Most of what AI says about your firm comes from pages you do not own

When an answer engine describes your firm, it is mostly reading other people’s pages about you. Not your practice-area pages. Not your about page. Other people’s.

BrandMentions analyzed roughly 410,000 mentions across 240 brands and found 85% of brand mentions came from third-party pages. A separate analysis of more than 21,000 brands landed on the same 85%, and found brands were around 6.5 times more likely to be cited through third-party sources than through their own domain.

The number that should change how you think about budget is this one: when a brand appeared in the evidence an engine had retrieved, its citation rate was 53.1%. When it did not, 10.6%. Being in the evidence set is most of the game, and the evidence set is largely not yours.

Why a good website is not enough

This is the part that frustrates firms who have already spent money. You can have a fast, well-structured, thoroughly written site and still be absent from answers, because the engine assembling the answer never had a reason to retrieve your pages.

Victorious tested 175 brands across eight AI engines and found something worth sitting with. Asked about a brand directly, the engines described it accurately 96% of the time. Asked a category or problem-aware question, the kind a real prospect actually asks, 89% of those brands never appeared at all.

Translate that to a firm. Ask an engine “tell me about Smith Injury Law” and it will probably do fine. Ask it “who should I call after a truck accident in Sacramento” and Smith Injury Law may simply not be in the conversation. Your prospects ask the second kind of question. Your website mostly answers the first.

Where the evidence actually lives

The composition of third-party citations is lopsided and predictable. Roughly 90% of third-party citations come from listicles, comparison pages, and review sites. Around 80% of cited brands appear in the first three positions on those pages.

For personal injury specifically, the evidence layer is:

• The state bar listing. The single highest-authority record of your firm’s existence, and one no marketing team can fake.
• Legal directories. Avvo, Justia, FindLaw, Martindale, Super Lawyers. Not because they send traffic, because they are read.
• Google Business Profile, complete and current, including categories, hours, photos, and service detail.
• Yelp. OpenAI licensed roughly 330 million Yelp reviews and more than 8 million business listings, and ChatGPT surfaces Yelp ratings and details in local answers. The deal is non-exclusive, so expect similar data to reach other engines.
• Local and trade press. Case results reported by someone other than you, verdicts covered locally, quotes in coverage of an issue you actually know something about.
• Community platforms. Around 22% of AI answers cite Reddit; Reddit, LinkedIn and YouTube together account for roughly 48% of citations across an analysis of 5.5 million answers.

The uncomfortable implication for a firm

The work that moves this is not work you fully control, and it is slower than publishing a page.

You cannot write your way into a “best truck accident lawyers in Sacramento” listicle you do not own. You can be worth including in it, be findable and verifiable when the writer checks, and have your record consistent enough that including you is easy. That is a different activity from content production, and most SEO retainers do not actually staff it.

It also means the honest first question when a firm is not being cited is not “what should we publish.” It is “what does the internet currently say about this firm, and is any of it wrong.” I have yet to audit a firm where the answer was “nothing and no.”

What to do about it, in order

1. Reconcile the record. Every profile, every directory, every listing. One canonical name, address, and phone. Fix contradictions before adding anything new, because contradictions actively cost you.
2. Fill the gaps that already exist. Most firms have half-completed directory profiles they were sold years ago and abandoned. Completing them is cheap and it is evidence.
3. Earn third-party mentions on purpose. Local press, bar association work, trade coverage, genuine expert commentary. Slow, unglamorous, and the thing that actually moves the 53.1% number.
4. Own the category question on your own site anyway. Educational, problem-aware content is what engines pull from to build category answers, and getting your firm named inside that content is the path in. It is necessary. It is just not sufficient.
5. Check what the engines currently say. Not once. On a cadence, per engine, against a set of prompts a real client would type.

The honest caveat

I cannot tell you exactly how any given engine weighs a bar listing against a Yelp review against a local news mention. Nobody outside those companies can, and anyone who gives you a weighting is guessing.

What the data supports is the direction: the evidence lives off your domain, being present in it roughly five times your odds of being cited, and no amount of on-site work substitutes for it.