AI visibility for immigration law firms
AI visibility is how often an AI assistant names your firm when someone asks it who to hire. You cannot buy a place in that answer. The models build it from the sites they trust — directories, review profiles, and pages that answer the question directly — so what you publish and how it is marked up decides whether you are in it.
What we measured
We ran the same questions a client actually types — what an immigration lawyer costs, who speaks Spanish, who handles asylum — against ChatGPT and Claude with live web search, in English and in Spanish. No firm paid to be included and no firm is named in the public numbers.
- 88
- firms audited
- 7
- cities
- 2320
- questions
- 42
- never appear at all
Where the answers come from
When the models answered, these are the sites they cited most. A firm's own website is almost never the first one, which is why the work is not only about your site.
- expertise.com1176 citations
- justia.com671 citations
- abogado.com554 citations
- yelp.com512 citations
- attorneys.superlawyers.com488 citations
What makes a firm citable
Four things decide it, and all four are visible from outside. These are the ones we find broken most often in real audits:
A Spanish version the model can recognize
Half the firms we measure serve clients in Spanish with a site marked English-only, or a Spanish section with no hreflang tags. To the model that section barely exists, so every Spanish question gets answered without them.
Headings written as questions
Models extract answers, not marketing paragraphs. A homepage with no heading phrased as a question gives them nothing to quote, so they quote the directory that does.
Structured data about the practice
Without local business markup the model has to guess what the firm is, where it practices, and which language it works in. Guessing usually means being left out.
Crawler access
Some firms block the crawlers that feed AI answers at the server. While that is true, being named is impossible, and it is a ten-minute fix.
Questions people ask
Can I pay to appear in ChatGPT answers?
No. There is no ad slot or paid placement inside the answer itself. The model assembles it from sources it trusts at that moment. That is why this work is about what those sources say about you, not about buying a position.
How is this different from SEO?
SEO works to put a link in a list of results. This works to get your name inside the answer, where there is no list. They overlap — both reward clear, well-marked content — but ranking first on Google does not mean an AI will name you, and we have measured firms where it does not.
Which AI engines do you measure?
ChatGPT and Claude, both with live web search, in English and in Spanish. We do not measure Gemini or Perplexity today, and we say so rather than list engines we do not run. Every number we publish comes from a query we actually made.
How long does it take to change?
Weeks, not days. Fixing the site is fast; the models picking it up is not, because they re-read sources on their own schedule. We re-run the same questions every month so the direction is visible instead of assumed.
Do you guarantee more clients?
No. We commit to the work and to the measurement, repeated monthly with the same questions so you can check it. Cases depend on your intake, your fees and your market. Anyone who promises you clients from this is guessing.
What does it cost?
The diagnostic is 497 USD and arrives in 72 hours: the full question set, the verbatim answers, the technical review and a 90-day plan. The ongoing work is 2,500 USD a month with a three-month minimum. The diagnostic is yours whether or not you continue.
Find out where you stand
Tell us the firm and the city and we run the questions. You get the answers as the models wrote them, not a score out of a hundred.
Request the diagnostic