How US Law Firms Get Found on ChatGPT, Google AI & AI Search

The part most law firm AI guides skip

Read the agency posts currently ranking for this query and you'll come away believing the whole answer is schema. Add Attorney schema. Add FAQPage. Tag practice areas with LegalService. Ship it.

That advice isn't wrong. It's just the smallest piece. Whether ChatGPT names your firm, or whether Google's AI Overview pulls from your page, is shaped largely by signals that live outside your website, especially when the model is deciding whether to recommend you rather than quote a page you wrote.

Consider the scale of what you're optimizing for. ChatGPT reached roughly 900 million weekly active users by early 2026 and crossed a billion monthly app users soon after, making AI assistants a mainstream first stop for research, including legal research.

At the same time, zero-click behavior on Google has climbed to around 68 percent of searches, with AI Overviews appearing on more than a fifth of queries and pushing zero-click rates above 80 percent when they show.

People are getting answers, and recommendations, without ever reaching a results page. If your firm isn't part of that answer, you're invisible at the exact moment a client forms an opinion.

ChatGPT runs in two modes, and the distinction matters. In its default mode it answers from training data drawn largely from public web corpora, so your firm appears only if it left a fingerprint big enough to survive.

In ChatGPT Search it combines live retrieval with the model's reasoning and returns answers with source links, leaning on web partnerships that include Bing-indexed results.

When somebody asks for the "best truck accident attorney in Texas," the names that surface are heavily shaped by what already lives on directories, legal press, Wikipedia, Reddit, and high-citation legal databases. Your site is one input. The internet's accumulated record of your firm is the bigger one. Our answer engine optimization approach treats both as one system.

Two different games: cited as a source vs. recommended as a firm

There are two distinct wins in AI search, and they require different work.

Cited as a source. A potential client asks an informational question, "what's the statute of limitations on a slip-and-fall in Pennsylvania," "how does parental relocation work in Florida custody cases," and the AI builds its answer from one or more attorney-authored pages.

This is where content quality, schema, and answer-first structure pay off. A well-structured practice-area page can win citations almost mechanically.

Recommended as a firm. A different query, "best litigation lawyer in New York," "who handles medical malpractice lawsuits," "top rated injury law firms near me," pulls from a different set of signals. The model is making a quasi-recommendation, and on legal topics it's conservative.

It leans on accumulated brand mentions, directory presence, reviews, press, and consistent firm-entity signals across the web. Flawless schema won't carry you here if your firm isn't a recognized entity beyond its own domain.

Most firms apply the same playbook to both. The playbook for source citation is editorial. The playbook for firm recommendation is reputational. They are not the same project, and conflating them is why so much law-firm AI spend underperforms.

This is exactly the gap we cover in why ChatGPT recommendations matter for law firms.

How each platform picks law firms

ChatGPT Search. Combines live web retrieval with the model's reasoning and returns sourced answers. Pages with clear claims, specific statistics, primary-source citations, and a direct answer near the top are easier to retrieve and quote.

Question-style H2s with the answer in the first 40 to 60 words do well. A focused ChatGPT SEO strategy lives at the intersection of Bing-index quality, answer-first structure, and recency.

ChatGPT default (no retrieval). Older training cuts. Your firm appears only if you've accumulated mentions, listings, press, and citations that survived in the corpus. You can't move this in the short term, only build toward it.

Google AI Overviews. Built on Gemini and rooted in core Search ranking and quality systems. Google's own documentation is explicit that the same things that make pages rank also drive AI Overview inclusion. Freshness helps when a topic genuinely changes, but it isn't a separate freshness-only system.

Gemini (standalone). Tightly integrated with Google's Knowledge Graph, Business Profile, and Maps. For local legal queries, Business Profile completeness and review volume matter more here than on any other platform, which is why Gemini SEO leans on profile depth and review velocity.

Perplexity. The most transparent about its sources. It rewards content structured like a research brief, with clear claims, sourced statistics, and links to primary law. Perplexity SEO is research-grade citation discipline applied to your own pages.

Claude. Doesn't browse by default in consumer chat and favors longer, nuanced content with full citations. Claude SEO rewards depth and attribution over snippet-bait.

Microsoft Copilot. Built on Bing's index and woven into Windows and Microsoft 365. Optimizing for the Bing index helps both Copilot and ChatGPT Search retrieval.

The 45-minute AI visibility audit

Before you change anything, find out where you stand. You can run this yourself, or have us benchmark it for you with a free AI visibility audit.

Step 1: Firm-recommendation queries. In ChatGPT, Gemini, Perplexity, and Claude, run prompts that ask the model to name a firm: "best personal injury lawyer near me," "top truck accident attorney in [your state]," "best litigation lawyer in [your city]," "law firms experienced with catastrophic injury cases in [your state]."
Run each twice. Outputs vary, and the variance tells you whether you're sitting on the edge of the model's confidence threshold.

Step 2: Informational queries. "Who handles medical malpractice lawsuits?" "What's the statute of limitations for personal injury in [your state]?" "What should I do after a car accident in [state]?" Note which sources the model cites.
If you only see FindLaw, Nolo, Justia, and .gov pages with no law-firm sites, that's the citation real estate currently available in your topic.

Step 3: Brand check. Search your firm name on Perplexity. If the citations are mostly your own site echoed back, you have a brand-mention problem, not a content problem.

Step 4: Competitor comparison. Run the same queries with a competitor's name in mind. The gap between where they appear and where you don't is your work plan.

Step 5: Document. Keep a simple two-column log of where you appear and where you don't, and repeat quarterly. Firms losing leads to ChatGPT and AI search almost always discover the leak here first.

The on-site foundation: the price of admission

On-site work is necessary but not sufficient. It's the layer most agencies sell as the whole solution when it's really one of three. Done well, AI SEO for law firms treats schema, structure, and attribution as a connected system.

Schema. A handful of Schema.org types carry most of the load. Use LegalService on practice-area pages with areaServed tied to specific cities and counties. Use Attorney or Person on every bio, with knowsAbout, alumniOf, and a visible bar-admission state. Use LocalBusiness on office pages with accurate NAP and geo.

Use FAQPage built from real client questions, not marketing questions. Use Article on every post, with the author connected by @id to the attorney's Person schema. Use Review and AggregateRating only when legitimate, faking these exposes you under ABA Model Rule 7.1.

Connect every block through @id so the whole site reads as one entity graph, and validate with Google's Rich Results Test. This entity discipline is the backbone of generative engine optimization.

Answer-first structure. Lead with the answer in the first 40 to 60 words, then expand into nuance, exceptions, and jurisdiction. Pages that bury the answer don't get cited.

Question-shaped H2s. "How long do I have to file a personal injury claim in Virginia?" outperforms "Personal Injury Statute of Limitations," because it matches how clients actually prompt AI tools.

Visible attorney attribution. Every substantive page should show a reviewing attorney's name, bar-admission state, and a link to their bio. ChatGPT in particular relies on visible content, because it doesn't always parse schema the way Google does.

Last-updated dates and crawl access. Refresh quarterly with substantive updates, not just touched timestamps. And in most cases, allow GPTBot, ClaudeBot, PerplexityBot, and Google-Extended in robots.txt, blocking them removes you from those models' training and retrieval.

Content that actually gets cited

The standard is narrower than most agencies imply. Specific beats general: "In California, personal injury claims must be filed within two years under Code of Civil Procedure §335.1" gets cited; "statutes of limitations vary by state" does not.
‍Name the jurisdiction in plain English, not in fine print, because models are conservative on this kind of high-stakes content.

‍Attribute to a named attorney with a bar number and a real bio, not "Our Legal Team." Link to primary law, statutes, regulations, and court opinions, most of which sit at the Cornell Legal Information Institute, rather than to other agency blog posts. Structure as Q&A in real client language, and add a labeled TL;DR block at the top of long content, AI extraction loves a clear summary.

What doesn't get cited: 600-word generic posts, pages that delay the answer, brochure-style practice areas, and anything ghostwritten and unattributed.

Off-site authority is the bigger lever

The training and retrieval data behind these models is dominated by Wikipedia, Reddit, established news, government domains, legal directories, and academic content. Your website matters; the rest of the internet matters more.

The US off-site authority stack starts with legal directories, Justia, FindLaw, Avvo, Martindale-Hubbell, Super Lawyers, Best Lawyers, Chambers USA, Legal 500, maintained as substantive profiles, not skeletons. Add state and local bar profiles, which are free, underused, and trusted because they sit on .org and .gov domains.

Pursue local and trade press: Law360, the ABA Journal, regional business journals. Where a firm or attorney genuinely qualifies, a Wikipedia or similar reference is high-leverage. Reddit, Quora, and forums absorb organic mentions, though never self-promotion, which violates both platform and bar rules. Podcasts and video with transcripts published on the web get pulled into retrieval.

And Google Business Profile is foundational for Gemini and local AI Overviews. Schema gets you to extractable; off-site presence gets you to known.

This is the heart of our AI SEO services for US law firms.

Case study: how a mid-size PI firm rebuilt AI visibility in two quarters

Composite scenario drawn from real engagement patterns; firm details anonymized.

A 14-attorney personal injury firm in a tier-2 Texas metro came to us after a partner watched ChatGPT recommend three competitors for "best truck accident attorney in [his city]" and never name his firm once. Organic traffic looked steady.

Phone volume had quietly slipped over six months, and the marketing director assumed it was seasonal. It wasn't.

The audit (week 1). Forty-five minutes of prompt testing returned a clear pattern: the firm appeared inconsistently on informational queries and never on firm-recommendation queries. Informational citations went to FindLaw, Nolo, and the Texas Department of Insurance, almost no law-firm sites.
On Perplexity, searching the firm name returned mostly its own pages.

Brand-mention problem confirmed. Underneath: strong Google rankings, thin schema (a single LegalService block on the homepage), generic "Our Legal Team" bylines, complete Avvo and Martindale profiles but a Justia page abandoned years earlier, a stale Business Profile, and no podcast or trade-press placements in two years.

Days 1 to 14. Schema rebuilt across attorney bios, practice-area pages, and three office locations, all cross-referenced via @id. Justia profile reclaimed and expanded within the bounds of Texas advertising rules.
Top three practice-area pages rewritten so the first 100 words answered the question implied by the title. Generic bylines replaced with named attribution plus bar-admission state.

Days 15 to 60. Real FAQ blocks added from actual intake calls. Five top posts refreshed with statute references, recent appellate decisions, and TL;DR blocks. Two attorneys placed on a regional plaintiffs'-bar podcast, transcripts published to both sites. Business Profile reactivated with weekly posts.

Days 61 to 90. A cadence of three attorney-authored pieces per month answering questions the audit showed AI was citing other sources for. A bylined op-ed in the regional business journal. State bar profile updated. Local press picked up a community sponsorship the firm had quietly run for years.

By week 16, the same prompts told a different story. The firm appeared in four of six ChatGPT firm-recommendation queries locally (still inconsistent on the most competitive statewide phrasing).
It was cited as a source on three informational queries where it had been invisible, including the truck-accident statute page rewritten in week two. Perplexity now returned third-party citations, the podcast, the op-ed, the updated Justia profile, alongside its own site. Gemini surfaced the firm on local queries.

The honest part. Phone volume recovered roughly two-thirds of the dip by the end of quarter two, with intake forms showing a small but growing share of clients answering "ChatGPT" or "Google AI" to the how-did-you-find-us question.

The firm hadn't yet cracked the most competitive statewide AI recommendations, that's the work of quarters three through six, and it depends mostly on more press and directory authority, not more on-site work.
What carried the result wasn't the schema alone. The biggest movers were named attorney bylines, answer-first openings, the reclaimed Justia profile, and the podcast placement: on-site editorial discipline and off-site authority, applied in the right order. You can see comparable patterns in our case study work.

Practice-area realities

High-value legal searches don't follow one pattern, and a generic playbook leaks budget in whichever lane it doesn't fit. Consumer practice areas reward on-site content depth and EEAT; commercial and high-stakes areas reward off-site authority, directories, press, expert credentials, and trial record.

Personal injury is the most competitive AI search environment for US firms, with AI Overviews triggering constantly, depth by injury type, insurance and settlement FAQs, reviews, and aggressive third-party citation are what pay off.

See how personal injury lawyers get found. Truck accident is urgent, regulation-heavy intent where models weight specific FMCSA citation and named-attorney credibility; we go deeper in truck accident lawyer AI SEO.

‍Medical malpractice is trust-heavy, where board certifications, expert affiliations, and standards-of-care content with primary citation move the needle, covered in medical malpractice AI SEO.

‍Catastrophic injury rewards documented trial experience and long-term damages depth, see catastrophic injury lawyer AI SEO. Criminal defense and DUI are urgent, local, and conservative, where same-day responsiveness and visible trust signals win, as in how criminal defense lawyers get found.

‍Family law rewards jurisdiction-specific content because the field is state-defined, detailed in how family law lawyers get found.

‍Commercial litigation, estate planning, immigration, employment, and bankruptcy each behave differently again, commercial firms especially lean on off-site authority, which is the focus of how commercial litigation firms get found.

Compliance: the part that can actually put your license at risk

ABA Model Rules aren't binding law. Each state adopts its own Rules of Professional Conduct, and those, with state ethics opinions, govern your practice. Treat the Model Rules and ABA opinions as persuasive; your state's rules are the law.

Model Rule 7.1 prohibits false or misleading communications about a lawyer's services, and AI tools can cross that line easily. If you publish AI-drafted content under an attorney's name, the attorney owns everything in it. Rules 7.2 and 7.3 govern advertising and solicitation, which is where AI-generated outreach and contact-initiating chatbots sit.

ABA Formal Opinion 512, issued July 29, 2024, is now the de facto baseline. It addresses competence, confidentiality, client communication, supervision, candor to the tribunal, and reasonable fees, and its central message is that lawyers must understand the tools they use, supervise outputs, and verify what those tools produce.
Publishing an AI-drafted post under an attorney byline with no real review is a supervisory failure.

The sanctions environment is no longer theoretical. Public trackers of AI-hallucination cases catalogued well over a thousand court decisions worldwide involving fabricated AI citations by early 2026, the majority in US courts, with several new cases added every day, and federal appellate courts, including the Sixth Circuit, have sanctioned attorneys for briefs built on fake citations.

The practical rules are simple: use AI for drafting and structure, not legal substance; list a named reviewing attorney and review date on every published page; keep a documented review workflow ("an AI wrote it" is not a defense); never fake reviews, awards, or ratings; and disclose AI use where your state requires it. AI changes what you can produce, not what you're responsible for.

Check the ABA Model Rules and your state's most recent guidance before deploying anything client-facing.

Tracking and measuring

Start free: run the 45-minute audit quarterly, watch Google Search Console and Business Profile insights, log manual prompt tests across the major models in a spreadsheet, and set up GA4 custom channel grouping for AI referrals (chat.openai.com, perplexity.ai, gemini.google.com, copilot.microsoft.com, claude.ai).

At scale, citation trackers like Profound, Otterly, and Peec.AI help, with Ahrefs and Semrush adding LLM modules. Track the queries where you appear each month, share of voice against three or four named competitors, AI-referral traffic, branded search volume, and the intake-form "how did you find us" field with an AI option. Ignore daily citation fluctuations (models are non-deterministic), pure schema scores, and any "guaranteed citation" pitch, no vendor can promise that.

Tie it together with how to measure AI search visibility through SkyScale's services.

The future of legal search is recommendation visibility

Traditional rankings still matter, but legal discovery is shifting toward recommendation-driven search. The firms that win over the next few years won't simply rank; they'll become trusted entities, repeatedly cited legal experts inside AI-generated answers, and locally reinforced brands.

Potential clients are already asking AI assistants who the best lawyer near them is before they ever scroll a list of blue links, and the firms appearing there shape trust earlier in the decision. That advantage compounds. The next phase of legal search won't be won by whoever publishes the most content. It'll be won by the firms AI systems trust enough to recommend, and the full AI SEO for lawyers playbook maps the rest of that build.