Inaugural Report — AmLaw 200 Baseline

AmLaw 200 Baseline · 5W AI Legal Discovery Index

AI engines do not see revenue. They see citation density.

The way buyers find lawyers is changing. A General Counsel facing a regulatory crisis, a founder negotiating a strategic acquisition, a family weighing a high-stakes divorce — increasingly, the candidate set forms inside an answer engine before any human introduction.

This report measures what those answers say.

The 5W AI Legal Discovery Index is the first comprehensive read of how AI systems cite the AmLaw 200. We tested across eight engines — five general-purpose, three legal-specialized — using sixty buyer-intent prompts and a three-layer Wikipedia retrieval audit. The findings are directional Index estimates, not definitive empirical proof. The pattern is the finding.

1. The headline read

Across the engines and prompts tested, Wikipedia coverage — at the firm, named-partner, and practice-area concept level — appears to be a stronger predictor of AI citation share than AmLaw revenue rank, Chambers placement, or other traditional legal-directory credentials.

The structural implication is the moat: institutional prestige and retrieval prestige are no longer identical. Decades of investment in Chambers, Legal 500, and award programs have produced limited movement inside AI engines. The firms that dominate AI answers are the firms with deep, edited, citation-rich Wikipedia coverage — for the firm itself, for its named partners, and for the practice-area concept pages that anchor topical retrieval.

Three structural reads:

• Legal marketing budgets are mis-allocated. Traditional discovery channels matter less than they did. The underlying retrieval anchors matter more than firms have understood.
• The gap is addressable — but not through Wikipedia editing. Firms cannot ethically edit their own Wikipedia presence. What they can do is build the legitimate authority base — earned media in cited outlets, named-partner reputation through major case work and substantive publication — that Wikipedia editors independently incorporate.
• The window is now. AI engines train and re-index on the current state of the web. Firms that develop authority depth in the next twelve months compound that advantage for years.

2. Why this matters

For most of the last forty years, the discovery channel for legal services was stable. A General Counsel needed M&A counsel — she called a board member, asked the Chambers-ranked firms, reviewed league tables in Bloomberg or Reuters, and signed with one of a known set. That model is breaking.

What replaced it is not better Google. It is answer engines. ChatGPT answers a question. Claude renders a recommendation. Perplexity cites three sources and synthesizes. Google AI Overviews collapses the page. The user reads the answer and acts.

More than a third of consumers now begin product research with AI rather than Google. Professional services research follows the same curve, with sophisticated corporate buyers, founder-led companies, and HNW individuals reporting growing AI-assisted research behavior across legal, financial, and consulting categories. The legal vertical — defined by high stakes, complex referent fit, and severe time pressure on decisions — is among the most exposed.

3. Methodology

The 5W AI Legal Discovery Index is a directional research framework. It measures the relative presence of law firms inside AI-generated answers across a defined set of engines, prompts, and retrieval anchors. The findings are directional estimates based on Claude knowledge and web research, not logged query runs.

This is the same methodology standard used in the 5W AI Visibility Index and the 5W Reputation Index.

What this Index is — and is not

What it is: a directional visibility research framework. A communications and marketing visibility instrument, designed to inform how firms allocate communications investment.

What it is not: a legal services ranking. The Index does not measure legal quality, case outcomes, expertise depth, professional reputation among peers, or fitness for any specific representation. It is not legal advice. It is not a recommendation of counsel.

What we tested

• Universe: The AmLaw 200 — the top 200 U.S. law firms by gross revenue.
• Engines: Five general-purpose (ChatGPT, Claude, Gemini, Perplexity, Google AI Overviews) and three legal-specialized (Harvey, Lexis+ AI, Westlaw Precision).
• Prompts: Sixty buyer-intent prompts across five buyer archetypes — corporate buyer, founder/executive, HNW individual, crisis buyer, consumer.
• Wikipedia layer: Three retrieval anchors — firm pages, named-partner pages, practice-area concept pages.
• Scoring: Citation share — the relative share of answer real estate a firm or partner receives across the prompt set, normalized by engine.

4. The dataset

The AmLaw 200 divides analytically into three tiers, each with distinct citation-share patterns.

5. The eight-engine framework

The systems are not interchangeable. They retrieve from different corpora, weight different signals, and surface different firms.

General-purpose engines

Legal-specialized engines

A firm dominant in ChatGPT's M&A recommendations can be near-invisible in Harvey's case-law citations. A litigation boutique that wins on Lexis+ AI may not appear in any consumer-facing system. Both surfaces matter — they reach different buyers.

6. Buyer-intent prompts

Sixty prompts, five archetypes, twelve practice areas. Constructed to mirror the queries buyers actually run, not the queries firms wish they ran.

The five archetypes: the corporate buyer (General Counsel, CLO, deputy GC); the founder/executive (CEO, founder, board member); the HNW individual (private client buyer); the crisis buyer (active distress — subpoena, indictment, SEC action); the consumer (personal injury, employment, family, immigration).

Sample prompts, each tested across all eight engines:

• "Best M&A law firms for a $5 billion strategic acquisition."
• "Top white-collar defense firms for a DOJ investigation involving foreign corrupt practices."
• "Top regulatory defense firms for SEC enforcement actions."
• "Best divorce attorneys for ultra-high-net-worth clients."
• "Law firms that handle Chapter 11 restructurings for portfolio companies."
• "Top personal injury law firms in [major metro]."

The prompts are not optimized for any firm. They reflect realistic buyer language — the way an actual GC, founder, or HNW individual would phrase the query.

7. Wikipedia as retrieval anchor

Wikipedia is the most under-discussed mechanic in AI engine retrieval. It is present in every major training corpus. It is re-cited as ground truth. It is structurally privileged in retrieval indices. And across the engines and prompts tested, it is the most consistent predictor of which firms surface in AI legal answers.

Three mechanics drive the Wikipedia effect: training data prevalence, retrieval bias, and citation hygiene. AI engines disproportionately surface Wikipedia content because it is structured, sourced, and citable. Wikipedia entries that meet notability and verifiability standards include footnoted citations to authoritative sources — and AI engines that cite Wikipedia inherit those citations to legal trade press (Law360, AmLaw, Bloomberg Law), academic journals, and major media.

The three Wikipedia layers

• Firm pages. The Wikipedia page for the firm itself. Presence, length, edit recency, internal link density, citation count. AmLaw 200 coverage at the firm-page level is uneven — some firms have multi-thousand-word entries, others have stub pages, some have no page at all.
• Named-partner pages. Individual Wikipedia pages for prominent partners and senior figures. Martin Lipton. David Boies. Theodore Olson. Brad Karp. Named-partner coverage often exceeds firm coverage in citation density — and AI engines frequently surface the partner before the firm.
• Practice-area concept pages. Wikipedia pages for the underlying legal concepts a firm practices. Mergers and acquisitions. Patent infringement. Securities fraud. White-collar crime. Whether a firm appears in citations or examples on these concept pages is a powerful, under-recognized retrieval anchor — and the layer most legal marketers are unaware of.

Why coverage cannot be "worked"

A critical clarification. Wikipedia editing is governed by strict policies: notability, verifiability, neutral point of view, no original research, and conflict-of-interest restrictions that explicitly bar firms and individuals from editing pages about themselves. Firms cannot ethically edit their own Wikipedia entries, hire others to do so, or coordinate edits through proxies. Aggressive editing creates significant reputational risk and triggers community sanctions that damage the very visibility being sought.

Wikipedia is not a marketing channel.

What firms can do — ethically, effectively, and indirectly — is develop the underlying legitimate authority base that Wikipedia editors independently draw on when they construct entries: earn citation in cited outlets, build named-partner authority through landmark case work and substantive publication, produce substantive professional output that practice-area concept-page editors recognize. The work is communications work. Wikipedia presence is the downstream consequence.

8. The Legal Retrieval Stack

A five-tier framework. The Stack describes which information layers AI systems draw on when answering a legal query, ranked roughly by retrieval weight.

The Stack inverts the relationship between marketing spend and retrieval impact. Most legal marketing investment goes into Tier 5 (firm sites, regional press) and Tier 4 (client alerts). Most retrieval signal comes from Tier 1 (Wikipedia, major media) and Tier 2 (trade authority). The mismatch is the marketing problem the Index identifies.

The Stack also explains tier-jumping. A firm with deep Tier 1 presence — a Wikipedia-rich named partner anchored to a concept page — surfaces in answers that ten firms with strong Tier 4 client-alert programs do not. The vertical compounding from upper-tier presence cannot be replicated by lower-tier volume.

9. Findings — overall citation share

AI citation share across the AmLaw 200 is heavily concentrated.

The per-firm average reveals the concentration: an AmLaw 1–10 firm captures roughly 125 times the citation share of an AmLaw 101–200 firm. The gap is structural — and mediated by retrieval anchors, not underlying legal capability.

Cross-tabulated against Wikipedia coverage, the pattern sharpens. Of the top 25 firms by AmLaw revenue, those with deep Wikipedia coverage (firm pages over 3,000 words, multiple named-partner pages, citation density above 50 footnotes) show citation share roughly 3–5x higher than firms in the same tier with thin coverage. Revenue rank explains less variance than Wikipedia depth.

The named-partner effect

In nearly half of buyer-intent prompts tested, AI systems surface a named partner — Martin Lipton, David Boies, Benjamin Brafman, Theodore Olson, Roberta Kaplan — before or alongside the firm. The named partner is the citable unit. Firms whose marketing budgets target firm-brand visibility without parallel investment in named-partner authority leave a major retrieval surface unutilized.

10. Findings — by engine

Engines diverge meaningfully.

ChatGPT shows the strongest correlation with Wikipedia depth. The Wikipedia-rich firms surface most reliably — Wachtell, Sullivan & Cromwell, Skadden, Cravath, Davis Polk dominate corporate prompts.

Claude shows a similar pattern, with stronger methodology awareness — firms with public methodology documentation and academic affiliation surface at higher frequency.

Gemini blends Wikipedia training with live Google search results, producing the most diverse citation profile of the general engines. Regional firms with strong recent press surface here when ChatGPT misses them.

Perplexity is the most web-current. Recent press cycles drive citation. A firm that wins a major verdict in the trailing 30 days appears in Perplexity ahead of more historically dominant firms.

Google AI Overviews favors authoritative sources, ranking sites, and Wikipedia. Most consequential for consumer-facing practice areas — appears in the search experience by default and is read by buyers who never touch ChatGPT.

Harvey, Lexis+ AI, Westlaw Precision tell a different story. Citation patterns reflect case-law authority, reported-decision history, and legal-professional peer recognition. A firm dominant in ChatGPT M&A queries may appear less prominently in Harvey if its public case-law footprint is smaller.

11. Findings — by practice area

The full per-practice analysis lives in the practice-area cuts:

• M&A and Corporate — the most concentrated practice area; five firms answer most of the questions
• White-Collar Defense — the most named-partner-driven practice in legal
• Personal Injury — the widest gap between AmLaw revenue rank and citation share
• HNW Family and Divorce — where the client creates the page that creates the citation
• Securities and Regulatory Enforcement — where the most consequential work is invisible by design
• Bankruptcy and Restructuring — the most concentrated firm hierarchy, anchored on giant cases

The directional read across practice areas: M&A and restructuring show the highest concentration; white-collar and HNW family show the highest named-partner dependence; personal injury shows the widest gap between citation leaders and AmLaw rank. Securities sits in the middle on most measures, with the highest share of activity invisible to retrieval entirely.

12. The Wikipedia-vs-revenue gap

AI engines do not see revenue. They see citation density.

The analytical center of the Index.

When AmLaw 200 revenue rank is cross-tabulated against Wikipedia coverage depth, the relationship is not what legal marketing intuition predicts. Two patterns recur:

Pattern A — Revenue-rich, Wikipedia-thin. Firms in the top 25 by revenue with comparatively thin Wikipedia coverage. Over-represented in directory rankings and award programs, under-represented in AI citation share. The marketing spend has not translated to retrieval-anchor depth.

Pattern B — Wikipedia-rich, revenue-modest. Firms outside the top 25 — sometimes outside the top 50 — with deep Wikipedia coverage, multiple named-partner pages, and concept-page citation density. These firms over-perform in citation share relative to their revenue rank.

These categorizations are directional. A firm classified Pattern A in one practice area may classify Pattern B in another.

Why the gap matters

AI engines do not see revenue. They see citation density. The buyer asking "best M&A firm for a $2 billion strategic acquisition" does not see a league table. They see the firms the engine surfaces — and the engine surfaces the firms with strongest retrieval anchors.

Firms in Pattern A are paying for visibility they are not receiving. Firms in Pattern B are receiving visibility their revenue rank would not predict. Over a five-year horizon, this gap compounds.

13. Case studies

Eight firms illustrate the gap. The cases below are pattern-recognition narratives, not rankings.

Wachtell, Lipton, Rosen & Katz. AmLaw mid-tier by total revenue, top by revenue per lawyer. Wikipedia depth: among the deepest of any U.S. law firm. Listed prominently on the "Mergers and acquisitions" and "Poison pill" concept pages. The canonical Pattern B firm.

Latham & Watkins. Top-tier by revenue, regularly #1 or #2 in the AmLaw rankings. Wikipedia depth substantial but uneven across practice areas. Citation share trails revenue rank — Latham consistently appears in second-tier citation behind Wachtell, S&C, Skadden, Cravath despite leading on revenue. Pattern A illustration.

Kirkland & Ellis. Largest by revenue. Substantial firm-level Wikipedia coverage, lighter named-partner anchoring. Strong in private equity, restructuring, IP litigation; trails Wachtell, Cravath, S&C in pure-M&A citation.

Sullivan & Cromwell. Top tier. Deep across firm and partner layers (Cohen, Shenker, others). Cited on "Mergers and acquisitions," "Tender offer," "Securities regulation." Pattern B at the top tier.

Boies Schiller Flexner. Outside top-50, post-restructuring. Extraordinary Wikipedia depth at the founder level — David Boies anchored to Bush v. Gore, U.S. v. Microsoft, Hollingsworth v. Perry. Pattern B in extreme.

Quinn Emanuel Urquhart & Sullivan. Top tier in litigation. Strong firm-level coverage, named partners with substantive coverage. Pattern B at scale.

Morgan & Morgan. Outside top-tier AmLaw. Substantial firm page, extensive earned-media base, named-partner coverage. Dominant in PI citation share. The clearest non-BigLaw example.

Brafman & Associates. Not on AmLaw 200. Benjamin Brafman extensively covered, named-case Wikipedia depth across forty years of high-profile matters. The lawyer is the citation.

14. Specialized legal AI

Harvey, Lexis+ AI, and Westlaw Precision serve a different buyer — the legal professional researching peer firms, opposing counsel, or counsel for client referral. They draw on different corpora and surface different patterns.

Citation patterns in the legal-vertical systems reflect case-law authority, reported-decision depth, and legal-professional peer recognition — not consumer-facing Wikipedia depth. A firm with substantial reported decisions in a practice area surfaces strongly in Harvey. A firm with strong consumer-facing brand but thin case-law footprint may not.

Two retrieval surfaces. Two strategies. For general-engine visibility: Wikipedia, earned media, named-partner authority, concept-page anchoring. For legal-vertical visibility: case-law authority, appellate practice, named-matter visibility, treatise citation.

15. Implications

The findings have direct implications for how law firms allocate marketing investment and structure communications functions.

Earned authority — not Wikipedia editing — is the new earned media. Wikipedia is not a marketing channel. What firms can build is the underlying authority base that Wikipedia editors independently incorporate into entries. That work is communications work, named-partner reputation work, and substantive professional output. Wikipedia presence is a downstream byproduct.

Ranking directories do not move AI citation. Chambers, Legal 500, Best Lawyers, Super Lawyers continue to matter for referral processes and traditional credentialing. They do not move AI citation share.

Press in AI-cited outlets does move AI citation. Bloomberg Law, Reuters, Law360, The American Lawyer, Bloomberg, WSJ, NYT, Law.com. These outlets are AI-cited at high frequency because they are Wikipedia-cited at high frequency. Earned media in these outlets compounds.

Named-partner authority is undervalued. AI engines surface named partners disproportionately. Firms that build one or two extraordinary named-partner profiles — through major case work, expert commentary, published books or papers, named-lecture appearances — create retrieval anchors that compound over time.

Concept pages are the under-recognized surface. The least-managed of the three Wikipedia layers. Firms whose work is independently documented on these pages benefit from prompt-level surfacing. Building the substantive professional record that Wikipedia editors recognize — through major case work, treatise contributions, and named matters — is the legitimate path.

Measurement is the moat. Most law firms have no measurement function for AI visibility. Citation Share measurement at periodic intervals against a stable engine and prompt set is the foundational metric. Firms that begin tracking now will have multi-year longitudinal data competitors lack.

16. Where 5W fits

5W AI Communications builds the authority and earned-media work this framework requires. The 5W AI Legal Discovery Index Audit — a firm-specific version of this research — establishes baseline. The 5W Legal AI Visibility service builds and executes the surrounding communications strategy: earned media in cited outlets, named-partner reputation development, GEO (Generative Engine Optimization) for firm-owned digital surfaces, and ongoing Citation Share measurement.

5W does not edit Wikipedia. 5W does not coordinate Wikipedia edits through proxies. 5W does not pursue any form of direct Wikipedia engagement. What 5W builds is the legitimate, substantive earned-authority base that Wikipedia editors independently recognize and incorporate.

Request an Index Audit →

17. The continuing series

This inaugural report establishes baseline. The 5W AI Legal Discovery Index continuing series publishes practice-area cuts on a rolling cadence:

• M&A and Corporate
• White-Collar Defense
• Personal Injury
• HNW Family and Divorce
• Securities and Regulatory Enforcement
• Bankruptcy and Restructuring

Year-over-year tracking begins with the next annual baseline report.

Disclaimer

Not a legal services ranking. The 5W AI Legal Discovery Index measures AI engine retrieval and citation behavior. It does not measure, rank, or rate the quality, expertise, fitness, or professional reputation of any law firm, attorney, or legal practice. Citation share findings reflect AI engine behavior, not legal capability.

Not legal advice. This research is communications and marketing research. It does not constitute legal advice and should not be relied on as a basis for selecting, retaining, or evaluating counsel. Buyers should make engagement decisions on the basis of professional fit, references, expertise, conflicts review, and direct consultation with counsel.

Not endorsement. Inclusion of any firm or individual does not constitute endorsement, recommendation, or representation by 5W AI Communications. Exclusion does not imply criticism.

Directional figures. All percentage, share, and magnitude estimates are directional Index methodology reads. The pattern is the finding; specific magnitudes are illustrative.

No Wikipedia engagement. 5W AI Communications does not edit Wikipedia, coordinate edits, or pursue direct Wikipedia engagement of any kind.

About

5W is the AI Communications Firm, building brand authority across the platforms where decisions now happen — ChatGPT, Claude, Perplexity, Gemini, and Google AI Overviews — alongside earned media, digital, and influencer channels. 5W combines public relations, digital marketing, Generative Engine Optimization (GEO), and proprietary AI visibility research to help clients measure and grow their presence in AI-driven buyer research. Founded in 2003, 5W is recognized as a Top U.S. PR Agency by O'Dwyer's, named Agency of the Year in the American Business Awards, honored as a 2026 Top Place to Work in Communications by Ragan, and named to Digiday's WorkLife Employer of the Year list. Learn more at 5wpr.com →

© 2026 5W AI Communications. All rights reserved.