How to get ChatGPT to recommend your law firm

We ran our readiness check across 182 sites and found a useful warning for firms seeking AI recommendations: the basic plumbing is usually in place, while the evidence a model needs to make a confident choice is thinner. That means a firm can be easy to crawl and still hard to recommend for a specific legal matter in a specific place. Organization schema passed on 113 sites, AI crawlers were allowed on 175, and 178 homepages answered a plain request. The weaker checks concerned a plain statement of what the firm does, comparison information, and whether buyer questions were answered on the page.

What should you expect ChatGPT to use for a local legal question?

Start with the shape of the question, not with your homepage. Legal buyers usually combine a place, a matter, a type of client and sometimes a practical constraint. A generic page saying that your firm provides expert legal advice gives a model little to work with. A directory entry that states the firm, office location and practice area plainly is easier for a model to lift as fact than a page full of marketing language.

There is no form where you submit a firm and get a reliable recommendation. ChatGPT assembles an answer from available sources, and the source may be a directory, a bar association, a local organisation, a news page or your own site. For local legal queries, the directory or bar listing can carry the identity and location facts, while your site supplies the detail about the work.

The citation data is a separate dataset, and it is not a law firm sample. In the pages recorded in our research on who owns AI answers, listicles and roundups account for 68.3% of citations, while competitor product pages account for 24.1%. The product's own site accounts for 1.7%. Do not turn those shares into a forecast for your firm, but do take the practical point: the answer may be built from a page you do not control.

You can still improve the chances. Keep the firm name, office details, practice areas and professional credentials consistent wherever you are listed. Ask the directory or bar association to correct old descriptions. On your own site, create pages that answer the exact local questions a buyer would ask, then make those pages easy for a crawler and a human to read.

How do you make outside listings support your firm?

Treat every important listing as a source that needs checking, not as a citation you can order. Look for the pages that already name your firm and the pages that appear for your target local questions. Compare the firm name, office location, areas of law, languages, client type and links. A mismatch, such as one listing saying you handle employment disputes while another only says commercial law, gives the model no easy way to decide which description is current, so it may default to the safer, vaguer answer.

Your own site should reinforce the same facts in plain language. Put the location and service area on the relevant practice page as well as on a contact page. Say who the page is for. Explain the matter you handle and the limits of the service. A useful page does not need to sound like a brochure. It needs to leave fewer gaps between the buyer's question and the firm's answer.

Our readiness data shows why this work is often more valuable than another round of small metadata edits. The check for a plain statement of what a site is, for whom, at what price, and against what passed on 33 of 182 sites, warned on 96 and failed on 49. Questions answered on the page passed on 99 and warned on 79. Comparison or alternatives pages passed on 91, warned on 50 and failed on 41. A law firm will not use product alternatives in the same way, but the underlying need is familiar: the page must answer the choice the buyer is trying to make.

Use a local question as the page brief. For example, a page for a particular city and landlord dispute should state the service area, the kind of dispute, the people the firm represents, the first step, and where the information does not apply. It should link to the relevant lawyer or office details. It should not force the reader, or an AI system, to infer the answer from a string of vague claims.

Here is a short procedure you can complete in under an hour:

1. Write down local legal questions your buyers actually ask, including the place and matter.
2. Search each question and record the directories, bar listings and other pages that appear repeatedly.
3. Check every important listing for the same firm name, location, practice area and destination link.
4. Open the matching page on your site and mark each fact the buyer still has to infer.
5. Rewrite the opening section so it states who you help, where you work and what matter you handle.
6. Run the changed pages through the free AI readiness check and save the result.

This process reduces avoidable ambiguity and gives the pages already used in local answers better facts to carry. It still cannot make an independent directory or bar listing accurate when its owner has not corrected it.

What can you fix on your site before you ask for a recommendation?

Fix the pages that answer a buyer's question, rather than starting with firm-wide claims. Give each important page one clear subject and one clear H1. Explain the service in text that remains available without JavaScript. Add the canonical and Open Graph tags, publish a sitemap, and make sure robots.txt does not block the relevant AI crawlers. These are simple checks, but simple is doing useful work here.

The sample shows a mixed picture. One clear H1 check passed on 151 of 182 sites. Content was present without JavaScript on 146. Canonical and Open Graph tags passed on 140, and a sitemap was published on 153. AI crawlers were allowed in robots.txt on 175. Those passes help a system reach and interpret a page, but they do not persuade it that the page is the best source for a narrow legal question.

The larger gap is the firm's statement. A page can have solid metadata and still avoid saying what the firm does, for whom, where and in what circumstances. Labels help a crawler classify the page. Specific sentences give it material for an answer.

A worked example makes the distinction clearer. Imagine a Bristol firm's landlord-dispute page, then review it line by line:

1. Page subject: The H1 says “Landlord dispute solicitors in Bristol”. The one-clear-H1 check passed on 151 of 182 audited sites, so this is basic hygiene rather than a differentiator.
2. Technical access: This page already clears the common access checks, with its text available without JavaScript and the relevant page signals in place.
3. Opening sentence: The page begins, “We offer expert legal advice.” It never says whether it is for tenants, landlords or both. That is the plain-statement weakness described above, so rewrite it as, “We advise Bristol landlords and tenants on possession claims, disrepair disputes and deposit issues.”
4. Buyer questions: Add direct answers about the first step, documents to gather and when the firm cannot help. The questions-on-page check passed on 99 sites and warned on 79, so this is a common gap.
5. Source choice: The firm's own page may still not be the page cited in an answer. In the readiness check, the site's own pages were cited for 25 of 182 sites. Check the relevant local directory and bar listing as well, then make their descriptions agree with this page.

The sequence gives you a practical order of work: make the page reachable, give it a clear subject, state the useful facts, answer the buyer's questions, then check which source appears in the result. A law firm's strongest citation path may still begin with a trusted directory or bar listing, so improve those sources alongside your site.

The chart below shows where the 182 sites pass, warn or fail on each readiness check.

Do not bury practical details. If you publish a pricing or consultation page, link to it from the relevant service page. In the wider readiness sample, price visibility or a linked pricing page passed on 145 of 182 sites. Legal fees are often conditional, so the useful point may be how an initial assessment works, what affects cost, and when a buyer should contact the firm. State what you can state, and label the boundary clearly.

How do you know if ChatGPT is actually recommending your firm?

Measure the questions, sources and locations separately. A single prompt such as “best law firm” is too broad to guide a local practice. Build a small set around the places you serve and the matters you want, then keep the wording stable while you check. Include questions that name a town, a legal problem and a client type. Record whether your firm is mentioned, which page is cited, and whether the source is your site, a directory or a bar association listing.

The free AI visibility checker can help you inspect the answer and its cited pages. The live check of which AI engine a site is built for is useful when the same page behaves differently across engines. Use the methodology behind the measurement when you need to explain internally what the check does and does not establish. The point is to turn a vague visibility score into a list of pages and facts you can change.

A useful report has a question, place, engine, date, mention outcome, cited URL and source type. It should show when a directory listing appears without your site, when your site appears without a useful practice page, and when neither appears. This lets you choose the next action. Correct the listing, improve the page, add a local question, or leave the query alone because it is outside the firm's work.

Do not treat one answer as proof. Our data comes from samples, and every sample has a margin of error. We cannot vouch for what we haven't measured. The readiness results come from 182 audited sites, while the citation shares come from the pages recorded in our research sample, not from all legal queries or all AI systems. A result can change with the wording, location, engine and source set.

That limit makes regular checks more useful, not less. Keep a small baseline of local questions and rerun it after a listing correction or substantial page change. Compare the cited source and assess more than whether the firm name appeared. If the same directory is repeatedly cited, that page deserves as much attention as your own homepage.

For ongoing work, review the weekly research on who owns AI answers and use the leaderboards only as a way to find patterns worth investigating. Rankings are not recommendations, and a strong general position says little about a specific legal matter in a specific place. Your useful unit is the local question and the page that answers it.

ChatGPT is more likely to recommend your law firm when trusted local listings establish who and where you are, while your own pages answer the specific legal question clearly enough to verify.