusing foundation.
1. Access: can machines reach the evidence?
Check that the important service and resource pages return successfully, appear in the sitemap, have a valid canonical address and are not blocked by robots or a security layer. Google advises keeping important content in textual form and making it discoverable through internal links. For ChatGPT summaries and snippets, OpenAI says OAI-SearchBot must not be blocked. Access creates eligibility, not a ranking promise.
2. Entity clarity: can the business be identified?
The homepage, About page, service pages, author profiles and external business listings should agree on the company name, specialist, audience, locations served and core offer. Describe the business in customer language before adding labels such as GEO or AIO. A CEO should understand the answer to three questions within seconds: who is this for, what commercial problem does it solve and why should I trust the provider?
3. Useful answers: do pages help with a real decision?
Build pages around distinct customer decisions: how much to invest, which channel fits, why leads are weak, how to enter a market or how to select a provider. Give the answer early, explain trade-offs and include a framework a decision-maker can use. Google recommends original, complete, people-first content rather than large volumes of pages created mainly for search traffic. Google's people-first content guidance.
4. Corroborated proof: can claims survive verification?
Use attributable case evidence, named expertise, customer reviews, relevant partnerships, independent coverage and accurate business profiles. Make the limits of evidence clear. Structured data can describe visible facts, but it cannot turn an unsupported claim into authority. Consistency across credible sources reduces ambiguity; repetition across pages on the same domain does not create independent proof.
5. Learning loop: does discovery create commercial value?
Review the questions that lead people to the site, which pages are cited, whether the business is described accurately and which visits become suitable enquiries. Update pages when the offer, evidence or buyer language changes. Do not refresh a date without a material improvement, and do not chase a single chatbot response as if it were a fixed ranking.
Diagnose the first weak layer before creating more content
| Business symptom | Likely first problem | Best next decision | Avoid |
|---|---|---|---|
| Important pages are absent from Google | Access or index eligibility | Check robots, status, canonical, sitemap and Search Console | Publishing more pages before fixing discovery |
| The company appears, but is described incorrectly | Entity clarity | Align core facts and explain the offer consistently | Adding markup that contradicts visible copy |
| Competitors are cited for customer questions | Answer usefulness or authority | Create one definitive decision page with original judgement | Copying competitor headings into a generic article |
| The site ranks, but AI tools rarely cite it | Evidence and corroboration | Strengthen authorship, proof and reputable third-party presence | Buying unverified mentions or mass backlinks |
| Visibility rises without qualified enquiries | Buyer fit and conversion path | Connect the page to the right service, proof and next step | Reporting mentions without commercial context |
This matrix prevents an expensive category error. A crawl problem needs technical repair. A vague positioning problem needs clearer language. A trust problem needs stronger evidence. A conversion problem needs a better path from answer to business decision.
Start with the AI growth systems page if customer insight and AI-assisted workflows are the main gap. If attention already exists but revenue does not, use the Traffic-to-Revenue Diagnostic.
Measure discoverability as a business system
Index eligibility, sitemap coverage and crawl access are healthy.
Name, offer, audience and evidence appear without material errors.
Priority customer questions lead to relevant business content.
Visits and referrals progress to suitable conversations and revenue.
Google includes AI-feature traffic within the Web search type in Search Console rather than providing a separate AI report. ChatGPT adds utm_source=chatgpt.com to referral URLs, according to OpenAI's current publisher guidance, so those visits can be analysed separately. OpenAI publisher FAQ.
For AI-answer monitoring, keep a small, stable set of high-value buyer questions and check them regularly. Record mention, description accuracy, cited URL and material changes. Treat the result as directional evidence because answers vary by query wording, context, location and product.
A practical 30-day discoverability plan
Confirm access
Check the priority URLs, robots rules, sitemap, canonical tags, text visibility and internal links.
Clarify the entity
Align company, founder, audience, offer, locations and proof across the core website and reliable profiles.
Strengthen decision pages
Improve the three pages closest to revenue with direct answers, original frameworks, evidence and a useful next step.
Build the learning loop
Connect Search Console, analytics, CRM outcomes and a controlled AI-answer review to one monthly decision report.
Use growth partnership services to see how paid acquisition, conversion, measurement and AI fit together. Review the case studies for evidence-led operating examples, and read about Thomas for the practitioner behind this framework.
Frequently asked questions
How can my business appear in AI search results?
Make the website publicly crawlable, publish clear pages about the customers and problems you serve, support claims with verifiable evidence and keep your business facts consistent across reputable sources. For ChatGPT search, do not block OAI-SearchBot. Eligibility helps discovery, but no platform guarantees inclusion or top placement.
Is GEO different from SEO?
GEO, or generative engine optimisation, focuses on how a business is represented and cited in AI-generated answers. The useful work overlaps heavily with sound SEO: crawlability, clear content, credible authorship, internal links and reliable evidence. Google explicitly says no additional optimisation is required for its AI features.
Do I need special AI schema or an llms.txt file?
Not for Google AI Overviews or AI Mode. Google says there is no special schema or AI text file required. Structured data should still match the visible page because it helps systems understand eligible content. An llms.txt file can document a site, but it should not replace crawlability, internal links or useful pages.
Should I create hundreds of pages for AI search questions?
No. Creating many lightly differentiated pages can dilute quality and create overlap. Build one useful page for each distinct customer decision, then strengthen it with first-hand judgement, supporting evidence, clear authorship and internal links. Expand only when a new page answers a genuinely different business need.
How should I measure visibility in Google and AI search?
Use Google Search Console for Google impressions, queries and clicks; analytics for visits and conversions; and controlled recurring prompts for directional AI-answer monitoring. Record whether the business is mentioned accurately, whether the correct page is cited and whether resulting visits create qualified enquiries. Do not treat one chatbot answer as a stable rank.
How long does it take to become discoverable?
There is no guaranteed timeline. Crawl and indexing can take days or longer, while earning authority and corroboration usually takes sustained work. Establish access and entity clarity first, publish the highest-value decision pages next, then review discovery and enquiry evidence monthly rather than expecting immediate visibility from a technical change.
Be understandable before trying to be everywhere
Google and AI search cannot reliably represent a business that is difficult to access, vague about its offer or unsupported by evidence. Build the discoverability stack in order: access, clarity, useful answers, corroborated proof and measurement. Then expand content only where a distinct customer decision and a credible route to revenue exist.