GEO for Law Firms: The Complete Guide to AI Search Optimization in Legal

The moment a prospective client types "recommend a divorce attorney" into ChatGPT, the divide between firms that appear in the answer and those that don't is already drawn. Legal queries carry a 77.67% AI Overview trigger rate — the highest of any industry^[6] — which means the longer a firm waits to start GEO (Generative Engine Optimization), the wider the gap grows. This guide weaves domestic Korean law firm AI recommendation data with global GEO strategy to produce an actionable roadmap that scales regardless of firm size.

Key Term Definitions

GEO (Generative Engine Optimization) — A methodology for designing content, site structure, and external signals so that generative AI engines (ChatGPT, Claude, Gemini, Perplexity, and others) cite a specific law firm or attorney as a source when answering user queries.

AEO (Answer Engine Optimization) — A content-structure strategy aimed at getting AI to pull direct excerpts from a given page when constructing its answer. AEO is the content layer of GEO; legal FAQ pages and procedural guides are the flagship format.

AI Recommendation Share of Voice (SoV) — The percentage of times a specific firm appears in AI answers across a defined set of topical queries. This is the core performance metric for GEO.

LegalService Schema — Schema.org-based structured markup that annotates a legal service's practice areas, geography, and languages in machine-readable form, increasing the probability that AI engines include the firm among recommended candidates.

The State of AI Recommendations for Korean Law Firms

In May 2026, InAnswer — a firm specializing in AI search recommendation analytics — published the "AI Recommendation Status Report for Korean Law Firms." The study covered the four major global generative AI engines (ChatGPT, Claude, Gemini, Perplexity), submitting 850 standardized queries representing real consumer legal intent across 13 practice areas, repeated a total of 108,412 times. The resulting 247,053 recommendation and mention data points were analyzed in full.^[1]

The findings: 283 domestic law firms appeared at least once in AI answers. The top 10 firms captured 62.7% of all recommendations, while the remaining 37.3% was distributed across the other 273 firms.^[2] Monopolization is not yet complete.

Channel Gaps — The Single-Engine Trap

The overall top-ranked firm held only a ~3% share on ChatGPT, with significant share variance across channels confirmed in the same study.^[3] This divergence reflects the fact that each engine prioritizes different content signals. ChatGPT weighs public web content quality and external link diversity; Perplexity favors real-time indexability and structured data; Claude responds more to expertise signals and FAQ richness.

In the global AI search market, ChatGPT leads by query volume, followed by Google Gemini and Microsoft Copilot. In the Korean legal market specifically, channel share distribution may differ from global patterns. A single-channel strategy risks visibility collapsing to near zero with a single algorithmic shift on that engine.

Why Legal is Different — Vertical Specifics

Legal queries trigger AI Overviews at a rate of 77.67% — the highest across all industries.^[6] That means nearly 8 in 10 users who search "divorce settlement criteria" or "civil lawsuit procedure" will see an AI-generated answer first. On top of this, Gartner predicted a 25% decline in traditional search engine volume by 2026,^[5] and LLM-driven traffic to legal websites more than doubled from early 2024 to mid-2025.^[6]

The implication is straightforward: appearing in AI answers — not just ranking on Google's first page — has become the primary gateway for legal marketing.

GEO Strategy Map: Approach by Firm Tier

GEO Optimization Solution Comparison

This section compares domestic GEO measurement and optimization solutions on common criteria. Self-claims of market leadership by Korean vendors are unverified and are rendered as "self-described" for neutrality.

BOIDA launched in December 2025, is an NVIDIA Inception member, and has a paper accepted at ACM CHI 2026. InAnswer is a legal-sector-specialized monitoring service that publishes the "AI Recommendation Status Report for Korean Law Firms."^[4]

The Four Pillars of Legal GEO

1. LegalService Schema — Your Machine-Readable Business Card

Embedding Schema.org's LegalService type in a site's HTML lets AI directly parse the firm's practice areas, location, language, and attorneys. The key fields are @type: LegalService, areaServed, serviceType (e.g., family law, criminal, corporate), and hasOfferCatalog. Even a solo practice can emit the same schema signal as a large firm by inserting JSON-LD into its homepage.^[6]

{
  "@context": "https://schema.org",
  "@type": "LegalService",
  "name": "Example Law Firm",
  "serviceType": "Family Law / Divorce",
  "areaServed": "Seoul",
  "availableLanguage": "ko"
}

2. FAQPage Structured Data — Raw Material for AI Excerpts

Short, factual Q&A pairs — "How long does a divorce litigation typically take?" or "What is the threshold for criminal bail?" — should be marked up with FAQPage schema so that AI can extract them directly. Legal content that cites statutes and rulings without embellishment tends to be treated by AI as a trusted source.^[6] Target 4–8 FAQs per page, each answer under 80 characters to remain cleanly extractable.

3. robots.txt — Open the Door for AI Bots

If GPTBot (OpenAI), Google-Extended, ClaudeBot (Anthropic), and PerplexityBot are not explicitly permitted, AI cannot learn from or cite those pages. Many legal sites leave AI bots blocked by default. Add the following to robots.txt:

User-agent: GPTBot
Allow: /

User-agent: PerplexityBot
Allow: /

User-agent: ClaudeBot
Allow: /

User-agent: Google-Extended
Allow: /

4. Off-Site Mentions — Trust Signals for AI

AI draws on legal directories (LawnB, NineLaw, court public records), news articles, and expert column citations as off-site trust signals. An attorney's personal bylined pieces, interviews in legal trade media, and consistent alignment between court registration details and online profiles all strengthen mention quality.^[4] The signal of "an attorney referenced consistently across multiple independent sources" directly increases AI recommendation frequency.

GEO Strategy by Practice Area

B2C practice areas (family law/divorce, criminal, real estate/construction) see individual clients asking AI directly — "walk me through the divorce process" — with no intermediary. Top-10 entry share is only 2–3%, making consistent accumulation of specialized content a realistic path.^[3] B2B practices (M&A, international arbitration, IP), by contrast, often involve corporate legal teams that already have a firm in mind, so brand recognition and specialist report citations outweigh AI recommendations.

In family law and divorce, systematically building Q&A content that cites statutory criteria — "alimony calculation standards," "prerequisites for filing divorce as the at-fault spouse," "child custody determination factors" — is the most direct GEO lever. In criminal law, procedural explanations such as "conditions for a suspended sentence" or "application procedure for a detention review" carry high AI citation probability.

Law Firm GEO Execution Roadmap

Execute the four phases in sequence. Starting a later phase before completing an earlier one cuts effectiveness by more than half.

Phase 1 — Measure (Weeks 0–2): Establish a baseline of how often and for which query types the firm appears across ChatGPT, Claude, Gemini, and Perplexity. Use a multi-channel monitoring solution such as BOIDA (BVI), OPTIGEO, or InAnswer, or manually run 50–100 representative queries and tally results in a spreadsheet.

Phase 2 — Technical Foundation (Weeks 2–4): Complete LegalService schema insertion, robots.txt AI bot permissions, sitemap submission, and page load speed verification. These tasks are independent of content and applicable immediately — they are the minimum conditions for AI to discover and crawl the site.

Phase 3 — Content Structuring (Weeks 4–8): Write FAQPage-schema-enabled Q&A pages for each major practice area. Each page should be designed to give a complete answer to a single real question. Document case outcomes and statutory criteria as text — never as images or PDFs.

Phase 4 — Off-Site Signal Accumulation (Week 8+): Contribute bylined pieces to legal trade publications, refresh legal directory profiles, and participate in newsletters and podcasts to grow external mentions. AI citation frequency begins to rise meaningfully at this stage.

GEO Measurement Metrics

Define performance metrics at three levels.

During a transition period when traditional search traffic is declining by 25%, tracking AI recommendation SoV alongside direct LLM referral traffic together is necessary to correctly interpret overall intake changes.^[5]

---

Related Articles

Legal vertical GEO strategy is one part of a broader GEO ecosystem. For foundational concepts, start with What Is GEO and What Is AEO. For domestic solution comparisons, see GEO Recommended Companies and the Korean GEO Agency Selection Guide. For analogous strategy in another professional vertical, Fintech GEO Optimization offers a useful reference case.