There is a pattern behind almost every AI Overview citation, and once you see it you cannot unsee it. AI search does not reward good writing in the way the old web did. It rewards structure. It cites content that is organised, specific, and verifiable, and it passes over content that is merely well intentioned.
For law firms competing to be quoted by Google AI Overviews and by assistants such as ChatGPT and Gemini, that distinction is the whole game. The clearest way to understand it is to look at a piece of legal-sector data built the way AI search likes to read.
AI does not cite prose. It cites structure.
When a search system assembles an answer, it is not looking for the most elegant paragraph. It is looking for a fact it can lift with confidence and attribute to a source. A specific figure, a clear definition, a step in a process, a number tied to a place and a date — these are extractable. A page of reassuring narrative, however polished, usually is not.
This is why two firms with similar expertise can have completely different AI visibility. One has written its knowledge down as structured, answerable content. The other has written a brochure. The search system can only quote the first.
A worked example: LawBoard
Consider LawBoard, a legal careers platform built on Solicitors Regulation Authority data. It holds verified profiles for nearly 9,000 law firms across England and Wales, and it runs a free salary estimator covering practice areas and regions.
Look at the shape of that, not just the subject. The data is organised by entity (the firm), structured by clear dimensions (practice area, region, seniority), grounded in an authoritative source (the SRA register), and exposed at scale across thousands of consistent records. You can browse the full firm directory and see the same structure repeat across every profile.
That is, almost to specification, what AI search is designed to pull from. When someone asks “what do conveyancing solicitors earn in Hampshire”, a source that holds salary data structured by practice area, region, and seniority is a strong candidate for citation. It can answer the exact question, with a number, attributed to verifiable data. A blog post musing on legal salaries in general cannot.
The same principle applies to your firm’s website
You do not need a directory of 9,000 firms to benefit from this. Every law firm site already contains the raw material AI search wants — it is usually just written in a form the system cannot use.
Your service pages, fee guidance, team profiles, and FAQ content are all potential citation sources. The question is whether they are written as answers. If your conveyancing page explains what the process involves, what it typically costs, how long it takes, and what a client should ask before instructing, that is content an AI assistant can cite. If it is three paragraphs of generic reassurance followed by a contact form, there is nothing to quote, and the citation goes to a competitor who did the work.
The practical translation of the LawBoard example looks like this:
- Attach real figures where you can — typical fee ranges, timelines, success measures — rather than describing services in the abstract.
- Structure content around specific questions, with clear question-led headings and direct answers underneath.
- Give named solicitors structured profiles with credentials, specialisms, and SRA registration, so the firm reads as a set of verifiable experts.
- Mark up FAQ content with FAQPage schema so the structure is explicit to the machine, not just visible to the reader.
Verifiability is the part most firms skip
The reason the LawBoard model works is not only that it is structured. It is that the structure rests on something checkable. Profiles built on SRA data can be trusted because the underlying register can be trusted. That verifiability is exactly what AI systems lean on when they decide whether to repeat a claim.
For a law firm, this is where structure meets E-E-A-T. Real data, named and credentialed authors, SRA registration, and recognisable authority signals are not decoration. They are the reasons a search system is willing to cite you on legal questions, which sit squarely in the “your money or your life” category where machines are most cautious. Invented figures and anonymous content fail this test, and rightly so.
The six signals behind a citation
Structured, verifiable data is the foundation, but it works alongside a handful of other factors. In short, the firms most likely to be cited combine strong existing organic rankings, question-led content, FAQ schema, named author credentials, genuine authority signals, and sound technical performance. I have set out how each one works, with the evidence behind it, in my analysis of Google AI Overview visibility for UK law firms.
The underlying lesson is the one LawBoard demonstrates at scale. AI search rewards content that is organised like data and grounded like evidence. Most law firm websites are neither, which is precisely why the opportunity is still open.
Where to start
Pick one practice area and rebuild it the way LawBoard builds a profile: specific, structured, verifiable, and answer-shaped. Add real figures, structure the page around the questions clients actually ask, mark it up properly, and put a credentialed solicitor’s name to it. Then watch whether it starts to appear in the Search Console AI report. That single page will teach you more about AI citation than any amount of theory.
If you would rather have the whole site assessed against how AI systems decide what to cite, that is the work I do for UK law firms — start with a SRA-compliant AI visibility audit, or book an intro call.
