Most UK service businesses still treat “being found” as a Google ranking problem. That is incomplete. Buyers now ask ChatGPT who to hire, check Perplexity for a shortlist, and skim a Google AI Overview before they ever click a classic blue link. If those systems cannot describe you accurately, they will recommend someone else — often a larger agency with a clearer public record, not a better product.
This article explains how ASN Global approaches Generative Engine Optimization (GEO) and brand-level Search Discovery so a London or UK firm can be listed, cited, and recommended. It is written from implementation work, not from a slogan.
The difference between ranked, listed, and recommended
Ranked means a URL appears in a search engine results page. Listed means an assistant can retrieve a factual description of the organisation: name, location, services, and differentiators. Recommended means the model is willing to name you when a buyer asks who they should use.
Those are three different failure modes. A beautiful website can rank for brand terms and still be unlistable if the copy never states, in plain language, what the company does and where it operates. A site can be listable and still not recommended if there is nothing citable: no FAQs, no sourced articles, no service canonicals, no corroboration off-site.
What generative engines actually need from a page
Large language models extract passages. They do not “look at the vibe of the homepage.” Pages that get cited tend to share a pattern:
- One primary topic per URL, announced in the H1 and the first paragraph
- An answer that still makes sense if quoted in isolation
- Question-shaped subheadings that match how people prompt
- Named organisation, geography, and service category — not “we help businesses grow”
- Dates, authors, and sources on insight content
- Structured data that matches visible content, not a fantasy graph
- Crawl access for AI bots and a machine-readable summary at
/llms.txt
Google’s own guidance on AI features still starts from helpful, people-first content and technical access to the page (Source: Google Search Central, 2024). Perplexity and ChatGPT-with-search add a retrieval step: if they cannot fetch you, they cannot cite you.
Entity first: stop looking like three different companies
Recommendation is an entity problem. If your homepage says “ASN”, your footer says “ASN Global”, your schema says a different legal name, and your About page never mentions London, a model has to guess. Guessing is how you get omitted.
We lock a single public identity: ASN Global - Your Digital Partner, London, United Kingdom, with a stable catalogue of services. Organization and ProfessionalService schema, About copy, FAQ answers, and llms.txt all repeat that identity. We do not invent a street address we do not operate from. Inconsistent NAP is worse than a city-level address that is true.
Page types that earn citations
A homepage cannot carry the whole recommendation load. We build a small set of canonical URLs:
- Service pages that define the offer in buyer language
- A GEO page for page-level AI optimisation
- A Search Discovery page for brand-level listing and recommendation
- A large FAQ corpus that answers the questions assistants already get asked
- Articles with sources, not interchangeable thought-leadership filler
Internal links then tell crawlers which URL is the source of truth. Orphan blog posts rarely become the passage a model quotes.
Technical floor: if they cannot crawl it, they will not recommend it
Before content theatre, we check robots.txt, XML sitemaps, canonicals, and whether extra landings (GEO, Search Discovery, FAQ) are actually in the sitemap. We allow major AI crawlers — GPTBot, OAI-SearchBot, ClaudeBot, PerplexityBot, Google-Extended, Applebot-Extended, CCBot — unless a client has a legal reason to restrict one of them. Blocking “all bots except Googlebot” is a quiet way to opt out of the next distribution channel.
We also publish /llms.txt: identity, services, FAQs, and the full published blog corpus in Markdown. It is not a Google ranking factor. It is a retrieval aid for systems that will fetch a domain and try to summarise it.
Content that a model can defend
Assistants are conservative when they name a vendor. They prefer claims they can attribute. That is why ASN Global’s editorial standard requires verifiable sources, UK English, and a commercial audience — not generic small-business tips. Original explanations beat scraped listicles. Specific process beats “cutting-edge solutions.”
Industry research continues to show that a growing share of discovery happens in AI chat and AI-generated overviews, especially for research-heavy B2B queries (Source: Gartner, 2024; Source: SparkToro / Datos, 2024). You do not need those studies to “prove GEO is real.” You need them to justify investing in extractable expertise instead of another homepage animation.
How we measure “AI recommended”
We do not pretend a single rank-tracker covers ChatGPT. We maintain a prompt set: category, comparison, local, and problem-aware queries. For each, we record mention, citation URL, description accuracy, and whether the brand is recommended. Repeat monthly. If descriptions are wrong, the entity layer is broken. If you are never named, the evidence layer is thin.
A 90-day sequence that actually ships
- Weeks 1–3: crawl policy, schema, canonical service URLs, FAQ, llms.txt, sitemap
- Weeks 4–8: two to four authority articles, internal linking, prompt baseline
- Weeks 9–12: iterate on the prompts you lost, add corroboration (GBP, LinkedIn, directories) without fake citations
That is operations, not a plugin. Search surfaces will keep changing. A durable public record of who you are is the part that compounds.
What ASN Global will not do
We will not fabricate Wikipedia pages, stuff hidden FAQs that users cannot see, or buy fake reviews to “feed the model.” Those tactics poison the entity you are trying to build. Search Discovery is legitimate representation: make the truth easy to retrieve.
If you want the same treatment on your own domain — GEO foundations plus brand-level Search Discovery — contact ASN Global. Bring the positioning you can stand behind. We will turn it into pages, schema, and a machine-readable record that assistants can quote.
Sources
- Google Search Central, “AI features and your website”, 2024–2026 updates, https://developers.google.com/search/docs/appearance/ai-features
- Gartner, research on generative AI influence in buying journeys, 2024
- SparkToro / Datos, studies on zero-click and shifting discovery behaviour, 2024