Twenty-nine citations. That is what ChatGPT produced when we asked it six California divorce questions on August 2026, and every single one went to the California courts or the State Bar. Not one law firm. Not one directory.
Google AI Mode, asked the same six questions the same day, named a family law firm on all six. Two engines, one set of questions, and completely different ideas about who is worth quoting.
Small California family law firms are winning Google AI Mode, which cited a named practice on all six of our divorce questions, while ChatGPT gave all 29 of its citations to the California courts and the State Bar.
Legal directories, widely reported as dominating AI citations, appeared on only one of the six questions.
Twenty-Nine Citations, and Not One Went to a Law Firm
Three questions were informational: property division, how custody gets decided, and whether you can divorce without a lawyer. Three were commercial: whether an uncontested divorce needs an attorney, what a divorce costs, and how to find a good divorce lawyer. Every question named California, because divorce is state law and an unqualified question produces an answer that is wrong everywhere.
| Source | ChatGPT citations |
|---|---|
| selfhelp.courts.ca.gov | 26 |
| calbar.ca.gov | 5 |
| Law firms | 0 |
| Directories | 0 |
Even the most commercial question in the set went to the government. Asked how to find a good divorce lawyer, ChatGPT pointed to the State Bar referral service and the courts’ free legal help page, and named no firm and no directory. One question, property division, returned no citations at all, because the model answered from what it already knew without searching.
That last point matters more than it looks. A firm cannot lose a citation contest that never happened.
Google AI Mode Cited a Family Law Firm on Every Question
The same six questions in Google AI Mode produced a different internet. Named practices carried all six answers, among them The Chula Vista Lawyer, Williams, Drexler & Mand, the Law Offices of Steven E. Springer, AF Law and Campuzano Law.
Property division is the demonstration case. Google AI Mode answered it with two law firm sources and nothing else, no court page, no directory, no publisher. Two small California practices own a top-of-funnel divorce question outright.
Not one firm cited across those six answers is a national name, which lines up with what we find when we ask whether a small firm can outrank BigLaw in an answer engine.
The Directory Finding Everyone Is Quoting Does Not Hold Here
InterCore, an agency that sells AI visibility services to law firms, ran 540 lawyer-hiring questions across six practice areas including family law in July 2026 and found that legal directories were cited first in 77.8 percent of answers. Justia alone took 1,834 citations out of 18,900. The most-cited individual law firm managed 36.
That number is now quoted across the legal marketing web as though it describes AI search. It describes Perplexity, which was the only engine InterCore tested.
In our run, directories appeared on one question out of six, and it was the question that literally asks how to find a lawyer. On the other five, no directory appeared in either engine. Directory dominance is a Perplexity finding, not a law of AI search, and a firm that gives up because of it is conceding two engines it could win.
Family Law Firms Score 92 on SEO and 45 on Answer-Engine Readiness
AISO Studio, a company that sells AI-visibility auditing and therefore has a commercial interest in the finding, scored 308 family law websites in March 2026. The average classic SEO score was 92 out of 100. The average answer-engine readiness score was 45 out of 100. Not one site in 308 earned an A grade, the highest scored 78, and 16 percent carried no schema markup at all.
The same firm audited 299 employment law sites on the same day using the same method. Family law beat employment on every measure and still produced no A grade, which is what makes this a readiness problem rather than a competence problem. The split we found in employment law is repeating in a category that is better run.
Family law firms have spent a decade getting good at the channel that is losing share.
Where a Family Law Firm Should Actually Spend
Stop competing with the court. The California courts self-help center took 26 of ChatGPT’s 29 citations because it answers procedure better than any firm will, and it is free, official and exhaustive.
Write instead for what a court will never publish. What a contested case costs when the other side will not cooperate, how long it actually takes, whether a set of facts is worth fighting over, what happens to the house, what happens to the children while it is pending. Those questions have no institutional owner on either engine, and they are the emotional and urgent queries that bring a client to a consultation rather than to a form packet.
Ground every claim in one state and one statute, then structure the page so a machine can lift a complete answer out of it, which is what family law practice area pages built for citation do differently from pages built for keywords.
Then measure each engine on its own. The answer engine optimization work that earns a Google AI Mode citation does nothing for an engine that answered our first question without searching, and a single-engine audit will keep producing surprises.
Frequently Asked Questions
Should a family law firm try to outrank the state court’s self-help site?
No. Court self-help pages are official, free and comprehensive on procedure. Target the cost, timeline and judgment questions the court declines to answer, where no institutional source currently holds the citation.
Do these California results apply to other states?
Treat them as directional. Divorce is state law, and community property states and equitable distribution states produce different answers, so a firm should run the same questions naming its own state before acting.
Why did ChatGPT cite nothing on one question?
It answered from model training rather than searching the live web. When an engine does not search, no page can be cited, which makes training-data presence a separate problem from on-page formatting.
Where Lexicon Comes In
Running these queries takes an afternoon. Turning the gap into pages that close it is the work, and that is ours. At Lexicon Legal Content, we have built content programs for more than 300 law firms across North America, and for family law that means answer-first pages on the cost, timeline and custody questions no court page will touch, grounded in your state and screened against ABA Model Rule 7.1 before delivery.
To see which sources currently occupy the answers your prospective clients are reading, start with our family law content or book a call and ask for the two-engine citation check. Our team is reachable at 877-486-8123.
David Arato, JD is the founder of Lexicon Legal Content, an attorney-owned legal content marketing agency serving law firms since 2012. He treats every AI visibility question as an evidence problem first, running live queries against the engines before recommending a single page, because the published studies rarely agree with what the engines actually return. David is a frequent contributor to Attorney at Law Magazine and Attorney at Work and a recurring guest on legal marketing podcasts.
