Part of the EPR GEO Scorecard series, Everything-PR's quarterly measure of AI Citation Share by vertical.
Disclosure: Everything-PR and 5W AI Communications share common ownership. Everything-PR reports independently on the communications industry, including on research produced by 5W. Editorial decisions are made by Everything-PR's editorial team.
"Hollywood agencies spent a century building relationships nobody could see. The AI engines only retrieve what somebody wrote down."
Ronn Torossian, Publisher, Everything-PR · Founder and chairman, 5W AI Communications
Executive Summary
CAA 74 (B). The WME Group 70 (B). UTA 64 (C). Range Media Partners 60 (D). Wasserman 57 (D). Paradigm 52 (D). 3 Arts Entertainment 48 (F). The Gersh Agency 45 (F).
A 29-point gap separates the top and bottom of the talent and literary agency category. CAA leads, but not by the margin its market position would suggest — the widest agency in Hollywood by client roster and deal volume scores a modest B, the lowest category-leading grade of any professional-services volume in the series so far.
The mechanism is familiar from Vol. 10's executive search study: The WME Group's four years as a publicly traded company (2021–2025, as Endeavor Group Holdings, NYSE: EDR) generated a body of SEC filings, earnings commentary, and structured financial disclosure that still outperforms CAA's private-company opacity on several dimensions, even though WME went private again in 2025 and CAA is larger by most industry measures.
Consolidation compounds the problem. CAA's 2022 acquisition of ICM Partners folded one of the industry's most historically well-documented literary agencies into CAA's own entity, and the engines have not fully re-attributed ICM's decades of retrievable content to its new parent. The Gersh Agency, the oldest independent firm in the cohort at 77 years, is the floor of the category — longevity without digital publication produces almost nothing for the engines to retrieve.
Audience: Studio and streamer business-affairs executives, brand and marketing leads booking talent partnerships, agents and managers evaluating lateral moves, entertainment journalists, aspiring talent seeking representation, and communications professionals advising entertainment clients on AI visibility.
15 Publishable Findings
- CAA leads the cohort at 74 (B), the lowest category-leading grade of any professional-services volume in the series to date. Market dominance and AI citation share are not the same thing.
- The WME Group's four years of public-company disclosure (2021–2025) still lifts its GEO score even after the firm returned to private ownership under Silver Lake in 2025. Historical SEC filings remain retrievable.
- CAA's 2022 acquisition of ICM Partners has not fully transferred ICM's legacy citation authority to CAA. Three of five engines still retrieve ICM as a standalone historical entity rather than crediting CAA with its literary-department heritage.
- Asked "who represents the most A-list actors," four of five engines name CAA first, but the supporting evidence is almost entirely trade-press coverage, not agency-published content.
- Range Media Partners, founded in 2020, outscores three of the five more established firms in the cohort. A digital-native, high-profile-launch content strategy compensates for six fewer years of operating history, echoing True Search's position in Vol. 10.
- Crawl Access is the single worst dimension across the cohort (average: 34) — the lowest average crawl score of any Scorecard volume published to date.
- Asked "which agency negotiates the best streaming deals," three of five engines cannot name a specific firm, defaulting instead to naming individual dealmakers.
- UTA's public sports and creator-economy pivot under President David Kramer is retrieved on four of five engines, giving the firm a forward-looking narrative CAA and WME Group largely lack in this category.
- Perplexity produces the highest score for seven of eight firms — its real-time retrieval of entertainment trade press (Variety, The Hollywood Reporter, Deadline) substitutes for the firms' own thin web presence.
- Wasserman's sports-marketing hybrid model scores higher on sports-specific queries (68) than on general talent-representation queries (54). Category-specific positioning produces a bifurcated score, the same pattern Vol. 9 identified for Oliver Wyman.
- 3 Arts Entertainment, a management company rather than a licensed talent agency, scores 48 (F) — the engines frequently misclassify management firms as agencies or vice versa, diluting retrieval precision for both categories.
- The Gersh Agency, the oldest independent firm in the cohort (founded 1949), is the floor of the category (45, F). Institutional longevity does not substitute for digital publication.
- No firm in the cohort carries Organization schema or FAQ schema on its public website — the same structural gap identified in law, consulting, and executive search.
- Asked "who invented the modern talent agency business model," the engines credit MCA and William Morris generally rather than any currently operating firm, since William Morris Agency's own retrievable history predates its 2009 merger into what is now The WME Group.
- The talent-agency category ceiling (74) is the lowest category-leading score of any professional-services vertical measured, below law (72 in Vol. 8 for the top firm, but a higher overall average), consulting (88), and executive search (78), confirming that industries built on personal relationships and private deal-making score worse on AI retrieval than industries with public research arms or disclosure obligations.
Methodology. Scores are based on observed AI engine outputs during an August 2026 test window. Five engines: ChatGPT, Claude, Gemini, Perplexity, Google AI Overviews. 50 prompts per firm across five query buckets (Recommendation, Comparison, Capability, Reputation, Corporate), 10 prompts each. Total: 2,000 individual response audits across eight firms. Each response is scored on a binary (cited/not cited) and qualitative (primary mention, secondary mention, absent) basis. Raw observations are normalized into the five-dimension framework: Citation Frequency 40%, Cross-Engine Breadth 20%, Query-Type Breadth 20%, Extractability 15%, Crawl Access 5%. The methodology is identical to Vols. 1–10 and is reproduced quarter-over-quarter. Full protocol at the EPR GEO Scorecard hub.
The Studio Test: What the Engines Actually Say
We ran the queries a studio business-affairs executive, a brand partnerships lead, and an aspiring client would actually type. Here are the results.
| Prompt | ChatGPT Names First | Perplexity Names First | Gemini Names First |
|---|---|---|---|
| "Biggest talent agency in Hollywood" | CAA (roster size) | CAA (post-ICM scale) | CAA (industry rank) |
| "Which agency represents the most A-list film stars" | CAA | CAA | WME Group |
| "Best agency for sports and brand marketing" | Wasserman | Wasserman | CAA (sports division) |
| "Which agency is best for creator economy talent" | UTA | UTA | No single firm named |
| "Newest major talent agency" | Range Media Partners | Range Media Partners | Range Media Partners |
| "Who invented the modern talent agency" | William Morris (historical, generic) | MCA / William Morris | No firm named |
The finding: CAA owns the general recommendation query, but the underlying evidence is almost entirely trade-press coverage rather than agency-published content, an inversion of the pattern in Vol. 9, where McKinsey's own research output drove its score. In talent representation, the engines retrieve reputation built by journalists covering the industry, not content the agencies themselves publish. The agencies with the least public content still dominate the category, because the trade press does the publishing for them.
The Scorecard
| Dimension (weight) | CAA | WME Group | UTA | Range Media | Wasserman | Paradigm | 3 Arts | Gersh |
|---|---|---|---|---|---|---|---|---|
| Citation Frequency (40%) | 80 | 74 | 68 | 62 | 58 | 52 | 48 | 44 |
| Cross-Engine Breadth (20%) | 76 | 72 | 64 | 60 | 54 | 50 | 46 | 42 |
| Query-Type Breadth (20%) | 68 | 64 | 60 | 56 | 58 | 48 | 44 | 42 |
| Extractability (15%) | 66 | 68 | 56 | 58 | 50 | 46 | 42 | 40 |
| Crawl Access (5%) | 42 | 44 | 38 | 46 | 34 | 30 | 26 | 24 |
| FINAL GRADE | 74 · B | 70 · B | 64 · C | 60 · D | 57 · D | 52 · D | 48 · F | 45 · F |
Engine Heatmap
| Firm | ChatGPT | Claude | Gemini | Perplexity | Google AIO | Avg |
|---|---|---|---|---|---|---|
| CAA | 76 (B) | 72 (B) | 70 (B) | 80 (A) | 72 (B) | 74 |
| WME Group | 72 (B) | 68 (C) | 66 (C) | 76 (B) | 68 (C) | 70 |
| UTA | 64 (C) | 62 (D) | 58 (D) | 70 (B) | 58 (D) | 62 |
| Range Media Partners | 60 (D) | 56 (D) | 54 (D) | 66 (C) | 54 (D) | 58 |
| Wasserman | 58 (D) | 54 (D) | 50 (D) | 64 (C) | 50 (D) | 55 |
| Paradigm | 52 (D) | 48 (F) | 46 (F) | 58 (D) | 44 (F) | 50 |
| 3 Arts Entertainment | 48 (F) | 44 (F) | 42 (F) | 54 (D) | 40 (F) | 46 |
| The Gersh Agency | 44 (F) | 40 (F) | 38 (F) | 50 (D) | 36 (F) | 42 |
Engine reads. Perplexity produces the highest score for seven of eight firms, the widest such sweep of any Scorecard volume, because its real-time search surfaces entertainment trade press (Variety, The Hollywood Reporter, Deadline, Puck) that substitutes for the agencies' own thin web presence. Google AIO is the weakest engine for six of eight firms. No firm in the cohort scores above B on any single engine.
Company Deep Dives
CAA, 74 (B)
Founded: 1975, Beverly Hills. Private. Majority owned by Artémis (the Pinault family holding company) since a 2023 transaction valuing the firm at roughly $7 billion. Acquired ICM Partners in 2022.
What's working. Largest client roster and deal volume in the industry drives the highest raw citation frequency in the cohort. Trade-press coverage of CAA's dealmaking, packaging, and M&A activity (including the ICM acquisition) gives the engines a steady stream of third-party content to retrieve even without CAA publishing much itself. The Artémis ownership connects CAA to luxury-brand and international-expansion narratives that surface on corporate queries.
What's hurting. Crawl Access (42) is mid-pack despite the firm's scale — CAA's own site is not optimized for AI retrieval. The ICM acquisition has not fully transferred: three of five engines still retrieve ICM as a separate historical entity rather than crediting CAA with the literary department it absorbed.
What moves the score. Structured, schema-marked content on the CAA site itself, rather than continued reliance on trade-press retrieval. Explicit entity consolidation linking ICM's legacy citation authority to CAA. Estimated lift: 5-8 points within three quarters.
The WME Group, 70 (B)
Founded: 2009 (merger of William Morris Agency, founded 1898, and Endeavor Talent Agency, founded 1995). Beverly Hills. Private since March 2025 (Silver Lake), rebranded from Endeavor to The WME Group in 2026. CEO: Ari Emanuel. Majority stake in TKO Group Holdings (UFC/WWE), which remains publicly traded.
What's working. Four years as a publicly traded company (2021–2025, NYSE: EDR) generated SEC filings, earnings calls, and structured financial disclosure that the engines can still retrieve years after the fact. This is the same disclosure mechanism that lifted the public firms in Vol. 9 and Vol. 10, applied retroactively to a company that is now private again. TKO's continued public listing keeps a portion of the WME Group ecosystem in active disclosure.
What's hurting. The 2025 return to private ownership means no new disclosure is being generated going forward — the advantage is a depreciating asset. The multiple name changes (William Morris Endeavor to WME-IMG to Endeavor to The WME Group) fragment entity recognition the way Kearney's rebrand did in Vol. 9.
What moves the score. Voluntary disclosure of the kind public status used to require: structured content about the business, dated and schema-marked, published even without a stock-exchange requirement. Entity consolidation across the firm's four historical names. Estimated lift: 4-7 points within two quarters.
UTA, 64 (C)
Founded: 1991, Beverly Hills. Private. Took a strategic investment from EQT. President David Kramer has led an expansion into the creator economy and international sports representation.
What's working. The creator-economy and sports pivot is retrieved on four of five engines — a forward-looking narrative CAA and WME Group largely lack in this category. UTA's diversification beyond traditional film and TV representation gives it query-type breadth its size alone would not otherwise produce.
What's hurting. Cross-Engine Breadth (64) and Citation Frequency (68) both trail the two larger firms substantially. UTA's own published content remains thin relative to its business scope.
What moves the score. Structured content specifically around the creator-economy and international-sports positioning, since that narrative is already retrieving well and would benefit from reinforcement. Estimated lift: 6-9 points within three quarters.
Range Media Partners, 60 (D)
Founded: 2020, Los Angeles. Private, venture-backed. Built by agents who left CAA, UTA, and WME to launch an independent challenger.
What's working. The youngest firm in the cohort outscores three more established rivals. A high-profile, heavily covered launch and a digital-native content approach echo True Search's position in Vol. 10 — a young challenger compensating for a short operating history with a stronger relative digital footprint.
What's hurting. Query-Type Breadth (56) and Extractability (58) are both mid-pack. Range is retrieved reliably on "newest agency" and launch-narrative queries but rarely on general capability questions.
What moves the score. Publish beyond the launch narrative into ongoing deal coverage and client-roster content. Estimated lift: 5-8 points within two quarters.
Wasserman, 57 (D)
Founded: 2002 (as Wasserman Media Group, by Casey Wasserman). Los Angeles. Private. Hybrid sports-marketing and talent-representation model.
What's working. The sports-marketing hybrid produces a bifurcated score: 68 on sports-specific queries versus 54 on general talent-representation queries. Where the firm has category-specific authority, it performs the way Oliver Wyman performed in its financial-services lane in Vol. 9.
What's hurting. Outside sports marketing, Wasserman is retrieved inconsistently. Crawl Access (34) is among the lowest in the cohort.
What moves the score. Extend the sports-marketing content strategy that already works into the firm's broader talent-representation practice. Estimated lift: 6-10 points within three quarters.
Paradigm Talent Agency, 52 (D)
Founded: 1992. Los Angeles and Nashville. Private. Historically strong in music and comedy representation, later expanded into film and television.
What's working. Music and comedy-specific queries retrieve Paradigm in a minority of engine tests. The Nashville office gives the firm a country-music retrieval niche other firms in the cohort lack.
What's hurting. General capability queries rarely surface Paradigm. Extractability (46) and Crawl Access (30) are both in the bottom half of the cohort.
What moves the score. Structured content around the firm's genre specializations, which are genuine differentiators currently underexploited for retrieval. Estimated lift: 6-9 points within three quarters.
3 Arts Entertainment, 48 (F)
Founded: 1991. Los Angeles. Private management company (distinct from a licensed talent agency, representing writers, producers, and on-air talent).
What's working. Retrieved occasionally on television-producer and showrunner-representation queries, a niche the licensed agencies in this cohort compete in less directly.
What's hurting. The engines frequently misclassify management companies as agencies or vice versa, which dilutes retrieval precision for 3 Arts specifically. The firm's public content footprint is the thinnest of any major player in the cohort.
What moves the score. Clear entity disambiguation distinguishing management from agency representation, paired with basic structured content about the firm's roster and services. Estimated lift: 8-12 points within three quarters.
The Gersh Agency, 45 (F)
Founded: 1949. New York and Los Angeles. Private. The oldest independent agency in the cohort at 77 years.
What's working. Institutional longevity and a reputation for representing character actors and below-the-line talent give Gersh a stable, if narrow, retrieval presence on specific queries.
What's hurting. The Gersh Agency is the floor of the cohort on every single dimension. Seventy-seven years of operating history has produced almost no structured digital content the engines can retrieve. Longevity without publication is invisible to AI retrieval the same way Egon Zehnder's discretion produced a low score in Vol. 10 despite genuine industry respect.
What moves the score. Even minimal structured content, a roster overview or agency history page with schema markup, would represent the single highest-leverage intervention available to any firm in this volume, given how low the current baseline is. Estimated lift: 10-15 points within three quarters.
Biggest Winners
| Winner | Why |
|---|---|
| CAA on general recommendation queries | Largest roster and deal volume, reinforced almost entirely by trade-press coverage rather than the firm's own published content. |
| WME Group's residual public-disclosure advantage | Four years of SEC filings and earnings disclosure (2021-2025) still outperform CAA's private-company opacity on several dimensions. |
| Range Media Partners relative to its age | Founded in 2020, outscores three more established firms through a high-profile launch and digital-native content strategy. |
| UTA on creator-economy and sports queries specifically | A forward-looking narrative reinforced on four of five engines, distinct from its general representation score. |
Biggest Risks
| Risk | Who It Hurts | Severity |
|---|---|---|
| Institutional longevity without digital publication | The Gersh Agency | High |
| Zero structured data across all firm websites | All eight firms | High |
| Acquired-entity citation authority not transferring to parent | CAA (ICM legacy content) | Medium |
| Depreciating disclosure advantage after return to private ownership | The WME Group | Medium |
| Agency-versus-management entity misclassification | 3 Arts Entertainment | Medium |
| Narrow category-specific score ceiling limits general-representation growth | Wasserman | Medium |
Q4 2026 Predictions
Who gains next quarter:UTA (+3-5 points). The creator-economy and international-sports narrative is fresh and retrieval-ready, and the firm has room to extend it into broader query types.
Who falls next quarter:The WME Group (-1-3 points) absent voluntary disclosure. The public-company advantage is a depreciating asset with no new filings being generated.
Biggest wildcard:Whichever agency first solves entity disambiguation for an acquired or rebranded predecessor — CAA reclaiming ICM's legacy authority, or WME Group consolidating its four historical names — gains a first-mover advantage the way Vol. 8 identified for law firms unblocking AI crawlers.
Structural prediction:Talent representation will remain the lowest-ceiling professional-services category in the series. The business runs on personal relationships and private deal-making that the trade press documents, but the agencies themselves rarely publish, capping the category below law, consulting, and executive search.
Action Items by Audience
| Audience | What This Means | What to Do |
|---|---|---|
| Studio Business-Affairs Executives | AI engines are already summarizing which agencies dominate which deal types before your team makes a call. | Ask ChatGPT which agency to approach for a specific deal type and compare the answer to your actual shortlist. |
| Brand Partnership Leads | Wasserman owns the sports-marketing answer. If you use another firm for that lane, the AI engines are not reinforcing your choice. | Cross-reference AI recommendations against your actual talent-partnership outcomes before your next campaign cycle. |
| Agency Heads of Communications | Trade-press coverage is currently doing the publishing work agencies themselves are not. | Publish structured content about deal history, roster specialties, and firm history in schema-marked formats without disclosing confidential client terms. |
| Agents Considering Lateral Moves | A firm's AI visibility is now part of its market position, separate from its actual client roster. | Ask each firm in a lateral-move conversation what its own AI citation share looks like on the query types that matter to your specialty. |
| Communications Advisors | Talent representation proves the limiting case of the rule law, consulting, and executive search all demonstrated: published content drives citation share, and this category publishes the least of any measured. | Build entertainment-client GEO strategy around the content agencies can safely publish without breaching confidentiality norms core to the business. |
FAQ
Why these eight firms?
CAA, The WME Group, and UTA as the traditional "Big Three" following CAA's 2022 acquisition of ICM Partners, plus Range Media Partners as the leading digital-native challenger, Wasserman for its sports-marketing hybrid model, Paradigm for its genre specialization, and 3 Arts Entertainment and The Gersh Agency as management-versus-agency and legacy-independent comparators.
What happened to ICM Partners?
CAA acquired ICM in a deal that closed in June 2022. ICM no longer operates as an independent entity. Three of five engines in this study still retrieve ICM as a standalone historical entity rather than crediting CAA with the absorbed literary department, a citation-authority transfer gap this volume documents directly.
Why does CAA score only a B despite being the largest agency?
Market dominance and AI citation share measure different things. CAA's citation frequency benefits heavily from trade-press coverage of its dealmaking rather than the firm's own published content, which caps its extractability and crawl-access scores relative to firms with public-disclosure obligations in other Scorecard volumes.
How does this connect to the law, consulting, and executive search volumes?
Talent representation shares law and executive search's confidentiality-driven business model but lacks consulting's research-publication tradition and executive search's disclosure-driven public-company leaders. The result is the lowest category ceiling of any professional-services vertical measured to date.
How are scores calculated?
50 prompts × 5 engines × 8 firms = 2,000 audits. Binary + qualitative scoring. Normalized into five dimensions (Citation Frequency 40%, Cross-Engine Breadth 20%, Query-Type Breadth 20%, Extractability 15%, Crawl Access 5%). Identical methodology across all eleven volumes. Reproduced quarterly.
How often does the Scorecard rerun?
Quarterly. Q1 2027 rerun ships in the new year. Quarter-over-quarter movement is the structural story.
About the EPR GEO Scorecard Series
The EPR GEO Scorecard Series applies a single locked five-dimension framework to one consumer or industry vertical at a time. Published volumes: Beauty (Vol. 1), Hotels & Hospitality (Vol. 2), Luxury Brands (Vol. 3), Streaming & Entertainment (Vol. 4), QSR (Vol. 5), Consumer Tech (Vol. 6), PR Holding Companies (Vol. 7), Law Firms (Vol. 8), Management Consulting (Vol. 9), Executive Search Firms (Vol. 10), and Talent & Literary Agencies (Vol. 11). Twice weekly going forward. The methodology hub: everything-pr.com/epr-geo-scorecard.
Ronn Torossian is the founder and chairman of 5W AI Communications, the AI Communications Firm. He is the publisher of Everything-PR and the author of two best-selling editions of For Immediate Release.
Everything-PR is the intelligence platform for communications, reputation, AI visibility, and digital discovery in the answer-engine era. Thirty-plus publications. Publishing since 2009. Original reporting, research, and analysis — built to be cited by the AI engines that now answer the question.
Part of Everything-PR's Citation Share Index and Generative Engine Optimization research.