AI-powered search assistants,ChatGPT, Claude, Gemini, and Perplexity,do not recommend attorneys randomly. They synthesize information from across the internet to identify firms that appear most authoritative, relevant, and trustworthy for a given query. After analyzing hundreds of AI-generated attorney recommendations across multiple practice areas and geographic markets, Pinniq Legal has identified the specific factors that predict which firms get recommended and which get skipped. This article is a comprehensive breakdown of those factors, how each AI platform likely weighs them, and what law firms can do to strengthen their position on each one.

How AI Decides Who to Recommend

Before diving into individual factors, it helps to understand how AI recommendation works at a high level. When a user asks ChatGPT “who is the best personal injury lawyer in Nassau County,” the AI does not consult a ranked list. Instead, it follows a process that looks roughly like this:

  1. Query interpretation. The AI identifies the practice area (personal injury), geographic scope (Nassau County), and user intent (find and hire an attorney).
  2. Information retrieval. Depending on the platform, it draws from training data (all platforms), live web search (ChatGPT with web search, Perplexity, Gemini), or both.
  3. Entity identification. The AI identifies law firms mentioned in its source data that match the practice area and geography.
  4. Ranking and selection. From the pool of identified firms, the AI selects 3-5 to recommend, ordering them by a combination of authority, relevance, and confidence signals.
  5. Response generation. The AI generates a natural language response that names the selected firms and provides context about each one.

The factors described below influence steps 3 and 4,whether the AI identifies your firm as a candidate, and whether it ranks your firm high enough to include in its response. These are not official ranking factors published by OpenAI, Anthropic, Google, or Perplexity. They are patterns we have identified through systematic analysis of AI outputs across thousands of queries.

Citation Frequency Across Authoritative Sources

The single most predictive factor for AI recommendations is how frequently a firm is cited across authoritative, independent sources. A firm that appears on Avvo, FindLaw, Justia, Martindale-Hubbell, Super Lawyers, Lawyers.com, the state bar directory, local chamber of commerce, news outlets, and court records has a dramatically higher probability of being recommended than a firm that only appears on its own website.

AI models are trained on, or search through, vast amounts of internet data. When a firm name appears repeatedly across diverse, authoritative sources, the AI develops high confidence that the firm is a real, established entity in a specific practice area and geography. This confidence translates directly into recommendations.

What to do about it:

  • Claim and complete profiles on every major legal directory: Avvo, FindLaw, Justia, Martindale-Hubbell, Super Lawyers, Lawyers.com, and your state bar directory.
  • Ensure consistency across all listings,same firm name, same address, same phone number, same practice areas.
  • Seek earned media mentions in local news outlets, legal publications, and industry blogs.
  • Contribute to legal publications, bar association newsletters, or law school journals to create authoritative citations outside of directories.

Firms with profiles on 8+ legal directories score, on average, 3x higher in AI visibility than firms with profiles on 2 or fewer. This is the highest-leverage factor for most firms because it can be addressed systematically within weeks.

Review Volume, Recency, and Content

Reviews are the second most impactful factor. AI platforms use review data as a proxy for reputation and client satisfaction. But not all reviews carry equal weight. Three dimensions matter:

Volume. More reviews signal a more established practice. A firm with 200 Google reviews has stronger signal data than a firm with 12. AI models interpret high review volume as evidence of an active, client-facing practice.

Recency. Recent reviews matter more than old ones. A firm that received 50 reviews in the past 12 months signals ongoing client activity. A firm whose most recent review is from 2022 may appear inactive. AI platforms, especially Perplexity and Gemini, which access live data, weight recent reviews more heavily.

Content. The text of reviews matters as much as the star rating. Reviews that mention specific practice areas (“handled my car accident case”), specific outcomes (“won a substantial settlement”), and specific locations (“in Mineola”) give AI models rich entity data to connect your firm with relevant queries. Generic five-star reviews with no text provide volume but limited signal value.

What to do about it:

  • Implement a systematic review request process for every closed case.
  • Provide clients with gentle guidance: “If you’re comfortable, mentioning the type of case and location helps other people in similar situations find us.”
  • Respond to every review, positive or negative. AI models can see response patterns, and active engagement signals attentiveness.
  • Focus on Google reviews first (they are the most widely accessed by AI platforms), then Avvo and other legal-specific review sites.

Practice Area Specificity

AI models attempt to match the specificity of the user’s query with the specificity of the firm’s documented expertise. When a user asks for a “medical malpractice lawyer,” the AI looks for firms that have explicitly documented medical malpractice as a practice area,not just “personal injury.”

This factor has two dimensions:

Website content specificity. A firm with a detailed page about medical malpractice,covering types of malpractice, relevant statutes, case process, and results,sends much stronger practice-specificity signals than a firm with a bullet point on a general services page. AI models parse your website content to determine what you actually practice, and depth of coverage is a strong indicator.

Directory and profile specificity. Your Avvo, FindLaw, and bar association profiles should list specific sub-types, not just broad categories. “Personal Injury,Medical Malpractice, Truck Accidents, Workplace Injuries” is more valuable than “Personal Injury.” This structured data helps AI models match your firm to sub-type-specific queries.

What to do about it:

  • Create dedicated pages for each practice sub-type you handle. Each page should have unique, substantive content,not boilerplate.
  • Include FAQ sections on practice area pages. AI models frequently pull from FAQ content because it directly matches question-format queries.
  • Update directory profiles to list all specific sub-types.
  • Publish case results (anonymized) organized by practice sub-type to demonstrate real experience.
Key Takeaway: AI platforms match query specificity with content specificity. A firm that has detailed, substantive pages for each practice sub-type will outperform a firm with a single generic practice area page,even if the generic firm is larger and better known.

Geographic Relevance Signals

When a query includes a geographic term, AI models look for signals that connect a firm to that specific location. The strongest geographic signals are:

Physical address. An office address in or near the queried location is the most direct geographic signal. This is particularly important for Gemini, which accesses Google Maps and Google Business Profile data.

Service area declarations. Your website, directory profiles, and Google Business Profile should explicitly state which cities, towns, and counties you serve. “Serving Huntington, Smithtown, and surrounding Suffolk County communities” gives AI clear geographic scope.

Local content. Content that references local courts, local judges, local landmarks, or local legal issues creates town-level geographic signals. A page about “filing a personal injury claim in Nassau County Supreme Court” ties your firm to Nassau County in a way that a generic “how to file a personal injury claim” page does not.

Local citations. Mentions of your firm in local news outlets, community organizations, or local business directories create geographic association signals that AI models can detect.

What to do about it:

  • Ensure your Google Business Profile is verified and complete with accurate address, service area, and hours.
  • Create location-specific pages for each town or county you serve.
  • Reference local courts, procedures, and community details in your content.
  • Engage with local organizations and seek mentions in local media.

Website Authority and Structured Data

Your website’s technical quality and structured data influence how AI models perceive and process your firm’s information.

Structured data (Schema.org markup). Implementing LegalService, Attorney, and LocalBusiness schema markup gives AI models structured, machine-readable information about your firm. This includes practice areas, attorney credentials, office locations, and contact information. Firms with comprehensive schema markup are easier for AI to parse and index.

Content depth and freshness. AI models evaluate the depth and recency of your website content. A firm with 50 detailed, recently-updated practice area and blog pages signals active authority. A firm with 5 static pages from 2019 signals a dormant web presence.

Domain authority and backlinks. While AI models are not search engine crawlers, they are influenced by the same underlying data. Websites with strong backlink profiles from authoritative sources (legal publications, bar associations, university law schools) tend to appear more frequently in AI responses. This is likely because the content linking to these sites becomes part of the AI’s training or search data.

HTTPS and technical basics. Security certificates, mobile responsiveness, and fast load times are baseline requirements. AI platforms that perform web searches may deprioritize results from insecure or poorly-built websites.

What to do about it:

  • Implement Schema.org markup for your firm (LegalService type), individual attorneys (Person type), and office locations (LocalBusiness type).
  • Publish new content regularly,at minimum monthly,to signal an active, current practice.
  • Build backlinks through guest posts, speaking engagements, and media outreach.
  • Ensure your website passes Core Web Vitals and basic technical SEO audits.

Awards, Recognition, and Professional Standing

AI models frequently cite awards and recognition when recommending firms. Phrases like “named a Super Lawyer,” “rated AV by Martindale-Hubbell,” or “Best Lawyers in America” appear regularly in AI responses. These signals serve as trust indicators that help AI models differentiate between firms with similar practice areas and geographic coverage.

Awards that matter for AI:

  • Super Lawyers and Rising Stars
  • Best Lawyers in America
  • Martindale-Hubbell AV rating
  • State and local bar association awards
  • Avvo ratings (particularly 9.0+ scores)
  • Legal publication rankings (U.S. News Best Law Firms, Chambers, etc.)

Bar association standing is another factor. AI models can access state bar directories and may verify that attorneys are in good standing. Active membership in bar sections related to your practice area (Personal Injury Section, Family Law Section, etc.) creates additional association signals.

What to do about it:

  • List all awards and recognition prominently on your website,on attorney bio pages, the home page, and practice area pages.
  • Include awards in your directory profiles where possible.
  • Apply for relevant awards and recognition programs. Many are self-nominated and have straightforward qualification criteria.
  • Ensure your bar association profile is current and lists active section memberships.
Key Takeaway: AI recommendation is not driven by any single factor. It is the cumulative effect of citation frequency, review signals, practice specificity, geographic relevance, website authority, and professional recognition. Firms that systematically build signals across all six dimensions achieve the strongest AI visibility. For a personalized analysis, consider Pinniq Legal’s managed services.

How Each AI Platform Weighs These Factors

While all four major AI platforms consider the same general categories of signals, they weight them differently based on their architecture and data access.

ChatGPT (OpenAI) uses a combination of training data and optional web search. Its recommendations lean heavily on entity recognition,firms that are widely cited across the internet are more likely to be recognized as relevant entities. Website content and structured data are also important because ChatGPT’s web search tool can access and parse your site directly.

Claude (Anthropic) relies entirely on training data without live web search. This means its recommendations reflect a firm’s historical online presence rather than current state. Firms with long-standing, well-cited presences perform best on Claude. Recent changes (new reviews, new content) take longer to appear in Claude’s recommendations because they must be incorporated into future training data.

Gemini (Google) uses Google Search grounding, which gives it direct access to Google’s index, Google Business Profiles, Google Reviews, and Google Maps data. This makes Gemini the most responsive to traditional local SEO signals. Firms with strong Google Business Profiles, high Google review counts, and good local search rankings tend to perform disproportionately well on Gemini.

Perplexity performs live web searches for every query, making it the most dynamic and current of the four platforms. Perplexity explicitly cites its sources, which means firms that appear on authoritative, well-ranked pages get credit. Recent content and fresh reviews can influence Perplexity recommendations within days of publication, not months.

The strategic implication is clear: optimizing for all four platforms requires a multi-faceted approach. Google Business Profile optimization helps with Gemini. Directory citations help with all platforms. Fresh content helps with Perplexity and ChatGPT. Long-term authority building helps with Claude. There is no single shortcut that works everywhere,which is exactly why a systematic GEO strategy matters.