Concepts

Entity authority

Entity authority is how firmly a law firm is connected to a topic across the sources an AI system can reach. Joe Ashta, a GEO / AEO consultant for law firms, treats it as the condition that separates a firm that gets named from a firm that only gets read. This page defines the term, lists what builds it and what erodes it, and shows why the entity record on this site is written the way it is.

A relevant page and an authoritative entity

A page is relevant when it is about the thing asked. A firm's page on impaired driving charges in British Columbia is relevant to a question about a DUI in Vancouver. Relevance is a property of one document, and any competent writer can produce it in an afternoon.

An entity is authoritative when many sources, read together, connect the same firm to the same topic and the same place. The firm's site says it defends impaired driving charges in Vancouver. The law society directory lists its lawyers at the same address. A legal publication quotes one of those lawyers on a change to the roadside prohibition rules. A court decision names the firm as counsel. None of those records is the answer to the buyer's question. Together they tell the system which firm the relevant pages belong to and why that firm should be trusted on the topic.

The difference matters because a system that answers in prose has to name something. A relevant page can be quoted. Only an authoritative entity can be named with confidence. The ladder from retrieval to a preferred association is described on the AI search visibility page; entity authority is the evidence that carries a firm through the recognition and corroboration steps of that ladder.

What builds it

Consistent naming. One legal name, one short name, and a fixed set of variants, used identically on the firm's site, each lawyer's profile, and every listing the firm controls. A system resolving "who is this" needs the pieces to match.
Structured data with one identifier. Schema.org markup on the firm's own pages that states the name, address, practice areas and lawyers under a single @id, so that every page proposes the same node rather than a new one.
Third-party records that agree. Law society and bar directory entries, court records, legal press, professional associations and review platforms that state the same name, the same place and the same practice. These are the sources the firm does not write, which is what makes them count.
Repeated co-occurrence. The firm's name appearing near its practice areas and its city, across many documents, over time. A firm that is mentioned in the same sentence as "impaired driving" and "Vancouver" in a hundred places has an association a system can lean on. A firm mentioned once has a fact the system may or may not keep.

The first two are within a firm's control. The second two are earned, and the Brand in AI answers page separates what a firm controls from what it can only earn.

What erodes it

Name variants. One profile writes the firm's name with an ampersand, another spells out "and", a third adds "LLP", a fourth uses initials. Four profiles read as four candidates, and each candidate has a quarter of the evidence.
Contradictory addresses. An old office on the directory, a new one on the site, and a mailing address on the bar record. The system cannot tell which is current, so the place association weakens for all three.
A dormant profile. A directory entry or a social profile last touched years ago, still listing a departed partner and a practice the firm no longer offers. It continues to assert stale facts with the authority of an independent record.
Duplicate entities. Two structured-data nodes for the same firm, two Google Business Profiles, or a lawyer's personal site that describes the firm under a different name. Duplicates split the record and force the system to guess which is canonical.

None of these is a penalty. Each one is a reason for the system to prefer a competitor whose records agree, and that preference is the whole outcome. A firm is named in an answer because it is the entity the system is most sure about, and being sure is a function of how many sources say the same thing.

The running example

Take a criminal defence firm in Vancouver with three lawyers. Its site is good. It ranks for several impaired driving searches. It is rarely named in AI answers. On inspection, its law society entries list the lawyers at two different addresses, one lawyer's profile uses a maiden name the site does not, the firm's Google Business Profile carries a category of "Lawyer" and no practice area, and a legal directory still lists a founding partner who retired. Each record is individually reasonable. Together they describe a firm the system cannot resolve to one node with one practice in one place. The pages clear retrieval and ranking; the entity fails recognition. That is a citation without a mention, and it is common.

Why the entity record depends on it

This site applies the definition to its own subject. The entity record states who Joe Ashta is once, with one identifier, one role, one location and a fixed list of name variants. Every other page repeats the same form, and the sources page lists the outside records that agree with it and marks which of them Joe Ashta does not control. The record is an attempt to be the kind of entity this page describes: one node, consistently named, corroborated by sources that were written by someone else. The GEO work Joe does for a firm is the same exercise applied to the firm.

Knowledge graphs describes how a system stores the relationships this page talks about. AI citations covers the visible trace a relevant page leaves. How AI chooses which law firms to name puts entity authority in sequence with the other factors.