Lexicon Legal Content | Blog

Immigration firms competing for AEO and GEO visibility are being sorted by whether their Spanish content is a website or a widget. Thirteen firms already cited by AI engines were read for how they handle Spanish, and the split is sharp: real Spanish sites in the Spanish answers, and mostly no Spanish at all in the English ones. This piece names what each firm built and gives a ten-minute check for your own.

Lawyer Growth Summit 2026 gave every attendee a 63-page workbook and spent two days asking law firm owners to score themselves honestly. The workbook is free to download, so instead of repeating it, here is what I concluded from each session and what I would act on first.

Google’s AI answered the cost question with two small law firms and no legal directory. One page runs 5,000 words with a fee calculator, the other runs 650 words and no FAQ. We read both and found five things they share, none of which is length.

Texas lets a firm call itself Travis County Criminal Lawyers and forbids Best Travis County Criminal Lawyers. Ask ChatGPT for the best criminal lawyer in the county and it answers anyway, using content the firm published. This piece works through what the advertising rules say about AI recommendations, why every published bar opinion answers a different question, and which sentences on a practice page create exposure when an engine compresses them.

Lexicon Legal Content asked ChatGPT, Claude and Gemini the same six California divorce questions in August 2026, recording not just which sources each engine cited but whether any law firm was named in the answer itself. The three engines behaved like three different products, all three declined to send searchers to a firm’s own site on the question about hiring, and four firms surfaced on two engines independently.

We put six California divorce questions to ChatGPT and Google AI Mode on the same day and recorded every source each engine cited. ChatGPT produced 29 citations and gave every one to the California courts or the State Bar. Google AI Mode named a family law firm on all six. This audit shows which firms are being cited, why the directory data everyone is quoting does not hold, and where a family law firm should spend.

The employment law firms answer engines cite are not the ones with the biggest budgets. One Dayton practice has been quoted in five separate sessions across two engines, from a page that breaks most of the rules our industry publishes. This piece names the firms recurring in our testing, takes apart the page that keeps winning, and explains why your own visibility check probably cannot see any of it.

Partners want to know what answer engine visibility returned, and the honest answer is that it cannot be proven the way they mean it. Google’s generative AI report contains impressions and no clicks, and across 54 sites AI referrals made up under 0.6% of traffic for nearly nine in ten. This piece covers the four measures that are genuinely observable, how to frame the investment, and which widely quoted numbers to leave out of the room.

We put six employment law questions to ChatGPT, Google AI Mode and Perplexity on the same day and recorded every source each one cited. ChatGPT named no law firm at all. Perplexity answered the most commercial question with eight local practices. This audit shows which firms are being cited, what their pages actually do, and why a single-engine strategy keeps producing surprises.

Siri is moving to Google’s Gemini models, which means voice search and AI search are now the same optimization problem. The same legal question, put to Google AI Mode and Perplexity once as speech and once as a keyword, returned two different sets of sources. This piece shows what each engine cited, why a government agency changes the answer, and what a firm should do about it.