How do AI assistants decide which law firms to mention?

From two inputs: what their training data absorbed about your firm (directories, reviews, articles, your site), and what live search retrieval finds when the user asks — which is largely the same content classic SEO optimizes, read by a machine that quotes rather than lists. Clean structure, real credentials, and consistent identity across the web drive both.

Should law firms block AI crawlers?

It’s a genuine trade-off: blocking (via robots.txt rules for GPTBot and similar) keeps your content out of training corpora but also out of the systems clients increasingly consult. For a marketing site — whose entire purpose is being found and repeated — most firms are better served by being legible to the machines.

Home › Guides › SEO & Design

Being the Answer: AI Search Visibility

The newest referral source doesn’t have a website. It has an opinion about you.

Law Firm SEO & Design · Chapter 4~6 min read

By Spencer McLennan · Published June 9, 2026

A growing share of “who should I call?” now happens inside a chat window. The client describes the problem in a sentence, and an AI assistant — ChatGPT, Gemini, Claude — names two or three firms with a sentence of reasoning each. There’s no page two. You’re in the answer or you’re not.

Where the machine’s opinion comes from

Two inputs. Training data: what the model absorbed about your firm from the open web — your site, directories, reviews, articles, news. Retrieval: for current questions, these systems run live searches and ground the answer in what they find — which is substantially the same content classic SEO surfaces.¹ Almost every AI optimization is also basic SEO, with a few additions like the llms.txt file.

What’s genuinely different

The machine quotes; it doesn’t list. A results page shows ten options; an answer names two. Direct, quotable statements — “the firm concentrates in immigration appeals; the founding partner has argued before the Federal Court of Appeal” — survive summarization. Adjective fog doesn’t.
Legibility is binary. Text baked into images, content that only exists after JavaScript runs, and crawlers you’ve blocked all read as nothing.² Clean semantic HTML with real text is the price of admission.
Schema is your sworn statement. Person, LegalService, FAQ structured data is the version of your identity a machine can’t misparse.³
You should hear what it says about you. Ask the major assistants what they’d tell a client about your firm, quarterly. Wrong practice areas, a departed partner, a confused merger — you can’t edit the model, but you can fix the sources it reads (usually the directories and stale profiles from Ch. 3). And an impersonator’s lookalike domain feeding the machine is Cybersecurity Ch. 8’s problem wearing a new costume.

FIGURE 1 — THE CLIENT’S NEW FIRST QUESTION

The prompt“My landlord in Bellingham is withholding my deposit — who should I talk to?”

The retrievalThe AI assistant searches; finds the firm whose page actually answers deposit-withholding questions, signed by a named lawyer

The answerTwo firms, one sentence each — assembled from bios, reviews, and quotable pages. No page two.

Everything the machine used was published by the firms — or about them — long before the question was asked.

What actually does it

1. Stay crawlable. Decide your AI-crawler posture deliberately.

2. Write quotable facts. One-sentence, verifiable statements of who does what, on bios and practice pages.

3. Ship the schema. Person + LegalService + honest FAQ, validated.

4. Audit the machine’s answer quarterly — and fix the sources, not the symptom.

Your firm’s AI-visibility checklist

Crawler posture decided; site legible as real text
Quotable, verifiable one-liners on bios and practice pages
Person/LegalService/FAQ schema valid
Quarterly “what does the AI say about us” audit; bad sources fixed

Spencer McLennan is the founder and lead webmaster of LegalWebmasters, and has handled websites, hosting, and security for law firms and professional practices across the United States and Canada since 2007. He holds an MBA, a PCM from the American Marketing Association, and is a graduate student in intelligence studies. He writes the Law Firm Cybersecurity and Law Firm SEO & Design guides. Connect on LinkedIn.

Footnotes

Retrieval-augmented systems ground answers in live web search; what they retrieve is the same content classic SEO surfaces — the difference is the machine reads and quotes it instead of listing it. ↩
Text rendered into images, JavaScript-only content, and blocked crawlers are all ways to be invisible to the systems doing the reading. ↩
Structured data (Person, LegalService, FAQ) is the machine-readable statement of who you are — see Chapter 1 and Chapter 2. ↩

← Chapter 3: The Map Pack Chapter 5: Design, Speed, First Impressions →