Law firms are better placed to win AI search than almost any other profession, and most have no idea why. The reason is regulatory. Since December 2018 the SRA has compelled firms to publish exactly the kind of structured, specific, verifiable information that AI engines need in order to recommend a business: prices, what is and is not included, key stages, likely timescales, and the qualifications of the named people doing the work. Most firms treat that as a compliance obligation, bury it in a PDF, and never think about it again. It is in fact the most valuable AI visibility asset the firm owns.
This guide covers what changes when clients ask an AI instead of Google, why the transparency rules are an advantage rather than a chore, how to audit your own AI visibility, and what to fix first.
Do clients actually ask AI for a solicitor?
Increasingly, yes - and the behaviour differs from search in a way that matters. Someone searching Google for a solicitor gets ten results and picks from them. Someone asking ChatGPT, Claude, Gemini or Perplexity gets a shortlist, often of one or two firms, with a short explanation of why. There is no page two. You are either in the answer or absent from it.
The queries are also more specific than keyword searches ever were. People no longer type employment solicitor London; they describe their situation in full - who their employer is, what happened, what they want - and ask which firm to approach. That rewards firms whose website answers detailed questions in plain language and punishes those whose practice-area pages say little beyond a list of services.
Why are law firms unusually well placed to win this?
Because the SRA already requires you to publish the data. Under Rule 1 of the SRA Transparency Rules, any authorised body that publishes the availability of certain services must publish cost information on its website covering:
the total cost of the service, or where that is not practicable, the average cost or a range of costs;
the basis of your charges, including hourly rates or fixed fees;
the experience and qualifications of anyone carrying out the work, and of their supervisors;
a description and cost of likely disbursements, with an average or range where the actual cost is unknown;
whether fees or disbursements attract VAT, and how much;
what is included in the displayed price, including the key stages of the matter and the likely timescale for each stage, and anything a client might reasonably expect to be included but is not;
for conditional fee or damages-based agreements, the circumstances in which clients may have to pay themselves, including out of damages.
Read that list again as a specification for an AI-citeable page rather than as a compliance rule. Specific figures. Named people with stated qualifications. Defined stages with timescales. Explicit scope boundaries. If you were designing from scratch the ideal page for a generative engine to quote when someone asks what an uncontested probate costs and how long it takes, you would produce very nearly this.
The rules apply where you publish that you work in these areas: residential conveyancing, uncontested probate, immigration excluding asylum, First-tier Tribunal immigration appeals excluding asylum, summary motoring offences at the Magistrates Court, and employment tribunal claims for unfair or wrongful dismissal brought by employees. For businesses: defending those employment tribunal claims, debt recovery up to £100,000, and licensing applications for business premises. They do not apply to publicly funded work.
What stops this working in practice?
Four failures, all common, all fixable.
1. The pricing lives in a PDF
A PDF is the single most effective way to hide compliant pricing from an AI engine. Some crawlers parse PDFs poorly, many skip them, and the content is detached from the practice-area page it belongs to. Rule 1.6 requires cost information to be clear, accessible and in a prominent place on your website - a position that happens to align exactly with what works for visibility. Put the figures in HTML, on the practice-area page itself.
2. The transparency page is orphaned
A single "Pricing" page listing every regulated service, linked only from the footer, satisfies a box-ticking reading of the rules and wastes the opportunity. One page per service, linked from the relevant practice-area page, gives each matter type its own URL that can rank and be cited independently.
3. The people are anonymous
The rules require you to publish the experience and qualifications of whoever does the work and of their supervisors. "Our experienced team" does not satisfy that, and it gives an AI engine nothing to attach authority to. Named solicitors, with admission dates, specialisms and real biographies, are how a model establishes that a firm genuinely does this type of work. This is also the clearest expression of E-E-A-T in any profession: you have regulated, verifiable credentials that almost no other industry can match.
4. Nobody has run an AI visibility audit
Most firms have never asked the engines what they report about the firm. In our audits the common findings are an old office address, a partner who left two years ago, practice areas the firm dropped, or - most often - no citation at all, with a competitor named instead. An AI visibility audit is not complicated, but it has to be done systematically: the same prompts, across all four engines, recorded and repeated, or you are reading noise.
How do AI engines decide a law firm is credible?
Broadly, they are asking whether consistent, verifiable evidence exists that this firm does this type of work in this place. Three things carry most of the weight.
Entity consistency. Firm name, address, phone number, SRA number and practice areas must agree across your website, Google Business Profile, the Law Society's Find a Solicitor, the SRA register, legal directories and your own LinkedIn. Disagreement is the most common reason an engine either gets a firm wrong or declines to name it. Rule 4.1 already requires your SRA number and the SRA digital badge in a prominent place on your website, which conveniently gives engines a verifiable identifier to anchor to - use the same number everywhere.
Specificity over claim. "Leading employment specialists" is unquotable. "Six employment solicitors; tribunal claims from £3,500 to £12,000 depending on complexity; typical timescale four to nine months" is quotable, and it is what the rules ask for anyway.
Reachability. A surprising number of firm websites block AI crawlers in robots.txt, usually without the partners knowing, often added by a developer worried about scraping. If ChatGPT and Perplexity cannot read your site, no amount of content will help.
Does structured data help a law firm get cited?
Yes, and schema is the most under-used lever in legal marketing. Structured data is a machine-readable restatement of facts already visible on the page, and AI systems lean on it heavily to work out what a page is actually about. Four schema types matter for a law firm:
LegalService or Attorney - confirms the business type, the practice areas and the geographic area served.
Person - attaches credentials, admission dates and supervisory responsibility to a named solicitor rather than to an anonymous firm.
FAQPage - marks up genuine client questions so the answers can be extracted whole, which is exactly the unit a generative engine wants to cite.
Offer or price markup on regulated-service pages - the structured-data counterpart of the costs information the rules already make you publish.
Schema is not a ranking trick and will not rescue a thin page. What structured data does is remove ambiguity, and ambiguity is the main reason an AI system declines to cite a specific law firm. A citation is an act of confidence: the engine has to be sure enough about who you are to put your name in an answer it is accountable for.
Does this work for law firms with several offices?
Yes, with discipline. Each office needs its own page with its own address, its own named solicitors and its own Google Business Profile, and those details must match the register. Where firms lose is listing five offices on one contact page: the engine cannot establish which solicitors sit where, so it either picks one at random or names a single-office competitor whose location is unambiguous.
Is any of this a regulatory risk?
Publishing clearer, more specific information about price, scope and people is the opposite of a risk - it is what the rules were designed to produce, and the SRA runs random web sweeps of firm websites to monitor compliance. Two cautions apply. Keep published figures current, because an out-of-date price is a compliance problem before it is a marketing one. And be careful with outcome-based claims: statements about success rates or results invite scrutiny under the Codes of Conduct in a way that factual statements about price, stages and qualifications do not.
How long does AI visibility work take, and what does it cost?
Entity and technical fixes - crawler access, consistent details across directories, the SRA number and badge, pricing moved out of PDFs - can register within weeks, and they are cheap because they are finite. Content and authority compound over months, which is exactly why they are defensible once established. Anyone promising that your firm will be named by ChatGPT by a particular date is guessing, because none of the engines publish their ranking criteria and all of them change.
The budget question is best answered by scale. Rewriting the regulated-service pages for a firm with three practice areas is a small project. Doing it across twelve practice areas and four offices, with directory consolidation and monthly monitoring, is a programme.
What should a law firm do first?
Ask the engines. Put your firm's name, then your practice area and town, to ChatGPT, Claude, Gemini and Perplexity. Record what each says, including what it gets wrong and which firms it names instead of you.
Check your robots.txt. Confirm GPTBot, ClaudeBot, PerplexityBot and Google-Extended are not blocked.
Move pricing into HTML, on the practice-area page, one page per regulated service.
Name your people, with qualifications, admission dates and supervisors, as the rules require.
Add schema for the firm, each solicitor and each regulated-service page.
Reconcile your details across the SRA register, Find a Solicitor, Google Business Profile and the legal directories.
Re-test monthly. Answers drift, and competitors are working on this too.
Common questions
Is SEO for law firms dead? No. AI engines are substantially built on the same signals - crawlable content, authority, consistent entity data, local relevance - so the work overlaps heavily. What has changed is the shape of the result: a shortlist instead of a page of links, which makes being second-best far less valuable than it used to be.
How is this different from ordinary SEO? SEO aims to rank a page. AI visibility, or GEO, aims to get your firm named inside an answer. In practice GEO puts more weight on entity consistency, structured factual statements and verifiable credentials, and less on keyword placement.
Can anyone guarantee we will be recommended for "solicitor" in our town? No, and a firm that guarantees it is telling you something about itself. What can be guaranteed is the work: the fixes, the monitoring and a clear record of what the engines said before and after.
Do we need to publish prices if we only do commercial work? The costs rules bite where you publish the availability of the listed services. If none of those apply, the transparency obligation is narrower - but the underlying lesson still holds, because specific published information about scope, stages and people is what gets a firm cited whether or not a regulator requires it.
See exactly where you stand
We audit what ChatGPT, Claude, Gemini and Perplexity currently say about your firm, where your details disagree across the web, and which firms are being recommended in your place. You get the findings and the fixes, not a PDF of problems.
