Even as I write these words, I am twitching. I started my career at a large ad agency that serviced some of the largest advertisers, such as AT&T, P&G and GM. And while it is painful to say this out loud, I know many people are quietly wondering the same thing: are holding companies living on borrowed time? Perhaps an honest look can point the way forward.
Is it too late to prevent an extinction event?
Sixty-six million years ago, a rock roughly six miles wide slammed into the Yucatán Peninsula. The inevitable happened. Most of the dinosaurs that survived the blast did not survive what came after. They were too big. They needed enormous amounts of food, they reproduced slowly, and they were built for a world that no longer existed. The animals that made it through were generally small, warm-blooded and adaptable. They needed less, moved faster and could live on whatever was left.
The dinosaurs never saw it coming. Even if they had, there was nothing they could have done. Their fate was sealed the moment the rock entered the atmosphere.
AI is the meteor hurtling toward the mega-agency groups but the meteor is still in the sky. The scale of what’s coming is hard to grasp, almost as hard as explaining a meteor to a dinosaur. Yet unlike the dinosaurs, the largest agencies can still act. AI does not have to be their extinction event.
Too big to run or hide?
AI changes the underlying economics of every agency’s creative and media work: what it costs to make things, how long it takes, and how many people are needed to do it.
The giant holding companies, built over four decades of acquisition and consolidation, are the most exposed species in the ecosystem because of their size, their structure and their client base.
Scale used to be the holding companies’ great advantage. It bought media at volume, absorbed talent from every acquired shop, and let a single parent offer a Fortune 100 client every discipline under one contract. In an AI-shaped market, that same scale has become a drag (literally and figuratively).
Consider what a holding company actually is: dozens of agency brands, each with its own leadership, P&L, culture, tech stack and client commitments, stitched together by finance and shared services. Every one of those units has to be retrained and restructured before the group can change how it works. Every layer of management that sits on top was justified by the headcount underneath it.
The instinctive response has been to consolidate further: merge networks, fold legacy brands into a few super-agencies, build a central AI platform and push it down to every office.
Even now, as we watch two large organizations merge into one even larger one, we know years of integration lie ahead, at exactly the moment the market rewards speed.
Large agencies may see getting bigger as a rational way to insulate it from threats, but this strategy does nothing to forestall the biggest threat to its survival – the AI extinction event.
Aside from AI dismantling agencies’ core business model, there is also a capital-markets problem. Public holding companies answer to shareholders who expect steady margins and predictable revenue every quarter. Genuine reinvention means accepting a period where revenue falls faster than costs can be cut. Few public boards will sign up for that voluntarily, so the big groups tend to reinvent at the edges while protecting the core that is being disrupted.
The model will be unrecognizable in five years
By 2031, the agency org chart will look less like a pyramid and more like a diamond, or even an inverted triangle. The traditional pyramid depended on a wide base of junior staff doing production, versioning, research, reporting and first drafts. That base is precisely the work AI does fastest and cheapest.
This does not mean agencies need no people. It means they need different people, doing different things, in different proportions.
| Function | How it works today | How it works by 2031 |
| Production and versioning | Large teams resize, adapt and localize assets by hand | Automated pipelines generate thousands of variants; a small team curates and approves |
| Copy and first drafts | Junior writers produce drafts for senior review | AI drafts in seconds; senior writers direct, edit and protect the brand voice |
| Research and insight | Analysts spend weeks compiling reports | Agents synthesize data continuously; strategists spend their time on interpretation |
| Media planning and buying | Planners build plans and optimize campaigns manually | Algorithms optimize in real time; people set strategy, guardrails and budgets |
| Reporting | Account teams assemble decks every month | Live dashboards; account leads spend time on what the numbers mean for the client |
| Account management | Layers of coordinators route work and manage timelines | Fewer, more senior client partners with real business and commercial judgment |
| New roles | Rare or nonexistent | AI workflow designers, model and prompt specialists, brand-system builders, data and governance leads |
The valuable person in the 2031 agency is the one who combines creative instinct, strategic judgment, and client understanding with the ability to direct machines. One strong creative director with a well-built AI workflow can do what used to take a department.
The question for every agency is how to insulate itself from the meteor strike while allowing it to thrive in a post-impact world of AI-dominated marketing.
The question for the mega agencies, watching the meteor streak across the sky, is sharper still. The world after impact will look nothing like the one the large holding companies were born into. What does that future look like, right after the strike and for the 10 years after that?
A structure built to bill hours, in a world where AI collapses hours
The core contradiction is simple. Large agencies make money by selling people’s time. AI makes time dramatically less necessary. The more efficiently a big agency uses AI, the faster it erodes its own revenue base.
Most large-agency relationships are priced on some form of labor: retainers built from staffing plans, full-time-equivalent rate cards, or hourly fees marked up to cover overhead and margin. Clients and procurement departments audit those plans line by line. When a task that took 40 hours now takes four, the client does not expect to keep paying for 40. They expect the savings.
That leaves the holding company in a bind. Its cost base is enormous and slow to shrink: real estate, legacy technology, multiple layers of management, and a corporate center whose cost is spread across every operating company. Fees fall quickly when AI enters the workflow, but those fixed costs fall slowly. Margins get squeezed from both directions.
The obvious answer is to move away from selling hours and toward selling outcomes, intellectual property or productized services. That is genuinely the right direction. But it is very hard to make that switch across thousands of client contracts, a network of agency brands and a finance function designed around utilization rates. The billing model is not just a pricing choice. It is wired into how the whole organization is measured, staffed and rewarded.
The insourcing war: why big agencies fight hardest and lose most
AI does more than squeeze agencies’ hourly billing business model; it also reopens the decades-old question of insourcing. Brands have been building in-house agencies, programmatic desks and content studios for a very long time. What AI changes is the efficiency equation. Work that once required an agency’s scale, specialist talent and expensive tools can now be done by a lean internal team with the right software.
The holding companies feel this more than anyone because of who their clients are. They serve the world’s largest advertisers: global brands with big marketing departments, deep pockets, their own data and the ability to hire senior AI and technology talent directly. These are exactly the clients most able to bring work inside. When they do, the first things to go are the high-volume, repeatable services that fund the big agencies’ margins, such as production, adaptation, social content and performance media.
So, the large groups fight insourcing aggressively. They offer hybrid models, embed their own teams inside client buildings, license proprietary platforms, and argue that only they can provide global scale and governance. Some of these arguments are sound. But the strategy is fundamentally defensive. It spends energy protecting a shrinking pool of work, and it often sets up the agency to train the very client team that will eventually replace it.
The mammals: why agencies under 200 people are primed for reinvention
By contrast, we see that smaller agencies are thriving in the age of AI, (our firm stands as one example). They are not as vulnerable to insourcing as large agencies because their clients are typically mid-sized companies and challenger brands that lack the budget or appetite to build in-house AI capability.
For those clients, a nimble partner that has already done the hard work of rebuilding its own workflows is more valuable, not less.
Smaller shops have significant advantages compared to their big-agency counterparts:
- Decision speed. There is no group CTO, regional committee or parent-company approval cycle. A new AI workflow can be tested on Monday and rolled out across the agency by the end of the month.
- Lighter cost base. Without layers of shared services and corporate overhead, savings from AI flow straight to margin or into lower prices that win new business.
- Freedom to reprice. Smaller agencies can move to fixed-fee, project, subscription or outcome-based pricing client by client, without renegotiating a global master agreement.
- Senior talent closer to the work. In a small shop, the most experienced people still touch the work. That is exactly the kind of judgment AI cannot replace and that clients increasingly want to pay for.
- A client base that needs them. Mid-market clients want AI-driven efficiency but cannot build it themselves, which makes a reinvented smaller agency a natural partner rather than a target for insourcing.
The most important shift is capacity. A 50-person team that once could handle 10 clients well may, with the right workflows, handle 25 or 30 with the same or better quality. Each client gets senior attention, faster turnaround and lower cost. The agency earns more per head than it ever did under the old model. That is not a marginal improvement. It is a different business.
None of this is automatic. Small agencies that simply add AI tools to old processes, or keep selling hours, will be squeezed just like the giants. The advantage belongs to those that use their size to redesign the work, the team and the pricing all at once.
What survival looks like for agency giants
The dinosaurs did not vanish overnight. The impact set off a long winter, and the big animals declined as the food chain they depended on collapsed. The holding companies will not disappear in a single year either. They will shrink, merge, sell off units and shed people, and some will find a smaller, sharper model that works. But the era in which size alone was a competitive moat is ending.
The agencies that come out of the next 10 years strongest will share a few traits. They will sell outcomes and ideas rather than hours. They will be built around small teams of senior, AI-fluent people rather than large pyramids of junior labor. They will treat AI as the foundation of how work gets done, not as a feature to pitch. And they will be organized to change again, because this will not be the last impact.
By that measure, the advantage sits with the small and the nimble. For the independent agency under 200 people, the meteor is not an extinction event. It is the moment the landscape opens up.
What the future holds for large holding companies
If the holding companies simply disappeared, nobody would celebrate. It would be bad for people: tens of thousands of jobs, and the careers and craft built inside them, would be lost. It would be bad for the industry, which would lose the institutions that have set standards, trained talent and led the business for decades. And it would be bad for the world’s largest advertisers, who depend on these groups for global reach, scale and coordination that few others can provide.
The real question is not whether the giants will die. It is whether they can reinvent themselves before the winter sets in. They can, but only by giving up the idea of what a holding company is and what it does.
a. From holding company to federation
The path forward is as a federation. Each agency in the network gets real autonomy: its own leadership, its own pricing, its own client relationships and the freedom to rebuild its workflows around AI at its own pace. In effect, every agency in the group is allowed to behave like a nimble, sub-200-person shop. There are already models of this new type of federation. These are networks of small agencies joined by a collaborative mandate, which lets them win more together than any one agency could alone.
The command center gives way to a collaborative hub.
The center stops trying to manage every unit and instead provides the few things that genuinely benefit from scale. These tend to be macro functions like research, media, and high-end creative production. Participation is a choice, not a mandate. Agencies use the shared services when they are the best option for their client, and buy elsewhere when they are not.
b. From media-agnostic to media-owners
Another feature the giants will embrace in this new era is media ownership.
Some might believe this is a new idea. It isn’t. In fact, before digital, the largest agencies created their own media networks as a way to serve their clients.
In the analog era, agencies were much more creative about brand media investments. Agencies produced shows for individual brands such as GM and AT&T. These properties were then aggregated to form a curated network of quality programming that supported many brands at once. This approach lasted for years until the 1980s, when production expenses made it economically unsustainable.
Agency-owned media networks also fell out of favor for another reason: brands’ conflict-of-interest (COI) concerns. Understandably, brands worried that if agencies owned media networks, they would invest brand media dollars in agency-owned properties first, even if it was not the best investment for brands. This raised potential transparency issues which, agencies in a hyper defensive move, wanted to avoid. This is the origin story of why agencies divested themselves of all owned media networks.
The height of irony is that the COI transparency concerns that worried brands are nothing compared with the transparency issues rampant in the digital age.
From MFAs (Made for Advertising) outlets to bots and fake traffic, when agencies exited the media business, the void was filled by a slew of alternative low-quality ad network that are unsafe for brands. Making matters worse, the supply chain is opaque by design, and that opacity is part of how the ad tech intermediaries make money.
Plainly, any trust concerns arising from agency-owned media pale in comparison to the trust gaps in digital media:
- Reaching real people is a trust gap
- Engagement by real people is a trust gap
- Lack of transparency in the amount of ad tech tax that depletes media budgets is a trust gap
- Brand safety is a trust gap
- Ensuring user privacy is a trust gap
While COI is a real issue, an agency-owned network, built properly, brings trust back to a level the industry has not seen in 20 years. Built from a tightly curated group of quality content and publishers, it is organized around topics rather than audience surveillance and scale.
More than that, an agency-owned network can be built from the ground up to be outcome-based. This type of network would be 100% privacy assured and would successfully reach real people based on high-intent topic targeting. Plus, there would be placement transparency and full accountability for attribution.
An agency-owned media network can offer what the open digital market cannot:
- Full transparency. Every placement, site, audience and cost line is visible to the client, with no black box of intermediaries taking a cut along the way.
- Guaranteed quality and brand safety. The agency curates the inventory itself, rather than trusting third-party networks with uneven standards.
- Clear, disclosed fees. No hidden markups or rebates, so the commercial relationship is out in the open.
- Real reach and real user privacy. Since targeting is topic-based, the network can reach real audiences while delivering complete user privacy.
- Cleaner reporting. One network means one consistent data set, so analytics can be built around each client’s real business questions instead of stitched together from many platforms.
- Networks tuned for outcomes. Different publisher groups can serve different goals, such as B2B lead generation or affiliate partnerships, with AI optimizing performance across them.
The technology to build these networks now costs a fraction of what it did a decade ago, which puts them within reach of a federation’s center.
Now is the time for agencies to revisit the outdated COI rule which has clearly outlived its usefulness.
Only the largest agencies can pursue this initiative, but that is exactly where the size of holding groups, heretofore a vulnerability, provides clear value to brands. As important, this approach is a significant new source of revenue for agencies, all the while serving clients extraordinarily well.
This is how the dinosaurs evolve rather than go extinct.
The large holding companies of tomorrow won’t look like they did just five years ago
By 2031, the mega holding company will be organized as a federation rather than an empire. Its member agencies will each have real autonomy, and the center will provide shared AI infrastructure, data, media networks and scale only where they truly add value. Its two core functions will be rebuilt from the ground up. Creative will move from producing assets by hand to directing AI systems that generate, test and refine work at speed. Media will shift from buying in an opaque marketplace to running curated, transparent media networks that agencies can choose to use, with performance measured against real business results.
The people will change too. Teams will be smaller, more senior and fluent in AI, with their time spent on strategy, judgment, creativity and client relationships rather than production and process. Most important, the value proposition will change. Instead of being agency partners paid for hours and deliverables, these companies will become true business partners, accountable for growth and paid for outcomes. In short, they will evolve from agencies into agents of growth (the tagline of our own agency – engageSimply).
New lands large holding companies must traverse to survive.
I speak from a “full circle” perspective. I worked at a large agency holding company for over a decade. Then I went client-side, working for some of the largest brands and tech firms on the planet. For the last 13 years, I have run a tech-agency: marketing tech at the core with agency services concentrated on delivering outcomes.
To borrow from Joni Mitchell, I have seen “the world from both sides now.” I am more optimistic than ever about how this plays out for those large holding companies with bravery, insight and a willingness to look at the world as it will be – not as it was.
The ideas are not new, but how they are expressed is – as is laid out in the chart below.
| 25 Years Ago | Today | Next 10 Years | |
| Revenue Model | 15% commission on media spend plus fee to cover labor | Tortured mix of arbitrage, labor fee and some data management/ special projects | 15% commission on owned media plus fee to cover OOP expenses. There are new productized technology services for reliable and repeatable revenue. |
| Agency’s main superpower | Business partners to clients | Marketing partners for clients | Business partners to clients |
| Staffing | Senior-leader centric | Junior-level centric | Tech-savvy centric |
| Creative | Slow and expensive | Fast and expensive | Fast and inexpensive |
| Insource/ Outsource balance | Clients mostly outsourced | Hybrid in/ out sourced | Shared tech stack run by clients and agencies together |
The technology is the unlock to save agencies from the AI meteor. Using technology to reclaim the agency’s standing as a trusted business partner, requires a willingness to travel to new territories that have enough pastures to sustain the species. It is a perilous journey, full of dangers, but the alternative is – frankly – a whole lot worse.
