You searched your firm's name and the right-hand panel on Google showed a competitor, an old address, or a mashup of your firm and a similarly named one in another state. Or you asked ChatGPT about your firm and it described a practice you do not have. In both cases the underlying problem is the same: the systems do not have a clean, single idea of who you are.
Entity SEO is the work of fixing that. It is less about keywords and more about making your firm and your attorneys unmistakable across every source a search engine or language model might read.
What Google (and a language model) thinks your firm is
Google's Knowledge Graph stores entities: things with a name, a type, attributes and relationships to other entities. A law firm entity has a type (law firm), a location, founders or partners, practice areas, a founding date, and links to the same firm on other sites. Google builds the record by reconciling many sources, and it only surfaces a knowledge panel when it is confident the record is coherent.
Language models learned about your firm the same way, from the text that existed about it. If the text is consistent, the model has a confident, narrow idea of you. If your firm has been "Miller & Associates," "Miller Law Group" and "The Miller Firm" across a decade of directory listings, the model has three fuzzy entities and will hedge or blend them.
The AEO for lawyers hub covers the full picture of AI search; this page is about the identity layer underneath it.
Knowledge panel or Business Profile panel: which one you are looking at
Two different panels appear in the same spot, and firms often confuse them. The local panel, with the map, hours, reviews and "Call" button, is driven by your Google Business Profile. You control it directly in the Business Profile dashboard.
The knowledge panel, with a description, founding date, "People also search for" and social links, comes from the Knowledge Graph. You do not edit it directly; you can claim it (there is a "Claim this knowledge panel" link at the bottom when Google recognizes an official presence) and then suggest edits, which Google may or may not accept.
A branded knowledge panel typically appears when there is enough independent, consistent coverage (bar directories, news, legal directories, sometimes Wikipedia) for Google to be sure. Chasing a knowledge panel for its own sake is rarely worth it; making the entity clean is worth it regardless, because it also affects how ChatGPT and similar tools describe you.
The five signals that make an entity unambiguous
A single canonical name. Pick the name as registered with the bar and use it everywhere, including the Business Profile, the site footer, schema legalName, and every directory. If you market under a shorter brand, put both in the schema (legalName and alternateName) and use the same pair everywhere.
One address format. "Suite 200" versus "Ste. 200" versus "#200" is a trivial difference to a human and a reconciliation problem for a machine. Choose the USPS form and copy it exactly. This is the core of citation and directory work, and it is tedious, and it matters.
Consistent practice area language. If the site says "car accidents," the bar profile says "motor vehicle torts" and Avvo says "personal injury," a model has to infer that these are the same thing. Use the same three to six practice area phrases everywhere.
Explicit links between sources. Your site's schema markup should carry sameAs links to the bar profile, Business Profile, Avvo, Justia, LinkedIn and any Wikipedia entry. The loop of links is what lets a system confirm the sources describe one entity.
An About page that reads like a reference entry. State the founding year, the founders, the office locations, the practice areas and the bar admissions in plain declarative sentences. This is the page Google and models pull descriptive text from, and most firms fill it with adjectives instead of facts.
Attorneys as entities, not just the firm
Each lawyer is an entity too, and often the more searched one. A client searches "Jane Doe attorney Richmond" more often than the firm name. Give each attorney a bio page with Person schema, bar admissions with years, education, memberOf for bar associations, and sameAs to the state bar record, Avvo and LinkedIn.
Common names need extra disambiguation. If there are four attorneys named Michael Brown in your state, the bio should anchor to the specific firm, city, law school and admission year, and the schema worksFor should point at the firm's @id.
Fixing a confused entity (mergers, renames, common names)
Entity problems most often come from change. A firm renames after a partner leaves, merges with another, or moves offices, and the old facts persist in a hundred directories. The fix is unglamorous: an inventory of every listing, corrections submitted one at a time, an About page that explains the change explicitly ("Founded in 2009 as Smith & Jones; renamed Smith Law in 2021"), and schema that includes the former name as alternateName for a while.
If the local panel shows a duplicate or a merged listing, use Business Profile's suggest-an-edit and ownership tools. If a knowledge panel shows wrong facts, claim it and submit corrections with sources. Expect weeks, not days, for changes to propagate. Google Maps optimization work usually runs in parallel, because the Business Profile is the single most influential source for the local entity.
Questions we get
Do we need a Wikipedia page?
No, and trying to create one for a typical firm usually fails Wikipedia's notability rules and gets deleted, which looks worse than never having one. Wikidata entries are less strict and can help, but bar directories, legal directories and consistent citations do most of the work.
Why does Google show a firm in another state under our name?
Because the two entities share a name and Google could not separate them confidently. Strengthen every distinguishing attribute: full address, state bar links, attorney names, and a distinct description. Over time the records usually split.
How long before ChatGPT describes us correctly?
Search-connected answers can update within weeks of your public sources changing. The model's built-in knowledge updates only when a new model version is trained, which you do not control. Fix the sources and let both catch up on their own schedules.
If your firm has been renamed, merged, or simply described wrong, tell us what you are working with and we will map out the cleanup.