Schema can help AI platforms identify your firm, not cite it

Schema markup is the most oversold item in AI search right now, and it’s also genuinely necessary. Firms waste money when they pay for schema as a way to get cited, and the identity work it’s actually good for never gets done.

Ahrefs ran the cleanest test available. They tracked 1,885 pages that added JSON-LD between August 2025 and March 2026 against roughly 4,000 matched control pages. Here’s what happened to AI citations:

Those are noise-level changes. Adding structured data didn’t move AI citations in either direction by enough to be worth paying for.

Where the three-times claim comes from

You’ll see it claimed that cited pages are around three times more likely to have schema markup. The statistic is real. The conclusion people draw from it, that schema causes the citations, doesn’t hold up.

Sites with thorough structured data also tend to keep their content current, publish regularly, have sound technical foundations and get referenced elsewhere. The schema comes along with a well-run site. Ahrefs’ comparison held site quality constant, and once it did, the effect disappeared.

FAQ markup, the schema type sold hardest to law firms, stopped producing rich results in Google on May 7, 2026. Search Console’s FAQ reporting and Rich Results Test support were removed in June. A proposal that still lists FAQ schema as a way to win space on the results page is selling a feature that no longer exists.

What schema is for: telling AI which firm is yours

Schema’s real job is making a machine certain who your firm is.

An AI platform putting together an answer about your firm has to decide whether the Smith Law Firm on your site, the one in the state bar directory, the one on Avvo and the one on a Google Business Profile are one entity or four. Structured data answers that without ambiguity.

That’s called entity resolution. It won’t make your pages more quotable, but it does make your firm hard to mistake for anyone else, which matters when an AI platform is deciding whether the reviews and results it found belong to you.

What to build for a law firm

Keep it lean and accurate, and point it outward. Three node types and one array do most of the work.

Organization or LegalService for the firm itself, with the exact legal name, address and phone that appear everywhere else. Use the name on the bar registration, not a marketing variation of it.

Person for each lawyer, with real credentials: bar admissions, admission dates, law school and practice areas. Link each one from the firm node, so the authorship on your content connects to a credentialed lawyer and not just a name on a byline.

The sameAs array does most of the work

sameAs is the part most implementations skip, and it’s the part that matters most. It’s a list of URLs that says “these profiles are also me.” For a personal injury firm it should point at:

• The state bar listing for the firm and each attorney
• The Google Business Profile
• LinkedIn, for the firm and the named attorneys
• Wikidata, if an entry exists
• Avvo, Justia, FindLaw, Martindale, and any legal directory profile you actually control
• Any local or trade press that maintains a profile page

A machine can check every one of those. A sameAs link to a state bar listing is a much stronger identity signal than anything you say about yourself, because the bar is an authority you don’t control.

Make your listings agree before you write any markup

Schema only claims the profiles are yours. The profiles matching each other is what backs that claim up.

If your site says “Smith & Associates Injury Law” at one address and the bar listing says “Smith and Associates, LLP” at another, your markup has told the machine to link two records that contradict each other. That’s worse than not linking them at all.

So before anyone writes a line of JSON-LD, pull every profile you can find and reconcile the name, address and phone against one canonical version. Then point the markup at profiles that agree. It’s tedious work, and it’s where most of the value is.

What to expect from schema

Don’t expect more citations. Expect that when an AI platform does surface your firm, it shows the right practice areas, the right locations, the right attorneys and reviews that actually belong to you.

That’s a real result, and it’s what this work can deliver. If someone pitches schema as a way to get cited by AI, ask what evidence they’re working from, because the best controlled test so far found it doesn’t move citations.