CORPUSMeasured study
Who AI assistants cite for immigration questions
We audited 67 US immigration law firm websites, then put 106 of the questions their clients actually ask to a retrieval answer engine and recorded every domain it cited. The firm being asked about was cited in 17% of them.
Every number here is our own measurement, taken on 2026-09-05 with the same audit and the same engine that run for customers. Nothing is modelled or inferred. No firm is named for a poor result.
What we did
We took 67 US immigration law firm websites and ran each through the free Corpus Plan audit. 53 returned pages we could read. 14 returned nothing at all — those are excluded from every figure below, because a site that gives our crawler nothing may be blocking automated requests rather than be broken, and we will not report a guess as a finding.
For 26 of the firms we then wrote the questions a prospective client would plausibly ask — not brand searches, and never the firm's own name, because a question containing the name retrieves the name and proves nothing. Each question went to a retrieval answer engine: one that searches the live web and returns the URLs it used. We recorded the full citation list for every answer.
The firm is usually not in the answer
Across 106 questions, the firm whose own practice area the question described was cited 17% of the time. The other 83% of answers pointed somewhere else — most often a government page, a nonprofit, or a different firm.
This is not a ranking. It is a record of what one engine returned on one day. Ask again next week and the list will move.
What gets cited instead
| Kind of source | Share of 1505 citations |
|---|---|
| Private firm or company site | 58.9% |
| Government or .edu | 19.8% |
| Nonprofit / legal aid (.org) | 14.2% |
| Directory, forum or reference | 7.1% |
Classified by domain: .gov and .edu as government, a fixed list of directories and reference sites as directories, remaining .org domains as nonprofit, everything else as a private site. A private site here means any company — not only law firms.
The most-cited domains
| Domain | Times cited |
|---|---|
| uscis.gov | 60 |
| usa.gov | 40 |
| travel.state.gov | 39 |
| reddit.com | 21 |
| boundless.com | 18 |
| visum-usa.com | 15 |
| indeed.com | 14 |
| justia.com | 14 |
| americanimmigrationcouncil.org | 13 |
The top three are the government sources you would expect. The interesting part is below them: a discussion forum and a venture-funded immigration startup are each cited more often than any individual law firm in the sample. The firms are competing for citation space against explainer content, not only against each other.
What the audits found
Across the 53 readable sites the median overall score was 73/100, but the search pillar sat lower at a median of 64, with 21 of 53 sites scoring under 50. 21% carried no structured data of any kind — the markup that tells an engine what the business is, where it is, and what it does. 4% actively turned away AI crawlers at the network edge, which their owners are unlikely to know.
What this does and does not show
- It is one engine at one moment. Different assistants retrieve differently, and results move on their own.
- Being cited is not the same as ranking, and neither is the same as getting a client.
- We did not test whether any fix causes a citation. We measured a state, not a cause.
- The sample is one practice area in one country. Do not read it as the whole web.
- 14 sites returned nothing to our crawler and are excluded rather than counted as failures.
Check your own site
The audit used here is the free one. It returns every finding, with no email required and nothing blurred out. If you want to see which domains an engine cites for your customers' questions, that is what the paid tier adds.