50 Personal Injury AI Search Statistics for Law Firms

A personal injury intake sheet records two dates that matter here: the day of the accident and the day the person first called the firm. Between them the person reads, asks around, and decides whether to call anyone, and more of that reading now happens in a chat window.

44% of US adults say they use ChatGPT, up from 34% a year earlier.¹ In Clio's own consumer survey, 24% of US adults had already used AI to answer a legal question, and another 40% would consider it.²

So in September 2026 we put the same accident scenarios, from rear-end crashes to dog bites to a misdiagnosis, to ChatGPT and Google AI Mode, Google's chat-style search answers, in two forms: what do I do now (the post-accident questions) and which lawyer should I call (the lawyer-choice questions).

Accident questions and lawyer-choice questions read different pages, so each needs its own audit.

This is one month of US questions, and it shows which pages appeared beside which questions, not why the engines chose them.

The Accident Question and the Lawyer Question Barely Share a Source

The scenarios covered eight kinds of injury: car, truck, and motorcycle crashes, slip-and-falls, dog bites, workplace injuries, injuries on someone else's property, and medical errors.

Each was asked once by a person deciding what to do ("What to do after a hit-and-run?") and once by a person choosing a firm ("I need a Houston hit-and-run lawyer who takes uninsured motorist claims on contingency").

The figures pool both engines, with ChatGPT's answers the larger share in both question types.

The list of sites a firm would check first for lawyer-choice questions covered 1.70% of the post-accident answers that cited any source. The post-accident list covered 7.65% of the lawyer-choice answers.

A list built from either question's answers covers almost none of the other's. Our shortest questions drifted: asked "What are my options now?", Google AI Mode answered with pages about stock options.

Asked What to Do Next, the Answers Read Government and Health Pages

The pool of people with these questions is large, even before anyone asks an assistant:

61.00% of the post-accident answers cited a government page. On ChatGPT the share was 89.17%, led by the CDC, the Centers for Disease Control and Prevention (cited in 18.33% of its post-accident answers), OSHA, the Occupational Safety and Health Administration (10.83%), and NHTSA (7.50%).

Those agencies publish the facts a person needs first: when a bite needs a doctor, how a workplace injury gets reported, what to do at a crash scene.

State pages carry the local part, such as the Texas Department of Insurance's "Injured employee FAQ" or the California Department of Insurance's "So You've Had an Accident, What's Next?".

34.50% of post-accident answers cited a state or local government page, about as many as cited a federal one (35.50%).

The other sources: health and safety groups such as the Red Cross and the HealthyChildren site for parents (12.00% of answers), Uber, Lyft, and insurers (8.00%), and video and forums (7.00%). Lawyer directories appeared in 1.50% of these answers. Set beside the lawyer questions, the pages the answers cite nearly invert:

12.00% of the post-accident answers cited no source at all, against 2.00% of the lawyer-choice answers. Those answers named no page at all, so no list of cited sites shows what they said.

Google says AI Mode and AI Overviews, the summaries above its results, may build an answer by running several related searches across subtopics and data sources at once, which it calls a "query fan-out"⁵.

A question about a dog bite has subtopics no lawyer owns (the wound, the report to animal control, the owner's insurance), and the pages that own them are mostly public ones. That is my reading; the data shows what was cited, not why.

Asked Which Lawyer to Call, the Answers Read Directories and Firm Sites

88.00% of the lawyer-choice answers cited at least one law firm's own site, and 37.00% cited a lawyer directory or a Google business listing. Super Lawyers was cited in 12.00% of them, Expertise.com in 9.50%, Avvo in 9.00%, and Justia in 7.00%.

The wording of the question changed which pages were cited. On the plainest questions ("Best truck accident lawyers San Diego"), directories and listings led.

Once the question carried the person's circumstance, a firm's own page appeared in 95.00% of answers and a directory or listing in 22.14%. Ask for a hit-and-run lawyer in Houston who takes uninsured motorist claims, and the answers cited firms' own pages about exactly that.

These are the pages most firms already work on: profiles, reviews, practice pages. It is crowded at the top, and what ChatGPT prints above those pages is often a card of about six rated firms.

The Firm's Accident Guide Gets Cited, and the Answer Rarely Says Lawyer

A quarter of the post-accident answers (25.00%) cited a law firm's page, and almost all of those pages were guides, with titles like "What Should You Do if You're the Victim of a Hit and Run?" and "Evidence to Preserve After a Slip-and-Fall Accident". 24.50% of the post-accident answers cited a firm's guide or explainer rather than a practice-area page.

No firm owns those pages. No single firm's site was cited in more than 1.50% of the post-accident answers. A firm with a clear guide for its own state competes there with other firms' guides and with the agencies, and barely with directories.

The engines differed here. ChatGPT cited a firm's page in 6.67% of its post-accident answers and leaned on the agencies instead. Google AI Mode cited firms' guides far more often.

17.50% of the post-accident answers cited a firm's page without mentioning a lawyer. In those answers the firm's link sits under guidance the firm wrote, and the text never suggests the person might need a lawyer.

The accident guide is cited for what it says about the first steps, and it is read by someone who has not yet decided to call any firm.

Give Each List Its Own Owner

The facts an accident guide states are set by legislatures and agencies, and they change.

Florida shortened its deadline and Louisiana lengthened its own. A guide written before either change and never revised gives an injured person today the wrong amount of time.

Accuracy is the worry firms already name: among US firms with fewer than 500 employees that use AI, it was the top reported challenge, at 46%.¹¹

Both kinds of question can be measured as separate topics, each with its own pages, when the same questions run every day.

So give the lawyer list to whoever keeps the firm's profiles and practice pages, and the accident list to the attorneys whose names sit on the guides. They are different pages, read by people at different moments.