When a general counsel asks ChatGPT, “Which attorneys handle cross-border M&A disputes in New York?,” AI systems return a shortlist. That shortlist is a synthesis of every topic the model has come to associate with each attorney. If those associations are wrong, outdated, or thin, your firm loses the matter before a partner ever sees the RFP.

Most large-firm marketing teams have started measuring AI visibility at the firm level through citation counts, brand mentions, and share of voice. That work is essential, and it should continue. But at firms with 200–2,000+ attorneys, firm-level visibility is the foundation, not the finish line. The next layer is the individual lawyer-to-topic association. This piece walks through a repeatable audit to inventory those associations, diagnose gaps against the practices and matters your firm actually wants to win, and convert the findings into a prioritized thought leadership plan.

Why Attorney-Level Topic Association Is the Audit That Actually Moves Revenue

AI systems don’t “rank” attorneys the way Google ranks websites. Instead, they build entity profiles, which are mathematical representations that connect each attorney to clusters of practice areas, matter types, industries, and expertise based on information from bios, bylines, third-party citations, directory data, and case coverage.

Adding Another Layer

Think of firm-level and attorney-level visibility as two different layers of the same problem. The firm layer is table stakes. Without it, nothing else matters.

AI systems don’t distribute firm-level visibility evenly across a practice. They build individual entity profiles for each attorney, and those profiles can diverge significantly from what the firm-level signal suggests. Your firm can be strongly associated with “intellectual property” while individual partners in your highest-growth IP practices are virtually unknown to the model. That divergence is invisible in firm-level data. It only surfaces when you audit at the attorney-topic level.

Similarly, when a prospect asks ChatGPT for “employment attorney experienced in WARN Act compliance,” the model doesn’t return all employment attorneys it knows about. It returns the handful it has most strongly associated with WARN Act expertise. If your firm has a leading WARN Act practice but the partners who handle that work aren’t clearly associated with it in AI systems, you don’t appear. A competing firm with less expertise but better topic association wins the inquiry.

This is the revenue lever that attorney-topic audits expose. It extends firm-level visibility work to ensure the right attorneys are surfaced for the right matters.

What “AI Topic Association” Actually Is

In general, AI topic association is the ability of an AI system to connect related ideas, concepts, or themes across vast datasets, mimicking human semantic memory.

This works through several mechanisms:

• Semantic Mapping — AI links words based on meaning, not just exact spelling or keywords. It recognizes that “M&A” and “merger” and “acquisition” and “deal” are related concepts, and learns which clusters appear together near a specific attorney’s name.
• Contextual Clues — AI understands that the same word means different things in different contexts. “Apple” near “iPhone” signals technology; “apple” near “pie” signals food. Similarly, “patent” near “biotech” activates different concept clusters than “patent” near “software,” and AI learns which clusters associate with which attorneys.
• Graph Relations — AI builds networks connecting entities, attributes, and actions. It learns not just that an attorney handles “M&A,” but that this attorney appears in coverage about healthcare acquisitions, PE-backed transactions, and cross-border deals, building a network of associated concepts.
• Vector Embeddings — AI converts words and concepts into mathematical representations that capture conceptual closeness. Two words with similar embeddings are semantically similar; distance between embeddings reveals how closely concepts relate.
  For law firms, AI topic association is the cluster of practice areas, matter types, industries, and concepts that an AI system has linked to the firm or a specific attorney entity, weighted by how often and how authoritatively those links appear across the model’s training and retrieval sources.

Three Inputs Build the Association

Three types of resources feed the model. Missing signals in any category weaken the association, both at the firm level and the attorney level.

• First-party sources — Your own website. Attorney bios, practice page copy, firm thought leadership bylined to the attorney, speaking engagement pages, client success stories. If your biotech patent partner’s bio says “intellectual property” but never mentions “biotech” or “pharmaceutical,” AI learns a weaker association than it should.
• Third-party sources — Legal trade press, bar association directories, Chambers and Legal 500 profiles, case databases, law firm rankings, podcast and event listings. When Above the Law covers your partner’s pharmaceutical patent work, or Chambers lists them for biotech patents, AI notes and weights those associations heavily.
• Structural signals — Schema markup on your website, internal linking from practice pages to attorney bios, consistent NAP (Name, Address, Phone) and entity data across the open web. If your biotech patent team page doesn’t link internally to individual partner bios with clear schema indicating specialization, you’re missing a ranking signal.

Why Brand Monitoring Tools Miss This

Most brand monitoring and AI citation tracking tools operate at the firm level. They tell you how often your firm gets mentioned and in what contexts. That data is valuable and tells you whether your firm has a foundation of visibility to build on. What it doesn’t tell you is whether that visibility is being attributed to the right attorneys for the right work.

A tool might report: “Your firm is cited 500 times across AI models for ‘intellectual property.'” That’s a strong signal. What you also need to know is: “Your biotech patent partners are cited 320 times for pharma patent work, but your software patent team is cited only 30 times despite being the same size. That $8M practice is invisible at the attorney level.”

The Four-Step Audit Methodology

An attorney-level audit is straightforward to run but requires precision in execution. It works best when marketing, practice leadership, and business development are aligned on priorities before testing begins.

Step 1: Build the Attorney–Target Topic Matrix

Start with an honest inventory of who you want surfaced in AI responses and what you want them known for.

Pull your priority attorney list. Focus on partners and senior counsel in your growth practices, plus rising stars the firm is investing in. Prioritization should come from business development strategy, not from “who’s already visible.”

For each attorney, define 3–7 target topics. These are the precise matters and expertise areas you want them associated with in AI responses. Not generic practice areas, but specific ones aligned with growth priorities and the work you want them hired for.

Examples of appropriately specific target topics:

• “ERISA litigation” (not just “employment litigation”)
• “SaaS commercial contract disputes” (not just “commercial contracts”)
• “pharmaceutical patent prosecution” (not just “intellectual property”)
• “cross-border M&A for PE-backed transactions” (not just “M&A”)
• “SPAC regulatory compliance” (not “regulatory”)

These should come from business development plans, not from what your current bios happen to say. The whole point is to audit the gap between what you want and what AI currently associates with your attorneys.

Step 2: Run a Cross-Model Prompt Set

Single-platform audits can give false readings. ChatGPT may associate your partner with pharmaceutical patents while Claude doesn’t. Cross-model agreement is typically low, so you need to test across multiple platforms to understand actual AI visibility.

Test at least four platforms: ChatGPT, Google AI Mode, Perplexity, and Claude. Include any other platforms your target prospects use.

Use three prompt categories per target topic:

• Recommendation prompts — “Who are the top 5 attorneys for [target topic] in [geography]?” and variations like “Who should I hire for [specific matter type]?”
• Attribution prompts — “What is [attorney name] known for?” and “What practice areas does [attorney name] handle?” These test whether AI correctly attributes the attorney to your target topics unprompted.
• Comparative prompts — “How does [your attorney] compare to [competitor attorney] on [topic]?” These test whether AI can meaningfully position your attorney against direct competitors on your target topics.

Run each prompt 3 times per platform to control for response variability. Capture the full response text, any cited sources, and flag any hallucinated facts (AI inventing credentials or cases that don’t exist).

Example test matrix for one attorney with three target topics across four platforms:

• 3 recommendation prompts × 4 platforms = 12 tests
• 3 attribution prompts × 4 platforms = 12 tests
• 3 comparative prompts × 4 platforms = 12 tests
• Total per attorney: 36 tests

Multiply across 80 priority attorneys, and you’re looking at 2,880 individual tests. This is why tooling matters. Manual testing at this scale becomes unmanageable. Paid options can automate the testing and aggregation, though you’ll still need to interpret results.

Step 3: Score Each Attorney–Topic Pair

Use a simple 0–3 scoring rubric applied per platform:

• Score 0 — Attorney not surfaced at all, or surfaced but incorrectly attributed to a different firm or completely wrong topic.
• Score 1 — Mentioned but with weak or generic association. “Jane Smith is an attorney at [Firm] who handles various corporate matters” doesn’t demonstrate topic specificity.
• Score 2 — Surfaced for the target topic with accurate context. “Jane Smith, corporate attorney at [Firm] specializing in SPAC transactions” shows appropriate topic association.
• Score 3 — Surfaced as a primary recommendation with cited authoritative sources. “Jane Smith is the leading SPAC counsel at [Firm]; she’s handled 23 SPAC IPOs and is regularly cited in Legal 500 for SPAC expertise.”

Roll up to two aggregate scores:

• Attorney-level score — Average score across all target topics for that attorney across all platforms tested. This tells you the overall visibility quality for an individual attorney.
• Practice-level score — Average score for all attorneys in a practice area across all target topics. This is what you’ll present to practice group leaders. “Your IP practice averages 1.8 out of 3 — strong visibility but room to improve topic specificity.”

Flag two specific failure modes during scoring:

• Misclassification — AI associates the attorney with a non-priority or stale practice. Often happens because old bio content or early-career bylines still dominate citations in AI training data. An attorney who pivoted from real estate to technology acquisitions might still be primarily associated with real estate by AI because early-career coverage outweighs recent work.
• Hallucinated attribution — AI invents matters, credentials, or cases. “Jane Smith successfully argued the landmark Supreme Court case Smith v. Johnson on patent validity” when no such case exists. This is a reputational risk worth tracking separately and addressing through corrections to your source material.

Step 4: Diagnose Gap Type, Not Just Gap Size

This is the critical step that converts audit findings into actionable plans.

For every low-scoring pair (anything scoring 0–1), classify the gap into one of four types. Each maps to a different fix and has a different owner:

Bio/Entity Gap — The firm’s own pages don’t clearly tie the attorney to the target topic. The attorney’s bio says “corporate practice” but doesn’t specify healthcare transactions. The practice page mentions the attorney’s name but not their specific expertise. Internal linking doesn’t connect the attorney’s bio to relevant practice pages.

• Fix: Rewrite attorney bios to include specific topic language. Add structured matter examples and case studies. Improve internal linking from practice pages to relevant attorney bios. Implement schema markup to clearly indicate attorney-to-topic relationships.
• Owner: Web/marketing operations and content strategy.

Content Gap — There’s no thought leadership bylined to this attorney on the target topic. The firm hasn’t published content establishing them as an authority on the specific work you want them known for. This is especially common for partners being repositioned or developed for new practice areas.

• Fix: Commission bylined content (articles, guides, webinars, white papers) from this attorney focused on the target topic. Distribute to authoritative third-party venues. Ensure each piece includes a rich author bio linking back to practice pages.
• Owner: Communications/editorial team.

Authority Gap — Content exists for this attorney on the target topic, but no third-party citations or coverage support it. The attorney has published articles or speaks at events, but no legal trade press, industry publications, or professional directories mention them in that capacity. AI relies on external validation; self-published content alone doesn’t build authority.

• Fix: Digital PR campaigns targeting trade publications that cover the attorney’s specialty. Speaking placements at industry conferences. Contributed bylines in Legal 500, Chambers, and industry trade publications. Guest appearances on legal podcasts. Bar association leadership roles.
• Owner: Public relations team.

Structural Gap — Schema markup is missing or incorrect. Entity data is inconsistent across the web (attorney listed as “Jane Smith” on the firm site but “Dr. J. Smith, PhD” on directories). NAP inconsistencies prevent entity recognition. Internal linking is sparse.

• Fix: Implement proper schema markup for attorney entities. Audit and correct NAP data across all directories and the open web. Improve page hierarchy and internal linking. Ensure consistent entity naming across all properties.
• Owner: Technical SEO and web operations.

This classification is the bridge from “we have problems” to “here’s what we do about them.” Without it, the output is a list of low scores with no clear owner or solution path. With it, you have a diagnostic that drives action.

Converting the Gap Map Into a Prioritized Thought Leadership Plan

Don’t optimize for the practice with the most gaps. Optimize for the practice where closing the gap will have the biggest business impact.
Five partners in a $50M M&A practice with bio/entity gaps represent more revenue potential than fifty associates in a $5M emerging practice with content gaps. Build a simple matrix ranking target attorneys and topics by practice revenue, new business potential, gap severity, and fix complexity. That combination surfaces high-impact, achievable wins.

Build a 90-Day Plan With Three Parallel Workstreams

Once priorities are sequenced, the work runs across three parallel workstreams that define timeline, owner, and type of impact.

• Quick Wins (Weeks 1–4) — Bio rewrites, schema markup, internal linking fixes, and NAP corrections. These are foundational and often shift AI associations within one to two retraining cycles. You can’t close authority gaps if entity gaps mean the attorney isn’t being recognized in the first place.
• Editorial Build (Weeks 1–12) — Bylined content against the highest-revenue topic gaps, placed in authoritative third-party venues — Legal 500, Law360, Above the Law, industry trade publications. One piece per attorney per quarter minimum, highly specific to the target topic, with a link back to their bio or practice page.
• Authority Earn (Ongoing) — Digital PR, speaking placements, and contributed bylines in the publications AI systems cite most. Research which outlets appear most frequently in AI responses for your target topics, then build a PR strategy around placing your attorneys there.

Don’t forget to assign clear ownership in the process. Without named owners and deadlines, audit findings don’t translate into action. Re-run the attorney-topic audit quarterly to monitor progress.

What This Audit Tells You That Nothing Else Will

A completed attorney-topic audit answers four critical questions that drive strategy:

Whether your highest-revenue partners are actually visible in AI for the work the firm wants. You may discover your top M&A partner is invisible in AI for SPAC work despite handling dozens of SPAC IPOs. Or that a partner you’re trying to develop for healthcare transactions is already being positioned correctly in AI despite limited recent bylines. The audit surfaces reality, not assumptions.

Which practice groups are losing opportunities at the AI-shortlist stage. If your litigation practice is invisible in AI for securities class action defense but competitors show up frequently, that’s a practice losing matters before partners even know about the engagement. Knowing where these gaps exist lets you prioritize investment.

Where thought leadership investment will move the needle versus where it’s noise. Many firms publish content that doesn’t move AI visibility. This audit reveals which bylines, speaking placements, and PR investments actually impact how AI associates your attorneys with target topics. You can stop publishing content nobody sees and redirect budget to channels that actually shape AI perception.

A reusable scorecard for presentation to firm leadership. “Our IP practice averages 2.1 out of 3 on target topic visibility; M&A averages 1.4. We’re investing in pharmaceutical patent bylines and SPAC transaction coverage to close those gaps. Projected impact: 15–20% increase in AI visibility for these practices in Q3.”

This scorecard is executive-friendly. It’s a clear diagnostic and action plan tied to business outcomes without technical jargon.

Getting Started

For large firms, firm-level visibility and attorney-level topic association are essential. Getting your firm cited is the foundation. Making sure the right attorneys are cited for the right work builds on that foundation to increase new business. This is how you ensure that when a general counsel asks ChatGPT for a recommendation, your firm’s name appears on their shortlist, and the right attorney gets the call.

If you want a benchmark before running the audit internally, the 9Sail Digital Visibility Index provides a firm-level read on where AI visibility stands today. If you need help executing and monitoring attorney visibility on an ongoing basis, our Law Firm GEO team is here to help.