London marketing teams are being sold three overlapping stories: “we’ll get you to the top of Google,” “we’ll get you into ChatGPT,” and “we’ll make your brand an AI entity.” Used loosely, those phrases all mean “more traffic.” Used properly, they are different work packages with different artefacts and different failure modes.
ASN Global sells all three because they stack. We do not sell them as synonyms. This piece is the briefing we give a marketing director or professional-services partner before they fund a year of “AI SEO.”
SEO: permission to be in the index
Search engine optimisation is still the floor. If Google cannot crawl, index, and understand a URL, most AI features that sit on Google’s index will not use it either. The 2026 basics have not gone away: canonical URLs, sitemap, robots, unique titles and descriptions, internal links, Core Web Vitals, and content that matches intent (Source: Google Search Central, 2024).
SEO fails when the site is a single vague homepage, when landing pages never enter the sitemap, or when query-string duplicates compete with real URLs. It also fails when local businesses refuse to name London or the UK on the page because it “feels less premium.” Models and crawlers cannot infer a market you will not state.
GEO: permission to be quoted
Generative Engine Optimization is page craft for answer engines. The unit of success is a passage, not a rank position. We structure headings as questions, put the answer early, add FAQs that humans can actually open, attach FAQPage and Article schema that matches the visible text, and write definitions a model can lift without inventing a third service you do not offer.
GEO fails when teams hide meaning in hero videos, when every H2 is a metaphor, or when schema describes a LocalBusiness that does not exist. It also fails when AI crawlers are blocked “for security” on a public marketing site. You cannot be quoted from a document the retriever is forbidden to fetch.
Independent explainers of GEO as a discipline emphasise extractability, citations, and machine-readable context rather than keyword density theatre (Source: Search Engine Land / industry GEO explainers, 2024–2025). That matches what we see in implementation: the brands that get cited have boring, specific pages.
Search Discovery: permission to be named
Generative Engine Search Discovery is brand-level work. The unit of success is the organisation, not a blog URL. We make the entity consistent across Organization schema, About, service catalogue, FAQ answers, and llms.txt. We monitor prompts such as “who should I hire for GEO in London” and check whether the description is accurate even when we are not recommended yet.
Discovery fails when the company uses five trading names, when services only exist as homepage cards with no URL, and when there is no public FAQ a model can treat as policy. It also fails when leadership wants recommendation without allowing any proof into the public web. Assistants are reluctant to name a vendor they cannot corroborate.
How the three should share a budget
For a typical UK professional-services firm with an existing site:
- Foundation (once): technical SEO + schema + crawl policy + sitemap completeness
- GEO (quarterly): service and insight pages rewritten for extractability, FAQ growth, article schema, internal links
- Search Discovery (ongoing): entity lock, llms.txt generated from live content, prompt set, off-site corroboration that is real
If the site cannot be crawled, do not buy content. If the entity is confused, do not buy more blogs. If the entity is clear but thin, then fund citable articles. Sequence is cheaper than doing everything as a rebrand.
What London buyers should ask an agency
- Which URLs are canonical for each service, including GEO and brand discovery?
- Are those URLs in the XML sitemap?
- Is there an
llms.txtgenerated from live FAQs and articles, or a static brochure file? - Does FAQ schema match visible questions, and are there enough questions to cover how buyers actually prompt?
- How will you measure recommendation as well as rankings?
- What will you refuse to do — fake citations, invisible text, invented addresses?
If the agency cannot answer those without a slide titled “synergy,” keep looking.
Where ASN Global sits
We design and engineer the site so SEO, GEO, and Search Discovery are the same information architecture. Web design, AI agent development, and search are not three disconnected retainers that fight in the CMS. A production agent that answers from your own FAQs should be reading the same corpus that Perplexity might cite. That is the point of a digital partner rather than a plugin vendor.
For the detailed service descriptions, see Generative Engine Optimization and Generative Engine Search Discovery, or the FAQ if you want the extractable version.
Ready to split the work properly instead of buying “AI SEO” as a mood? Talk to ASN Global.
Sources
- Google Search Central, documentation on crawling, indexing, and AI features, 2024–2026, https://developers.google.com/search/docs
- Search Engine Land and related industry publications on Generative Engine Optimization, 2024–2025
- Nielsen Norman Group, research on how users scan headings and extract answers, ongoing — relevant to passage-level GEO