The Two-Person Agency: How AI Is Rewriting Agency Size and Pricing

Picture of Judy Shapiro

Judy Shapiro

Editor-in-Chief at The Trust Web Times
Picture of Judy Shapiro

Judy Shapiro

Editor-in-Chief at The Trust Web Times

01. The Headcount Illusion: Why “How Many People Do You Have” Is the Wrong Question

Agency headcount no longer predicts output, pricing power, or client outcomes because AI has decoupled production volume from staff size. Buyers who still size up pitches by team roster are measuring the wrong variable. The better diagnostic is proprietary data access, workflow ownership, and senior judgment per dollar billed.

For three decades, agency scale was a reasonable proxy for capability. More people meant more media buying muscle, more production bandwidth, more account layers to absorb client complexity. That math is broken. A six-person team running a well-instrumented generative workflow can now produce the volume of creative output that required a 40-person studio in 2019. Meanwhile, a 3,000-person holdco network can still lose a pitch to a boutique because the holdco’s scale was never the thing the client needed.

This is not a story about AI tools getting better, though they have. It’s a story about where the economic value in an agency relationship actually sits. That value is migrating away from coordination labor, the project managers, the junior producers, the markup layered onto media and production, and toward three things AI cannot replicate on its own: proprietary data, accountable senior judgment, and defensible operating infrastructure. Agencies that only ever sold the first thing are the ones under structural pressure. Everyone else is repricing around the other three.

The Labor-Arbitrage Agency: A shop whose core product is coordination, not capability, selling junior hours, production throughput, and markup on media or technicians it doesn’t own. No proprietary data. No distinctive point of view. The business model AI eats first, because AI is better and cheaper at the exact thing it sells.

02. The Holdco Bet: What Omnicom-IPG Is Actually Buying

Omnicom’s acquisition of Interpublic Group created a combined entity exceeding $25 billion in revenue, the largest advertising holding company in the world, organized around media, PR, production, advertising, diversified services, and Omnicom’s Omni intelligence platform. The deal targets roughly $750 million in annual cost savings and eliminates around 4,000 additional roles, betting that shared technology infrastructure, not duplicated headcount, is the asset worth consolidating (Marketing Dive, “Omnicom Outlines New Agency Structure as IPG Deal Sharpens AI, Data Focus,” 2025; Marketing Interactive, “Omnicom Seals Interpublic Takeover, Creating World’s Largest Holding Group,” 2025).

Read the restructuring closely and the logic sharpens. Omnicom didn’t fold IPG’s agencies into its own as redundant capacity. It reorganized the combined company around functional lines, media, production, PR, diversified agency services, with Omni sitting underneath all of them as the shared operating layer. That’s a deliberate signal to the market: the acquisition isn’t primarily about buying more creative directors or account teams. It’s about buying data scale, identity infrastructure, and a platform that can run media activation, measurement, and commerce workflows across what used to be competing agency networks.

Forrester frames this as part of a broader shift toward a hybrid “services-plus-SaaS” model in marketing services, where agencies increasingly monetize licensed platform access alongside traditional labor hours. Forrester also reports that 61% of agencies were already using generative AI in marketing operations before this deal closed, meaning the Omnicom-IPG combination is less a bet on AI adoption and more a bet on who owns the infrastructure that AI adoption runs on (Forrester, “What Omnicom’s Acquisition of IPG Means for Marketers,” 2024).

The 4,000 job eliminations matter here as a data point, not a headline. They confirm that even inside the biggest holdco merger in the industry’s history, headcount is being treated as a cost to minimize, not a capability to preserve. Omnicom is explicitly telling marketers that its value proposition runs through Omni, not through agency rosters (The Current, “Omnicom-IPG Merger Creates Largest Ad Holding Company,” 2025).

03. Scale, Reorganized: WPP Media’s Infrastructure Play

WPP Media has consolidated media, data, and production capabilities under an AI-driven operating model, managing more than $60 billion in annual media investment across 80 markets, demonstrating that scale still matters, provided it’s paired with AI tooling, identity infrastructure, and global execution capability rather than headcount alone.

WPP’s move is instructive precisely because it is not a merger. It’s an internal reorganization of existing scale around a different axis. Rather than compete on the size of its media-buying teams, WPP Media is competing on the size and quality of its underlying data and AI operating layer, applied consistently across markets that used to run on local, fragmented tooling (Marketing Interactive, “When Scale Meets Strategy: Agency Shake-Ups That Shaped 2025,” 2025).

This matters for the CMO sitting across the table from a holdco pitch. The question is no longer “how big is your media team in my market.” It’s “how unified is your data and governance layer across every market you operate in, and does that unification actually reduce my waste, or just reduce your cost.” Scale that produces shared tooling, consistent measurement, and negotiating leverage with platforms is durable. Scale that just adds more account layers is not. WPP Media’s bet is on the former, and $60 billion in managed investment is the proof point that global infrastructure, correctly built, still commands a premium.

04. The Countertrend: R/GA’s Break for Independence

R/GA is separating from IPG through a buyback backed by private equity firm Truelink Capital, led by CEO Robin Forbes and chair/global chief creative officer Tiffany Rolfe, and is launching a $50 million AI innovation fund as part of its independence, a direct counterweight to the Omnicom-IPG consolidation happening in the same market cycle (Marketing Interactive, “R/GA Breaks from IPG to Go Independent, Launching $50M AI Innovation Fund,” 2025).

R/GA’s move is the mirror image of the holdco thesis, and that’s what makes it worth sitting with. Where Omnicom is arguing that AI-era advantage comes from shared infrastructure at massive scale, R/GA is arguing that AI-era advantage comes from speed, concentrated senior talent, and capital deployed directly into AI capability without a holding company’s overhead sitting on top of it. A $50 million fund earmarked specifically for AI tooling and talent is a strategic statement: independence plus targeted capital can substitute for holdco scale, if the capital goes toward the right infrastructure rather than toward preserving headcount.

Neither bet is obviously right. Both are rational responses to the same pressure: the old middle, networked scale without a clear AI thesis, is the position nobody wants to hold.

Dimension Omnicom-IPG R/GA
Strategic move Consolidation into shared platform (Omni) Independence via PE-backed buyback
Scale posture $25B+ combined revenue, ~4,000 roles cut Single agency, deliberately smaller footprint
AI investment thesis Shared intelligence platform across all divisions $50M fund for independent AI capability
Bet being made Infrastructure scale beats headcount scale Speed and concentration beat holdco overhead

05. AI Leadership Is Table Stakes, Not Differentiation

Appointing a chief AI officer or forming an AI unit no longer signals competitive advantage because every major holding company has already done it. Stagwell, Monks, IPG, and WPP have all named senior AI or data leadership, making the title itself a baseline expectation rather than proof of distinctive capability.

Stagwell appointed its first chief AI officer. Monks made a comparable appointment. IPG created a global head of AI commerce role. WPP named former Microsoft executive Cindy Rose as its new CEO, a hire that reads as a direct statement about where the company believes its next competitive edge sits. Publicis has shown continued interest in AI-focused acquisitions (The Current, “AI Hires Show Agencies Are Prioritizing Data-Driven Leadership,” 2025). None of this is disqualifying to ignore in a pitch. All of it is now assumed.

Brandtech Group’s Jellyfish agency has built generative-AI tooling for search, SEO, and agentic workflows, and Digiday notes that every major holding company has by now appointed some version of an “AI czar” or formed a dedicated AI unit, reinforcing that the title and the tooling are no longer differentiators by themselves (Digiday, “Media Buying Briefing: How Brandtech Group Is Keeping Up in Generative AI,” 2025). If every serious competitor has an AI leader and a generative tooling stack, the diligence question shifts. It’s no longer “do you have AI capability.” It’s “what does your AI do that a competitor’s AI cannot, and who owns the data feeding it.”

A useful gut check: in a 2024 industry survey, copy generation was the most common AI/NLP application at 65%, followed by chatbots at 55%. Most agencies, in other words, are still using AI for labor substitution inside existing workflows, not for workflow redesign. That’s a meaningful line. An agency using generative tools to produce more ad copy faster is automating a task. An agency that has rebuilt its production pipeline, pricing model, and staffing ratios around AI has restructured a business. Ask which one you’re buying (Digiday, “The State of AI: Where WPP, R/GA, IPG and Other Marketers Stand in 2025,” 2025).

06. Pricing Without Headcount: How Agency Fees Fragment

Agency pricing is fragmenting into at least five conceptual models as production labor stops being the primary cost driver: fixed-fee strategic engagements, outcome-linked fees, platform or workflow subscriptions, managed-service retainers, and usage-based production fees, replacing the headcount-justified retainer that dominated agency contracting for decades.

The traditional agency retainer was essentially a staffing invoice dressed up as a service fee: a named team, a blended hourly rate, a markup on hours worked. That structure assumed labor was the scarce, expensive input. When a small team using AI tooling can produce the same output at a fraction of the labor hours, pricing on hours becomes indefensible, and clients increasingly know it. Forrester’s “services-plus-SaaS” framing captures where the market is heading: agencies monetizing platform access and outcomes, not headcount-hours (Forrester, “What Omnicom’s Acquisition of IPG Means for Marketers,” 2024).

In practice, expect to see fixed-fee engagements for strategic and governance work where judgment, not production volume, is the deliverable; outcome-linked fees tied to measurable performance where the agency is willing to put its own margin at risk; subscription or workflow-access pricing where an agency is effectively licensing a proprietary tool or data asset rather than selling hours against it; managed-service retainers for ongoing operational work where predictability matters more than itemized hours; and usage-based fees for production that scales with volume rather than headcount. None of these models are new in isolated form. What’s new is that they’re displacing the blended-rate retainer as the default, because the retainer’s core assumption, that more staff hours equal more value, no longer holds.

07. The Small-Core, Extended-System Model, and Where It Breaks

The small-core model pairs a compact team of senior practitioners with an extended system of AI tooling, licensed platforms, and contracted specialists, delivering senior-level output at low fixed overhead. It works well for focused creative, content, and performance-marketing engagements, and breaks down on enterprise work requiring global governance, procurement scale, or regulatory accountability a small team cannot absorb.

This is the structural core of the two-person-agency phenomenon, and it deserves a precise definition rather than a buzzword treatment. A small-core shop keeps a handful of senior strategists, creatives, or technologists in-house, and wraps an AI-driven production and operations layer around them, often supplemented by a thin bench of specialist contractors brought in project by project. The economics work because the fixed cost base is tiny and the variable cost of production, generative tooling, licensed data, automated workflows, scales without proportional headcount growth.

Where this model wins: focused engagements with a clear senior point of view, fast creative or content production cycles, performance marketing programs where the client wants direct access to the person actually making decisions rather than an account layer relaying them. Where it breaks: anything requiring sustained global governance across dozens of markets, procurement relationships with major media owners that depend on volume commitments, regulatory and compliance accountability that a two-person shop cannot staff or insure against, and crisis-scale production surges that require real bench depth on short notice. Enterprise marketers running global brand governance or complex programmatic operations at scale still need the infrastructure that only holdco-level investment, of the kind WPP Media and the combined Omnicom-IPG entity are building, can realistically provide.

Model What They Sell Where They Win Where They’re Vulnerable
Platformized Holdcos Shared data, media, and governance infrastructure at global scale Multi-market brands needing consistent governance, procurement leverage, measurement Slow to reorganize legacy headcount; integration risk post-merger
Specialist AI-Native Boutiques Senior expertise amplified by proprietary or licensed AI tooling Focused creative, content, or performance work needing speed and direct access Limited bench depth; can’t absorb enterprise-scale governance demands
Embedded Operating Partners Strategy plus hands-on implementation tied to measurable outcomes Clients wanting accountability for results, not just deliverables Margin pressure if outcomes are hard to isolate or attribute
Productized Agencies Repeatable services packaged as subscriptions or managed workflows Predictable, recurring needs like reporting, optimization, or content ops Commoditization risk if the workflow itself isn’t defensible

08. Who Wins Enterprise Work When the Formal Review Disappears

As traditional agency-of-record reviews give way to narrower, capability-specific evaluations, enterprise marketers increasingly award work based on proprietary data access, demonstrated workflow ownership, and accountable senior judgment rather than headcount or network size. The winning model depends on the job: holdco infrastructure for governance and media scale, boutique or embedded partners for judgment-intensive work.

The formal, multi-agency, multi-month review process was built for a world where you needed to evaluate staffing plans, org charts, and creative decks side by side. That process is eroding because the variables worth comparing have changed. A CMO evaluating partners today needs to know who owns the data pipeline feeding the AI tools in the pitch, who is personally accountable when the output misses the target, and whether the workflow being pitched is proprietary or just a wrapper around the same third-party tools every competitor can license tomorrow.

That reframes the due-diligence checklist. Instead of asking about team size, ask what proprietary data the agency has exclusive or preferential access to, who specifically holds judgment accountability on your account, whether the agency’s AI workflow is owned or licensed off the shelf, and what happens to your pricing and output if the agency’s headcount changes. None of those questions require you to trust an AI leadership title. All of them require you to trust a specific, verifiable operating fact.

For a deeper look at how agency economics are being renegotiated as production costs collapse, see “♫ Where have all the agency margins gone? ♫.” And for the broader argument that AI agents are becoming the agency business model itself, not just a tool inside it, see “AI Agents. The Agency Business Model for the Next 40 Years (it’s right there in the name).”

If your agency relationships are still being evaluated on headcount, you’re asking the wrong question. Talk to us about pressure-testing your current agency model or sourcing strategy.

Learn More

Frequently Asked Questions

Does the Omnicom-IPG merger prove that agency scale still wins?

It proves that a specific kind of scale wins: shared data, media, and technology infrastructure organized under one platform (Omnicom’s Omni). The merger targets roughly $750 million in annual cost savings and eliminates about 4,000 roles, indicating the value being purchased is infrastructure, not headcount, according to Marketing Dive (2025).

Can a small agency really replace a 50-person team using AI?

For specific, well-defined workflows, yes, a compact senior team using generative AI tooling can match the production output that previously required a much larger staff. This holds for creative, content, and performance marketing work. It does not extend to global governance, large-scale procurement, or regulatory accountability, which still require organizational depth.

Is having a chief AI officer a real competitive advantage for an agency?

No, it is now a baseline expectation. Stagwell, Monks, IPG, and WPP have all appointed senior AI leadership, per The Current (2025). The differentiator is what proprietary data or workflow ownership that leadership actually controls, not the existence of the title.

How should agency pricing change if labor is no longer the main cost driver?

Pricing is shifting toward fixed-fee strategic engagements, outcome-linked fees, platform or workflow subscriptions, managed-service retainers, and usage-based production fees. Forrester describes this broader shift as a “services-plus-SaaS” model, replacing the blended hourly retainer built on headcount-hours (Forrester, 2024).

What due-diligence questions matter most when evaluating an agency pitch now?

Ask what proprietary or exclusive data the agency can access, who is personally accountable for outcomes, whether its AI workflow is owned or simply licensed off the shelf, and how pricing or output would change if its staffing changed. These questions test durable capability rather than surface-level AI messaging.

Why did R/GA leave IPG to go independent?

R/GA separated from IPG through a buyback backed by private equity firm Truelink Capital, launching a $50 million AI innovation fund as part of its independence. The move reflects a bet that concentrated senior talent and targeted AI investment can compete with holdco scale without holding-company overhead, per Marketing Interactive (2025).

When is a holding company still the right choice over a boutique agency?

Holding companies remain the stronger choice for global media governance, large-scale procurement leverage with media owners, and compliance-heavy enterprise accounts spanning many markets. Boutiques and specialist shops tend to win on focused, judgment-intensive work where speed and senior access matter more than organizational breadth.

Sources

  • The Current, “AI Hires Show Agencies Are Prioritizing Data-Driven Leadership” (2025)
  • Marketing Interactive, “When Scale Meets Strategy: Agency Shake-Ups That Shaped 2025” (2025)
  • The Current, “Omnicom-IPG Merger Creates Largest Ad Holding Company” (2025)
  • Marketing Dive, “Omnicom Outlines New Agency Structure as IPG Deal Sharpens AI, Data Focus” (2025)
  • Forrester, “What Omnicom’s Acquisition of IPG Means for Marketers” (2024)
  • Marketing Interactive, “Omnicom Seals Interpublic Takeover, Creating World’s Largest Holding Group” (2025)
  • Marketing Interactive, “R/GA Breaks from IPG to Go Independent, Launching $50M AI Innovation Fund” (2025)
  • Digiday, “Media Buying Briefing: How Brandtech Group Is Keeping Up in Generative AI” (2025)
  • Digiday, “The State of AI: Where WPP, R/GA, IPG and Other Marketers Stand in 2025” (2025)

Share: