Search engines and AI systems don't score a single website in isolation. They build a picture of an entity — a person, a company, a brand — from evidence scattered across the whole internet, and then decide how confident they are to recommend it. Authority is that confidence. We build it across five distinct types, because each answers a different question the AI is silently asking.
Is this a real, verifiable entity, and is it the same one everywhere I look?
Is it legitimate, safe and endorsed by independent third parties?
Do real, credentialed people stand behind it who know this subject?
Has it produced original, useful, citable material of its own?
Do many independent sources reference it — so no single claim stands alone?
Platforms are just the vehicles. The real objective is strengthening these signals. New platforms appear and old ones fade — the signals don't.
The type of authority: credibility of the real humans behind the brand. People trust people before they trust companies — and AI now models named experts as entities in their own right.
Answers: "Is there a real, qualified expert accountable for this?"
The type of authority: the legal and commercial legitimacy of the business as an entity. Built once — every website the company owns benefits from it.
Answers: "Is this a real, registered, reputable organisation?"
The type of authority: the technical quality, trust and topical relevance of each individual website. If a company owns five sites, each one earns this layer independently.
Answers: "Is this specific site well-built, trustworthy and machine-readable?"
The type of authority: the original, useful evidence a brand produces itself. This is where long-term authority compounds — and what AI systems most love to cite.
Answers: "Has this entity created something original worth quoting?"
The type of authority: the web of independent third-party references that corroborate everything else. This is where AI confidence compounds — because no single claim has to stand on its own.
Answers: "Do lots of sources I already trust independently point to this?"
Every layer above builds signals. This one proves they work — and it's what separates a methodology from a checklist. We track two halves:
Server-log evidence that GPTBot, ClaudeBot, PerplexityBot and Google-Extended are actually fetching your pages. If they never arrive, nothing downstream matters.
Firing real buyer questions at ChatGPT, Perplexity, Gemini, Claude and Copilot and recording whether you're mentioned, cited and named ahead of competitors — over time.
The message to a client stops being "we submit you to directories" and becomes "we measure and grow your recommendation confidence."
Most agencies confuse company authority with website authority. They're two separate but connected systems. The company is the entity; the websites are assets it owns.
Person Authority
│
▼
Company (built once)
│
┌──────────┬──────────┬──────────┬──────────┐
▼ ▼ ▼ ▼ ▼
Website A Website B Website C Website D Website E
│ │ │ │ │
Technical Technical Technical Technical Technical
Content Content Content Content Content
Authority Authority Authority Authority Authority
The company builds authority once. Each website builds its own authority independently — different audience, content, backlinks, research and PR. Together they reinforce the entity.
The type of authority: presence on the platforms where your tools, apps or datasets actually live. Powerful if you build products — skip if you don't.
Every action strengthens one or more of the signals AI systems use to decide whether your people, company, websites and content deserve to be recommended. This is future-proof: platforms change — signals don't. Clients buy a durable strategy, not a list of websites.