Five layers of AI search visibility

The problem

Large language models only surface brands they can disambiguate. If a model cannot tell you apart from three similarly named competitors, or cannot tell whether you sell to SMB or enterprise, in the US or Europe, to developers or marketers, it will default to the brand it can resolve. Ambiguity is the single largest cause of AI invisibility.

Identity is the foundation layer because every layer above it inherits the confusion. No volume of content, citations or schema compensates for a brand the model cannot name with certainty.

What we work on

• Canonical brand definition, a single sentence the model learns
• Audience and ICP disambiguation by segment, geography, persona and use case
• Category placement, the “we are an X that does Y” statement
• Wikidata entity creation and claims architecture
• Disambiguation against look-alike brands
• Consistent identity propagation across owned surfaces

Primary service · Entity & Knowledge Graph

The problem

Your brand does not own how it is described in AI answers, the collective internet does. When analysts, journalists, reviewers, forums and competitors use a different vocabulary from your marketing site, the model averages across them and lands closer to the consensus than to your messaging.

Reinforcement is the act of getting third-party vocabulary to match your canonical vocabulary, without writing a word of the third-party copy yourself.

What we work on

• Messaging and vocabulary audit across owned versus earned sources
• Phrase-level drift detection across review sites and forums
• Earned-media placement with vocabulary-aligned angles
• Analyst and podcast briefing kits carrying canonical phrasing
• Community narrative work where your category is actually discussed
• Customer-quote and case-study language normalisation

Primary service · Branding & Positioning Content

The problem

Every frontier model is trained on a measurable subset of the internet, crawl corpora, reference works, a short list of high-authority publishers, code repositories, technical communities and academic indexes. If your brand is absent from that corpus it is structurally absent from the model’s memory, regardless of how much you publish on your own domain.

Distribution is the ground-truth layer. It determines whether you exist in the weights at all.

What we work on

• Training-corpus presence audit, platform by platform
• Reference-work notability and citation architecture
• High-authority publisher placement across trade and national press
• Open-data and technical contributions where the category supports them
• Transcript distribution to indexable surfaces
• Syndication strategy for long-tail authority accumulation

Primary service · LLM Citation & Visibility

The problem

Retrieval-augmented systems, which increasingly dominate AI search, do not read your prose, they read your structure. Pricing, specifications, founding date, headquarters, integrations, supported platforms, certifications: every factual claim needs to sit in a format a retrieval system can extract in milliseconds.

Data is the layer where most B2B sites lose citations without ever knowing it.

What we work on

• Structured-data deployment across JSON-LD and Schema.org
• Fact-sheet pages built for passage-level extraction
• Comparison tables, feature matrices and pricing grids
• FAQ schema and question-answer pair engineering
• Product and catalogue feeds for agent access
• llms.txt and AI-crawler-readable content infrastructure

Primary service · Technical Audit & Schema

The problem

Frontier models discount sources they judge low-credibility, thin content, affiliate-driven pages, sites with contradictory claims across pages, and authors without verifiable expertise. A model may know you exist and know your facts and still decline to recommend you, because the confidence signal is too low.

Integrity is the layer that separates “mentioned” from “recommended”. It is the difference between being in the corpus and being the answer.

What we work on

• Experience, expertise, authoritativeness and trust audit
• Author bio, credential and bylined-expertise architecture
• Claim-consistency auditing across owned pages
• Primary-source attribution on every factual claim
• Review-ecosystem hygiene across the platforms buyers check
• Original research published to be cited

Primary service · AI Brand Monitoring