Brand in AI answers

To an AI system a brand is an entity: a name, a location, a field, a set of relationships, and sources that agree about all of them. Joe Ashta, a GEO / AEO consultant for law firms, works on the mechanism by which such an entity becomes the default answer to a buyer's question. This page describes that mechanism, separates consistency from independence from corroboration, and states what a firm can control and what it can only earn.

What a brand is to an AI system

The previous page, From SEO to brand, argued that generative search names entities and so pushes search work toward brand work. To an AI system a brand is a record. It holds a name and the aliases that name goes by. It holds a location. It holds a field: the practice areas, the kind of client. It holds relationships: founded by, works at, member of, published in, based in. And it holds the sources from which each of those facts was learned, with a sense of how many sources say the same thing.

When the facts agree, the entity is sharp and the system names it with confidence. When the facts conflict, or exist only on the firm's own pages, the entity is blurred and the system names something else or names nothing.

The ladder from retrieval to default answer

A brand becomes the default answer in stages. Each stage responds to different evidence.

• Retrieval. The firm's pages, or pages about the firm, are relevant enough to enter the set the system reads for a question. This is grounding, and the buyer never sees it.
• Ranking. Among the retrieved pages, the firm's are strong enough on fit and authority to be used rather than discarded. A citation is the visible trace of this stage.
• Recognition. The system connects the pages to one entity and can state who the firm is. This is where a mention becomes possible. It depends on the same relationships being stated the same way in every record.
• Corroboration. Those relationships appear on sources the firm does not control, and the more authoritative and independent the sources, the stronger the stage.
• Dominant association. The evidence for one entity outweighs the evidence for its competitors so consistently that the system prefers it by default. This is where a recommendation comes from.

Most firms with a website have reached retrieval. The firms an AI system names first have reached dominant association for the questions that bring in clients.

Consistency, independence, corroboration

Three words get used as if they meant the same thing. They do not, and the ladder above depends on the difference.

Consistency is necessary and is the easiest of the three, because the firm can do it alone. Independence is a property of the source. Only corroboration moves a claim from asserted to established. One consequence follows: every domain a firm owns counts as one origin. A law firm with a main site, a blog on a second domain, and a lawyer's personal site has stated its facts once, three times over. Consistency across them helps. Corroboration has to come from somewhere else.

What a firm controls and what it earns

A firm controls three things. Its own naming: one legal name, one short name, a fixed set of aliases, used identically. Its structured data: schema.org markup on its own pages stating the name, location, field, and relationships in a form a machine reads directly. And its pages: content plain and specific enough that a system can quote it in an answer to a real question. All three are within the firm's power.

A firm can only earn the rest. Directory records, bar records, press, profiles on platforms it does not own, and mentions by other people are the sources that corroborate. The firm can make itself easy to describe correctly, supply the facts, and do work worth writing about. The record itself has to be written by someone else. This is the point where search work and reputation work stop being separable.

How Joe Ashta applies this to himself

This site is built on the same rules it describes. The entity record states the facts about Joe Ashta once: role, market, location, company, aliases, and boundaries. The sources page lists every record about him that sits outside this site, marks who controls each one, and separates independent records from owned pages. Every other page links back to those two and states his role the same way. The site publishes its own evidence so a reader can check whether it has been corroborated.

Next: the discipline

Doing all of this, page by page and record by record, is a discipline with a name. Generative Engine Optimization, or GEO, operationalizes the ladder: it treats retrieval, grounding, citation, mention, and recommendation as outcomes that can be produced and measured. The next page defines it.