AI Search Optimization for Law Firms
I’m Carlos Arias. I help law firms get found, cited and accurately recommended across AI search by combining technical SEO, legal content strategy, software engineering and data — then connect that visibility to the rest of your marketing system. The goal isn’t another mention. It’s more qualified opportunities and more signed cases.
The gap
Someone asks ChatGPT who the best truck accident lawyer in Miami is. Someone else asks Perplexity which immigration attorney handles EB-5. A third compares medical malpractice firms in Google AI Mode before making a single call.
The answer names your competitors, a directory, a couple of articles. Your firm can be absent from it while holding a strong website and established Google rankings.
That doesn’t mean traditional SEO is dead. It means your search strategy now has to account for another way people discover and evaluate law firms. I treat AI search as an extension of your existing marketing infrastructure, not a replacement for it — your website, content, local presence, reputation and technical SEO all still matter.
What’s added is understanding how these systems retrieve information, choose sources, describe your firm, and connect those answers to a potential client. The objective is simple enough to state: be a credible, accessible, relevant source when an AI system answers a question connected to your practice.
No guarantees, straight answers
Tell me your practice area and your market. I'll test how your firm actually shows up across the AI platforms, tell you who's being named instead of you, and say plainly whether the problem is AI search or something further down your funnel.
The work
Most agencies treat GEO as a content service — a few questions added to a page, a schema plugin installed, and it gets called AI optimization. Content matters, but it’s one part of a system. I came up through software engineering as well as marketing, so I can investigate the technical problem instead of recommending another article.
Before writing more content, find out whether the systems that need your information can reach it.
Robots.txt, indexing directives, canonical URLs, server responses, redirects, XML sitemaps, internal linking and how your content renders. I also check the CDN, firewall and bot-protection settings that quietly block legitimate crawlers.
For ChatGPT search, OpenAI identifies OAI-SearchBot as the crawler used for search discovery, and allowing it matters if you want to be eligible for ChatGPT's summaries and snippets. Google's AI features run on the existing Search technical requirements — indexability and snippet eligibility. Neither platform guarantees an eligible page gets cited.
I don't assume that because a page looks right in a browser, every crawler receives the same useful content. I inspect what the site actually delivers.
AI systems don't need a special version of your website. They need information they can parse and an answer to the question being asked.
Semantic HTML, descriptive headings, clear page relationships, crawlable links, and content that exists as readable text. Important information shouldn't depend on an interaction or a JavaScript implementation that makes it hard to retrieve.
I also check whether pages are organised around real legal questions. A truck accident page shouldn't bury what's useful under generic marketing copy. An EB-5 page should explain the process, eligibility, risks and the attorney's role with enough precision to be worth reading.
The goal isn't writing for a robot. It's making expert information easier for both people and retrieval systems to find.
An AI system may meet your firm through your site, your profile, a directory, an attorney bio or a publication. Those sources should agree.
Firm name, attorneys, practice areas, office locations, contact details, credentials and experience. I review those signals and correct what contradicts.
Where it fits, I implement structured data — Organization, LegalService, attorney information and other types that accurately describe what's actually on the page. Schema is useful for describing entities and relationships. It is not a shortcut to an AI recommendation and I won't sell it as one: Google states plainly that no special schema is required for its AI features, and structured data has to match what users can see.
Publishing another “What to Do After a Car Accident” doesn't make a firm more visible. The question is what your page contributes that the others don't.
I work from your attorneys and your firm's actual knowledge: real client questions, legal research, practice-area experience, your process, jurisdiction-specific guidance, attorney-authored analysis, approved case examples, data that helps someone understand their situation.
Accuracy is not negotiable in legal content. Statutes, regulations and decisions get checked against current authoritative sources. AI can assist with research and drafting; it does not invent citations or publish unverified legal advice under an attorney's name.
Google's guidance separates useful AI-assisted content from scaled content made to manipulate rankings. The problem was never the tool. It's publishing volume with no benefit to the reader.
Your site isn't the only source an AI system reads to work out who you are.
Reputable legal directories, professional associations, relevant publications, local organisations, attorney profiles — the credible sources that can substantiate what your firm claims about itself.
This is not buying hundreds of listings or manufacturing mentions. It's identifying which sources actually matter in your practice area and market. For a personal injury firm that may be local authority and legal publications; for an immigration practice it may be international business communities and credentials tied to the services the firm genuinely provides.
Not one strategy
A personal injury firm, an EB-5 practice and a medical malpractice firm don’t acquire clients the same way, so their AI search strategies shouldn’t be identical either. The same holds for family law, bankruptcy and tax resolution.
They ask what to do after a crash, whether they have a claim, how liability works, who handles serious trucking cases in their city. The content has to answer the legal question, establish why the firm is relevant, and make the next step obvious — then survive intake, because an unanswered inquiry on a high-value case is an expensive way to generate nothing.
EB-5 prospects research hard before they contact anyone: investment requirements, processing, regional centres, source-of-funds documentation, how the options differ. That needs deeper information architecture, accurate regulatory sources, international search considerations, and content built for a long decision — often across languages and across investors, families and business owners who ask different questions.
Prospects are weighing a serious medical and legal question at once. They need credible information on the injury, the potential negligence, the process, and whether this firm has relevant experience. Attorney-reviewed content, accurate medical and legal references, and evidence behind the claims — generic pages and inflated case results are not expertise.
I build the strategy around the way your clients search, evaluate and hire — not by changing the practice-area name at the top of the same template.
Where this sits
A website is no longer a stand-alone tool. It’s one part of a system that connects research, content, visibility, lead acquisition, intake and business outcomes — which is the direction of the AI-driven platform I’ve been building.
What the market asks, what competitors answer, what your firm knows
Attorney judgment before anything is published, not after
Drafted with your brand and practice context, not a generic template
Into your existing site, on your existing platform
The sources and surfaces where your firm should be visible
Prompts re-run, citations tracked, descriptions checked for drift
The inquiry answered, qualified and followed up
The only number that settles whether any of it worked
I’ve built AI-assisted systems that research a client, develop content plans, draft articles in the client’s brand and industry context, review that content and support publishing. The wider agentic platform is being built to connect those to SEO analysis, competitive research, distribution, monitoring and reporting.
The point isn’t to replace your attorneys or remove human judgment. It’s to cut repetitive work, hold consistency, and make the process responsive — research informing content, content reviewed before it publishes, published pages monitored, and the resulting inquiries measured against real business outcomes.
You don’t need to rebuild your technology stack to start. I work with the website, CRM, analytics and tools you already have.
How it works
I define the questions that matter to your practice and market, then test them across the platforms in scope — recording whether your firm appears, which competitors do, and what sources are being cited. Alongside that, a review of your SEO, architecture, technical accessibility and analytics. A baseline, not a screenshot of one flattering ChatGPT answer.
Current visibility measured against the opportunities that actually carry commercial value. That usually surfaces missing practice-area content, thin attorney profiles, crawl problems, inconsistent business information, weak local authority, or a competitor with genuinely better resources. Work gets prioritised by value, difficulty and what we can realistically execute.
I coordinate the technical changes, content development, internal linking, entity information and external authority work — using AI-assisted workflows where they speed up research, production and quality control. Legal accuracy, factual claims and attorney approval stay in the process. The objective is better and more accessible information about your firm, not more pages.
We re-run the prompt set, review citation changes, monitor referral data and compare it against your broader search and intake performance. When a platform describes your firm wrongly, I go find the source. When a competitor keeps appearing for an important question, I work out what supports that and whether there's a legitimate way to compete. AI search moves; the strategy gets adjusted rather than declared finished.
The scorecard
An AI visibility score can be useful. It is not revenue. What I want to know is whether the work is improving your firm’s ability to attract and convert the right clients.
A defined prompt set, tested per platform and recorded separately — a mention in ChatGPT tells you nothing about Google AI Mode or Perplexity. Mention coverage, citation frequency, competitor presence, and whether your firm is described accurately. Prompt, platform, date and conditions all logged, because these answers change.
Which URLs get cited, and what job they do in the answer. A citation to an attorney biography, a legal article and a third-party directory are three different opportunities. This is how we learn which assets are actually carrying the visibility and where nothing exists yet.
Where the platforms allow it: AI-referral sessions, landing pages, calls, form submissions and assisted conversions. Bing Webmaster Tools added AI Performance reporting in February 2026 — citation activity, cited URLs and grounding queries across supported Microsoft AI experiences. Citations are not clicks and not cases, but they are evidence.
Marketing data connected to qualified inquiries, consultations, signed cases and acquisition cost where the information exists. AI attribution is incomplete — someone may find you in an AI answer, then call directly a week later from another device. I report what we can detect, say plainly where the data stops, and use the whole picture to make decisions.
The uncomfortable part
More visibility is only worth something when the firm can turn what it produces into clients. I’ve seen how misleading it is to measure only the top of the funnel.
That’s why I look at the whole path from search to signed case. If your AI visibility improves while intake misses calls, delays follow-up or fails to qualify prospects, those problems need addressing before more visibility is worth buying.
My objective isn’t a report that looks impressive. It’s helping your firm understand what is working and where the next improvement should come from.
Straight terms
Nobody controls these systems. Anyone telling you otherwise is either guessing or selling. Here is the line, in both directions.
I won’t
I will
Those refusals aren’t modesty. Claims like a guaranteed AI ranking, or an llms.txt file that unlocks one, oversimplify how retrieval works — and a strategy built on a misunderstanding is a strategy that stops working the moment the platform changes. A technical foundation, better information about your firm, content with a reason to exist and honest measurement is a more useful basis for a long-term investment.
Why me
More than 25 years across software engineering and digital marketing: building websites and software platforms, developing SEO strategies, working with law firms, integrating marketing technology, and improving the systems that connect a lead to a business outcome.
That combination lets me work both sides of the problem. I can look at how your website is built, how your content is retrieved, how your data is organised, how your marketing tools talk to each other, and how your intake handles what all of it generates.
A law firm doesn’t need another disconnected platform or another agency layer. It needs a strategy, competent execution, and a way to tell whether the work is producing anything.
Questions firms ask
AI Search Optimization is the process of improving how your law firm is discovered, understood, mentioned, and cited in AI-powered search experiences such as ChatGPT, Google AI Mode, Gemini, and Perplexity. It combines technical SEO, content quality, entity consistency, credible external sources, and ongoing measurement.
GEO and SEO overlap substantially. Traditional SEO focuses on search visibility and website performance, while GEO also examines how AI-generated answers retrieve, summarize, and reference information. Strong SEO remains an important foundation. AI search adds another layer of measurement and content-discovery considerations rather than replacing the work you already do.
No. There is no single, fixed ChatGPT ranking that an agency can guarantee. Answers can vary by question, location, model, and available sources. I focus on improving your firm's eligibility, relevance, and credibility, then measuring how often it appears across a defined set of relevant prompts.
Not necessarily. I first evaluate your existing website's technical accessibility, content structure, indexing, and architecture. Many improvements can be made within the current platform. A rebuild should only be recommended when the existing technology creates meaningful limitations that cannot be addressed efficiently.
There is no universal requirement for an llms.txt file or special AI schema to appear in Google AI Overviews or AI Mode. I prioritize established technical SEO, accessible content, accurate structured data, and the crawler controls relevant to each platform. Experimental or optional technologies can be evaluated, but they should not replace the fundamentals.
There is no guaranteed submission process that makes ChatGPT recommend a firm. I work on making your website accessible to relevant crawlers, improving the quality and clarity of your content, strengthening your firm's digital identity, and building credible supporting sources. I then test relevant questions and monitor the results to identify opportunities for improvement.
AI-assisted content is not inherently a problem. Publishing large amounts of repetitive, inaccurate, or low-value content primarily to manipulate rankings can be. For law firms, I emphasize legal research, attorney expertise, original information, factual review, and meaningful value to the reader. Automation should improve the process, not eliminate quality control.
There is no reliable universal timeline. Technical issues may be corrected relatively quickly, while content authority, external signals, and changes in AI visibility can take longer. The starting condition of your website, your practice area, competition, and the platforms being measured all affect the results. I establish a baseline and prioritize the work so progress can be evaluated over time.
I look at platform-specific mentions and citations, the sources AI systems reference, description accuracy, available AI-referral traffic, and the qualified leads and signed cases that can be connected to those channels. I also compare the results with your broader SEO and marketing performance. Attribution is incomplete, so I distinguish measurable outcomes from estimates.
Law firms are the primary focus of this service. My broader experience includes software engineering, digital marketing, and automation across other industries, but the strategy for this page is specifically built around legal client acquisition and the different requirements of individual practice areas.
Let’s talk
Your firm may already have the experience, reputation and legal knowledge people are looking for. The question is whether any of it is being discovered and described accurately when they ask an AI. We’ll look at what your competitors are doing, what you can improve, and which opportunities are worth pursuing — and if the problem turns out not to be AI search, I’ll tell you that too.
Carlos Arias · Marketing Engineer · 25+ years in software engineering & digital marketing
