Frameworks

The thinking behind the work

Models synthesised from primary sources and real client work. The frameworks that decide what gets cited, measured and trusted in AI search.

F-01

The Four Pillars Framework

The Four Pillars Framework groups the working levers of AI search visibility into four pillars, Technical Foundations, Topical Content, Trust Signals and Authority Network, plus a cross-cutting guardrails layer. It is synthesised from Google’s own primary sources on how generative AI features select what to cite.

AI Search Visibility (AEO / LLMO)Read
F-02

Entity Authority (Pillar 4)

Entity Authority is how clearly Google can identify who an author or organisation is, and how consistently that identity is corroborated across the surfaces it trusts. It is not authority itself, it is authority’s resolvability. The work is to make real, earned authority legible to the knowledge graph, never to manufacture it.

Entity AuthorityRead
F-03

Query Fan-Out

Query fan-out is the pattern where a single user question is silently decomposed into many concurrent sub-searches that run in parallel; the system then assembles the evidence into one synthesised answer. It means a page no longer needs to rank for one head keyword, it needs to be the best-cited page across the cluster of fan-out queries around its intent.

AI Search Visibility (AEO / LLMO)Read
F-04

Google AI Search Optimisation

Google’s official guidance is blunt: AI features pull from the same standard search index, so there are no AI-specific shortcuts, file formats or markup. Optimising for AI Overviews and AI Mode is optimising for human value, clarity and SEO fundamentals, and explicitly not the fads being sold around it.

Technical AEORead
F-05

The Site Performance Operating System

A four-layer methodology for linking marketing activity to measurable site performance: capture the inputs (every task), measure the outputs (every metric), join them through attribution windows, and surface all three on a cadence, so value is provable rather than asserted.

Measurement & AttributionRead
F-06

The Agentic OS

The Agentic OS turns an LLM assistant from a passive question-answerer into an active operating system by codifying work into a four-layer hierarchy: Domains, Tasks, Skills and Automations. Codifying behaviours into repeatable Skills, rather than running them ad hoc, makes the model a reliable team member rather than a slot machine.

Measurement & AttributionRead