How law firms can improve their visibility in ChatGPT

Citation, mention, recommendation: three different results

When a partner asks me “are we in ChatGPT?”, my first answer is a question: in which way? ChatGPT can surface a law firm three ways, and they’re not worth the same.

Many AI visibility reports blend all three into one score. I report them as separate lines. A firm with lots of citations and no recommendations has a different problem from a firm that’s named but never cited, and the fix is different too.

How ChatGPT decides which law firms to name

ChatGPT picks firms and lawyers based on trust and authority, and it builds that picture from signals it finds around the web. Are you talked about? In articles and publications, on video, on podcasts, in reviews, in directories? The more places that say the same true thing about your firm, the easier it is for an AI system to repeat it.

The sources ChatGPT leans on change with the kind of question. This is the pattern I see in testing:

Two more things from testing. Each platform pulls from slightly different sources, so ChatGPT, Gemini, Perplexity and Google’s AI answers can disagree about the same firm. And the sources change constantly. I re-check them instead of relying on an old map.

One thing is fixed: ChatGPT can only cite pages it can reach. OpenAI’s search crawler is called OAI-SearchBot, and OpenAI’s documentation says sites that block it won’t be shown in ChatGPT search answers. I check robots.txt early in every review, because a security plugin or an old blanket rule can block it without anyone noticing.

Why some firms show up in ChatGPT and others don’t

The question I hear most is some version of this: we rank in Google, so why does ChatGPT name our competitor? When I dig in, it’s usually one or more of these.

• The search was hyper-local. A firm can show up for a prompt about one neighborhood and disappear when the prompt widens to the county or the state.
• The firm is missing sources AI pulls from. Ranking in Google doesn’t mean the firm has the directory listings, news coverage or specific citations ChatGPT uses to check a firm.
• The firm’s own site doesn’t say what AI is looking for. Experience, practice areas, case types, location. If it isn’t on the site in plain words, AI has to guess or go somewhere else.
• The firm’s data disagrees with itself. An old firm name, an attorney who left, an address that changed two moves ago.
• The firm is quiet everywhere except its website. No video, no podcasts, no articles, few reviews. Nothing for AI to cross-check.

Entity recognition: does ChatGPT know who your firm is?

An entity is a distinct thing a search system can identify: your firm, each attorney, each office. Before I worry about prompts, I check whether AI can tell who you are and get the facts right.

I start with manual searches to see what AI sees. I’ve watched ChatGPT bring up a firm’s old name, or say an attorney works at a firm they left. That happens when data isn’t updated, or when the old name built up a lot of authority over the years. It takes a while to change. The new information has to outweigh years of old information across many sites.

What I check:

• The firm’s name, address and phone match across citations and directories. (A citation, in local SEO, is any listing of your firm’s name, address and phone on another site.)
• Schema on the site matches those citations. Schema is code that labels facts for machines: this is a law firm, this is its address, these are its attorneys.
• Attorney bios name the current firm, practice areas and background, and link to the attorney’s profiles elsewhere.
• Old firm names, former attorneys and closed offices are corrected anywhere the firm controls them.
• Social and directory profiles exist and say the same thing as the website.

This is where my semantic SEO work carries straight over. I moved into search intent and semantic SEO in 2015. Semantic SEO helps search systems understand how entities relate: this attorney works at this firm, this firm handles these cases, in this city. AI uses those relationships to connect the dots between the data it pulls from around the web. Your website is the one source of truth. Everything else should agree with it.

Website content: what each page does for AI

The most common gap I find is content. Firms don’t write the pages that give AI the information it’s looking for. Each page type has a different job:

Third-party sources: be where AI expects to find you

AI isn’t going away, and it expects an established firm to be visible in many places. Social profiles and third-party sites carry far more weight than they did years ago, because AI uses them to pull information about a law firm.

The places that matter most for law firms, in my experience:

• Legal directories and review sites. They feed the recommendation and experience parts of an answer.
• Google reviews. Volume, recency and what clients actually say about the case.
• News and publications. Local press, legal publications, articles your attorneys write or are quoted in.
• Video and podcasts. Another media type saying the same thing about who you are and what you handle.
• Social profiles. Even on platforms you don’t actively use, a complete and accurate profile is one more place your facts can be confirmed.

I call that last layer a “Social Fortress”: consistent, authoritative profiles on the platforms search engines and AI assistants expect an established brand to have.

The tradeoff is upkeep. Every profile is one more place your details can go stale. A handful of accurate profiles beats a long list of abandoned ones showing an old address. And Reddit is one place I don’t try to manufacture. Planted posts get removed, and it’s a bad look for a law firm.

Local relevance: location still decides who gets named

Location matters in ChatGPT, especially when someone asks for a local firm or attorney. Location pages and citations carry the local signals ChatGPT pulls from to check a firm’s details.

• Location pages with real local detail, schema and the office’s correct name, address and phone.
• Google Business Profiles filled out in full, each pointing to the right location page.
• Geographic backlinks: links from local organizations, news and businesses near each office.
• Consistent citations for every office, not just the main one.

Multi-office firms need this per office. If only the home office has strong signals, the firm can be recommended in one city and missing in the next. This is also why I test from the firm’s office location and not just from my desk.

How I test a law firm’s ChatGPT visibility

I don’t start with a dashboard. I start by searching the way a potential client would.

1. Build the prompt list from real demand

   I draw prompts from intake data, call recordings, search queries and practice area and city combinations. Most are unbranded, because that’s how new clients search.

2. Run the searches by hand

   I run the basic searches potential clients perform, logged in and logged out, and from the firm’s office location. Each setting can return a different answer.

3. Record how the firm appears

   For each prompt: cited, mentioned, recommended or absent. I also note which outside sources the answer drew on, because those are the gaps to close.

4. Check the other platforms separately

   Gemini, Perplexity and Google’s AI answers each pull from slightly different sources, so a win on one doesn’t carry to the others.

5. Set a baseline with the best data available

   AI visibility tools change constantly, so I don’t rely on one. I use the best data I can pull from each to see how the firm is performing.

6. Fix what’s missing, then re-run the same prompts

   Once I have a baseline, I work on the areas the firm is missing out on: entity data, site content, third-party sources or local signals. Then I run the same prompt list again so the comparison is fair.

Measuring ChatGPT visibility without fooling yourself

I tell every firm this up front: it’s hard to find and count every prompt a firm is mentioned or cited in. Nobody can see them all. So the goal isn’t a perfect count. It’s a consistent read on whether visibility is getting better.

What I report:

• Overall numbers from the tracked prompt list, split into citations, mentions and recommendations.
• The prompt list itself, showing which prompts the firm shows up for and which it doesn’t.
• Search Console queries. More people now search Google the way they talk to ChatGPT, in longer, full-question searches. When those queries grow and the firm starts appearing for them, it’s a useful signal. The queries also feed the next round of ChatGPT prompts.
• AI referrals in analytics. Visits that arrive from ChatGPT and other AI platforms. Links from ChatGPT often carry a utm_source=chatgpt.com tag, which makes them easier to separate.

None of this matters if it doesn’t turn into signed cases. Visibility is the start of the funnel, not the end of it. I’d rather see a firm add “ChatGPT or another AI tool” as an option when intake asks how a caller found them.

What law firms get wrong about ChatGPT visibility

• Treating Google rankings as proof. Rankings help, but ChatGPT checks a firm against sources Google rankings don’t cover.
• Skipping the content. This is the biggest one. Firms don’t give AI the information it’s looking for on their own site.
• Staying quiet outside the website. No signals from other media types or platforms means nothing for AI to cross-check.
• Trusting one blended score. A rising “visibility score” can hide the fact that you’re cited but never recommended.
• Calling one test a result. Answers change from run to run. One screenshot proves very little.
• Expecting entity problems to clear quickly. An old firm name with years of authority doesn’t disappear because you updated your website.

The honest limits of ChatGPT visibility work

I’d rather you hear these from me now than from a report later.

• Answers change. The same prompt can return different firms from one run to the next, and logged-in users can see different answers from logged-out ones.
• You can’t see every prompt. Any count of mentions or citations is a sample, not a census.
• The sources move. What ChatGPT pulls from this quarter may not be what it pulls from next quarter.
• Analytics undercounts. Some AI-influenced visits arrive looking like organic or direct traffic.
• Old data lingers. Correcting firm and attorney information is slow, and you can’t control every site that lists you.
• Nobody controls the output. Not me, not an agency. A guaranteed spot in ChatGPT is a promise about something no one owns.

One more condition where this advice stops working: if intake is losing calls, AI recommendations just send more people into the same leak. In my experience, most firms are shocked to learn how much of their marketing problem happens after the lead comes in. That’s a fractional CMO problem, not a ChatGPT problem.

Working with me on AI visibility

There are a few ways in, depending on how your firm is staffed.

• A review, run like a consultation. I check what AI sees about your firm, show you where you’re cited, mentioned, recommended or missing, and talk through your options.
• Ongoing work. If your firm isn’t staffed to do the work, I do it and keep improving your AI visibility over time.
• Part of a fractional CMO engagement. AI search is one of the pieces I own alongside intake, paid search and vendor oversight. How fractional CMO work runs.
• Direction for your agency. If you already have an agency, I make the recommendations and your agency does the work. You don’t have to switch vendors.

Questions law firms ask about ChatGPT