Law firms show up in ChatGPT and Perplexity recommendations when the engine can find a page that answers the person's exact question, confirm the firm's name, location and practice areas across directories, reviews and third-party mentions, and see a recent date on the page. We asked both engines this question 12 times a month for two months and logged every source they cited.
Data as of October 2026: September and October runs, 12 per engine per month, from our own citation tracker. We update this page after each monthly run.
How do law firms show up in ChatGPT recommendations?
Perplexity's answers to this question, read across all 12 October runs, keep returning to the same short list: pages that answer specific questions, consistent directory and review profiles, third-party mentions, structured data and recency. ChatGPT's answers cover the same ground in looser form. Here is each factor with what our own tracking shows about it.
1. Pages that answer one specific question
Both engines retrieve pages and then quote them. A page titled for a practice area, such as "Houston Car Accident Lawyer," competes with every firm in the city for the same few words. A page titled for a question a client actually asks, such as "Can I still recover if I was partly at fault for a wreck in Texas?", competes with very few pages and has a precise answer to give: under the Texas proportionate responsibility statute, an injured person can recover only if their share of fault is 50 percent or less, and the award is reduced by that share.
The sources tell the same story. Of the 10 pages Perplexity cited for this question in October, 7 carried both "law firm" and "ChatGPT" in the URL itself. The engine matched the question's own words. A firm that wants to be named for a question needs a page that asks it.
2. The same identity everywhere the engine looks
An engine is more willing to name a firm when several independent sources agree on who the firm is. It finds those sources on the state bar listing, Google Business Profile, Avvo, Justia, Martindale-Hubbell, FindLaw and Super Lawyers. Small drift adds up. If the bar lists the PLLC name, Avvo lists a shortened brand, and Google lists the brand plus "Injury Lawyers," the engine is looking at three candidates with thin evidence each instead of one firm with strong evidence.
The fix is clerical, not creative. Choose one legal name and one display name. Then make every profile match the website footer character for character, including the suite number, the phone number and the list of practice areas.
3. Third-party mentions and reviews
A firm's own website is one witness. Engines look for others: reviews on Google and Avvo, local news coverage, bar association pages, podcast appearances and community threads. The two engines differ sharply on that last one. Across the 54 consumer legal questions we track each month for two injury law practices, ChatGPT cited Reddit on 10 of them in October. Perplexity cited Reddit on none.
So a firm that only courts reviews may do well on Perplexity and still be missing from the threads ChatGPT reads. Answering questions in local and practice-area forums as a named attorney, within your state bar's advertising rules, is how a firm earns a place there.
4. Structured data that matches the visible page
Structured data, the schema markup in a page's code, tells a machine what the page says without making it guess. For a law practice that means LegalService markup, which schema.org prefers over its older Attorney type, carrying the same name, address, phone number, practice areas and attorney names a visitor can read on the page.
The rule that matters is agreement. Markup that claims a fourth office, a practice area the page never mentions, or review counts nobody can see gives the engine two different stories about the firm. Fix the visible page first, then make the markup describe it.
5. A recent, honest date
Perplexity's cited set for this question turned over almost completely in one month. Of the 22 pages it cited in September, 3 were still cited in October. We cannot prove the date is the reason, and some pages that dropped out had not changed at all. Freshness is still cheap to signal honestly: a visible "last updated" date, a dateModified value in the markup that matches it, and a real change behind both. Changing the date without changing the page fails the reader, who can see nothing new.
6. Directories the engines actually cite
Not every directory pulls weight, and the two engines lean on different ones. This is how many of the same 54 consumer legal questions cited each directory at least once in October, at 12 runs per question per engine:
| Source | ChatGPT (questions of 54) | Perplexity (questions of 54) |
|---|---|---|
| Justia | 19 | 5 |
| Super Lawyers | 16 | 7 |
| Avvo | 14 | 5 |
| Best Lawyers | 10 | 2 |
| Reddit (not a directory) | 10 | 0 |
| Martindale-Hubbell | 9 | 2 |
| FindLaw | 7 | 12 |
| Lawyers.com | 7 | 5 |
| LawInfo | 0 | 8 |
Two readings. First, complete and matching profiles on Justia, Super Lawyers, Avvo and FindLaw cover most of what either engine cited. Second, FindLaw and LawInfo matter far more to Perplexity than to ChatGPT, so checking one engine hides half the picture. These numbers come from two practices in two practice areas. Your field's mix will differ, which is the reason for the test at the end of this page.
7. A page the engine can quote
Engines lift sentences, list items and table rows. They struggle with a 400-word paragraph that circles its answer. Put the direct answer in the first two sentences under each heading, keep paragraphs short, and use a list or a table wherever the content really is a list or a table. Our guide to AEO for legal services covers that page structure in more depth. This page follows it: the answer to its title question is its first sentence.
How we measured this
We track a fixed set of questions with our own citation tracker. Each question runs 12 times a month on ChatGPT and 12 times on Perplexity through DataForSEO's LLM responses API, and we log every URL each answer cites. Repeating the run matters because one answer is an anecdote. In October, the page ChatGPT cited most often for this question appeared in 9 of its 12 runs, and no page appeared in all 12.
Every rate gets a 95% confidence interval. With 12 runs, a source cited 0 times has a true rate somewhere below about 24%, and a source cited 12 times sits above about 76%. Two limits: answers through the API can differ from what a logged-in user sees in the ChatGPT app, and the client practices behind the directory figures are not named.
Which sources Perplexity cited for this question, and how fast they change
| Measure | September 2026 | October 2026 |
|---|---|---|
| Unique URLs cited | 22 | 10 |
| Unique domains cited | 20 | 10 |
| URLs cited in all 12 runs | 16 | 10 |
| URLs carried over from September | n/a | 3 |
| Domains Perplexity had not cited the month before | n/a | 5 of 10 |
Inside a month, Perplexity is close to deterministic on this question: in October, every cited page appeared in all 12 runs. Between months, it is not. Only three exact pages survived from September to October (one on webris.org, one on resultfirst.com and one LinkedIn post), two more sites stayed with a different page, and five of October's ten sources were sites Perplexity had not cited in September.
ChatGPT behaved differently on the same question. It cited 29 different URLs in each month, mostly general posts about how ChatGPT recommends businesses plus OpenAI's own help pages, and it shared zero domains with Perplexity in either month. A firm can be a steady source on one engine and absent from the other.
For a firm, the turnover is good news: seats at this table open up every month. November is the first run that can include this page, and we will add the result here.
Why most firms are invisible here
The firm sites we review usually miss on the same four points:
- They target head terms. "Best personal injury lawyer in Houston" is the question every firm chases, and engines tend to answer it from directories and ranked lists rather than from firm websites.
- Their site answers nothing specific. Practice-area pages describe services; they rarely answer the question a client actually typed.
- Their dates are old. A 2019 blog post with no update date competes against pages revised this quarter.
- Their markup disagrees with their page. Old plugins leave behind offices, phone numbers and review counts the page no longer shows.
Run this on your own firm this week
You do not need a tool for a first reading. You need 30 minutes and a spreadsheet.
- Write three questions a client would type, using your city and practice area: one "best [practice] lawyer in [city]" question, one "how do I choose a [practice] lawyer" question, and one question about a specific situation, such as "can I sue if I was partly at fault for a wreck in [state]?"
- Ask each question five times in ChatGPT with search on and five times in Perplexity, each time in a new chat. One answer is not a measurement.
- For every answer, record three things: whether your firm is named, whether your site is linked as a source, and which URLs are cited.
- Read the counts honestly. 0 of 5 means you are not a regular answer to that question. 2 or 3 of 5 means you are in the rotation but not settled in it. 5 of 5 is a position worth protecting.
- Open the cited pages and list what they have that your site lacks: the question in the title, a recent date, a table, matching directory profiles. Fix one gap on one page, then run the same questions next month.
If you want a second set of eyes on a specific page, our AI visibility audit scores it and sends back a fix list. Our law firm marketing page explains how we work with firms, and our comparison of SEO and AEO companies for law firms covers who else does this work.
Frequently asked questions
How does ChatGPT decide which law firms to recommend?
Does ChatGPT use Google rankings to recommend lawyers?
Which directories does ChatGPT read for law firms?
How long does it take for a law firm to show up in ChatGPT or Perplexity?
Can a law firm pay to be recommended by ChatGPT?
Matthew (Matt) Bertram helps energy and industrial companies get found, trusted, and chosen as AI reshapes how buyers decide. As owner and CEO of EWR Digital, he leads the agency's AI search and SEO work for oil and gas and other capital-intensive industries; he is also President of ModalPoint. He is also Chief Marketing Officer of the Oil & Gas Global Network (OGGN) and the author of multiple books, including LLM Visibility: A Decision-Grade System for Winning AI-Mediated Discovery and the co-authored Oil & Gas Sales & Marketing: The Energy Growth Playbook for Oil and Gas Leaders. He is a member of the American Petroleum Institute's Houston Chapter and the International Association of Privacy Professionals (IAPP).


