Forrester's prediction is already obsolete — not because they're wrong, but because they're late. The 15% is happening now. The agencies dying are dying of the same thing: they built staffing pyramids, called it a business, and handed clients execution instead of outcomes. The agencies surviving are doing something different. They are building growth engines their clients cannot run without. That distinction is the whole game.
What Forrester Actually Said
In late 2025, Forrester threw out its own prior forecast. Back in 2022, they projected 7.5% of US agency jobs automated by 2030 — roughly 32,000 positions across eight years. Then they doubled the severity and compressed the timeline to one year. Their new call: 15% of agency jobs gone in 2026, following an average 8% headcount cut across holding companies in 2025.
WPP cut 9,000 roles. IPG shed 3,200 before the Omnicom deal closed. Omnicom then announced 4,000 more cuts immediately after absorbing IPG and retired DDB, FCB, and MullenLowe as independent brands. The "big six" holding companies are collapsing toward a "big three." This is not a forecast. It is a casualty drill playing out in quarterly earnings calls.
Here is the number that tells the real story: worldwide ad spending grew 8.6% in 2025. Holding company revenues fell 1.2% over the same period. The market grew. The agencies lost ground. That gap is structural, not cyclical.
The Workforce Inversion
Forget the headline job loss number. Understand the mechanism.
The traditional agency made money through labor arbitrage. Expensive senior talent supervised cheap junior talent. Junior staff did the volume . research, coordination, reporting, data entry, proposal drafts. The spread between what clients paid and what juniors cost was profit. Forrester calls what is happening now the "workforce inversion." AI erases the junior layer. The model flips to small senior teams working directly with AI systems. You do not need ten researchers when three people with the right tools produce the same output in a third of the time.
Forrester breaks down the 2026 losses: 28% are clerical and administrative roles. 22% are sales and business development roles. 18% are market research positions. These are not creative or strategic losses. They are execution losses. And execution is exactly what agencies have been selling.
I Learned This at Hartford
When I worked with Hartford, I got a look inside Munich Re's innovation operation. Fifty-five thousand employees globally. Around fifteen dedicated innovation scouts. The ratio is jarring until you understand what the scouts actually did. They were not executing tasks. They were building systems . identifying signals, creating frameworks, running experiments that scaled. The company did not reward the people who did the most work. It rewarded the people who built the most use. The scouts who just executed projects eventually got absorbed or replaced. The ones who built systems became indispensable.
That is the same bifurcation happening in agencies right now. The execution-only shops are getting absorbed or replaced. The ones building systems are becoming infrastructure.
The Owner-Operator Frame: Two Kinds of Agency
Run this through the Owner-Operator Frame and the split is clean.
Type one: the execution agency. Retainer covers headcount. More clients means more hires. Value proposition is "we do the work." When AI can do the same work faster and cheaper, the client does not need the agency anymore. The owner is trapped in the business. The business is not sellable at any meaningful multiple because there is nothing proprietary to sell.
Type two: the growth infrastructure agency. Retainer covers a system . AI-powered delivery, outcome tracking, proprietary playbooks, compounding client data. More clients does not necessarily mean more headcount; it means the system runs more accounts. Value proposition is "we run your growth operation." When AI accelerates the system, the agency becomes more valuable, not less. The owner builds something acquirable.
PE buyers understand this already. 78% of the top 80 digital media agencies have taken PE or venture capital. Forrester predicts 10 major creative agency acquisitions in 2026 alone . deals designed to bolt AI-delivery capability onto existing PE holdings. The agencies getting acquired are not the execution shops. They are the ones with systems, recurring revenue, and defensible client relationships.
The Valuation Math Is Brutally Clear
An execution agency billing $3M a year on retainer trades at 3-4x revenue if it trades at all. Client concentration risk is high. Churn risk is high. Margin is thin. There is no real multiple to speak of because there is no real asset . just a roster of clients who could leave tomorrow and a team that walks out the door with them.
An infrastructure agency billing $3M a year on recurring outcome-based contracts trades differently. Gross revenue retention above 90%. Gross margins in the 60-70% range. Proprietary delivery playbooks. Documented client results. That business starts at 6-8x EBITDA and can reach higher when a strategic acquirer needs to bolt on delivery capability. The owner has built something sellable. The multiple reflects a real asset.
The math on compounding is even starker. Agencies that shifted to productized, AI-powered delivery report deployment times dropping 83% . from 30 days to under 5. Client retention moves from 62% to 88%. That retention gap compounds over three years into a business that is structurally worth two to three times more than the execution shop next door.
Who Survives: The Doctrine Check
Competence beats credentials. That is the doctrine here. The agencies surviving are not the ones with the most awards, the biggest pitch decks, or the longest client lists. They are the ones who can demonstrate that they run their clients' growth better than the clients can run it themselves . with or without a full team.
Here is what the surviving agencies look like in practice. They walk into new business with a specific playbook for the client's category. They deliver 30-60-90 day result milestones. They show compounding returns: client data gets richer, attribution gets sharper, and campaigns get more efficient over time because the system learns. The retainer is not the product. The growth engine is the product. The retainer is just the access fee.
This is the Owner's Exit Engine in reverse. Instead of building toward an exit, these agencies are building the kind of infrastructure that makes them worth buying . and the kind of ongoing value that keeps clients from leaving. Both outcomes depend on the same thing: a system that runs, not a team that executes.
The Agencies Already Dead
The execution-only shops are already dead. Most just have not filed the paperwork yet.
Their bottleneck is structural. They cannot cut enough headcount to survive the margin compression without also cutting the delivery capacity their clients pay for. They cannot raise prices without demonstrating differentiated outcomes they do not produce. They cannot sell because there is nothing to acquire that the buyer could not build cheaper.
One global holding company CEO put it plainly: "By 2028, we will double profits and halve the people." That is not a strategy. That is an autopsy.
The Agencies Getting Rich
The agencies building AI-delivery models are not just surviving the compression . they are accelerating through it. When your competitors are cutting 15% of their capacity, your capacity stays constant or grows because your engine room does not depend on headcount. You take their clients. You take their talent. You run both through a system that compounds.
85% of US B2C marketing executives plan to review their media agencies in 2026. That is a massive buyer pool actively shopping for a better answer. The agencies with documented systems, outcome track records, and recurring-revenue structures are going to win a disproportionate share of those reviews. The execution shops are going to lose them.
The doctrine check is simple: can you show a client what their growth trajectory looks like if they work with you for three years? Can you show them what the system produces, not just what your team does? If yes, you are in the right half of this split. If no, the casualty drill is already running.
FAQ
Q: Does the 15% Forrester forecast apply to all agencies or just holding companies?
The forecast applies broadly. Forrester's analysis covers the full spectrum from holding company giants to independent shops. The structural forces . AI replacing execution roles, clients insourcing, procurement pressure . hit every agency running a labor-based model. The 15% number reflects the industry average. Execution-only shops will see cuts steeper than 15%. Agencies with AI-delivery systems will see headcount stay flat or grow as they absorb displaced clients from failing competitors. The split is not about size. It is about model.
Q: What does an "AI-delivery model" actually mean for a mid-size agency?
It means your agency delivers outcomes through a documented system, not through billable hours. You have playbooks that define what you do for each client category. You use AI to run research, content, reporting, and optimization at speed. You charge for results or access to the system, not for time. Concretely: a 10-person agency with AI delivery can run 30-40 client accounts that a 40-person execution agency would need twice the headcount to service. That margin gap is what makes you acquirable.
Q: How does this affect agency valuations and exit potential?
Directly and significantly. Execution agencies trade at low multiples . often 3-4x revenue . with high churn risk and no proprietary assets. AI-delivery agencies with 60%+ recurring revenue, documented outcome track records, and scalable systems can trade at 6-8x EBITDA or higher when PE buyers are consolidating delivery capability. The difference in exit value between the two models, on identical revenue, can be $5-10M or more. Building toward that multiple starts with one decision: sell the system, not the team.
Q: Is this really about AI, or is it about business model?
Both, inseparably. AI is the mechanism that makes outcome-based delivery economically viable at scale. Without AI, you cannot produce the volume of work required to justify a system-based retainer without also hiring the headcount that kills your margins. With AI, a small team runs the engine room and the system compounds. The business model shift . from execution to infrastructure . is only possible at the margin structures you need because AI changes the unit economics. You cannot do one without the other in 2026.
Q: What is the first move for an agency owner who recognizes they are in the execution trap?
Audit your retainers. For each client, ask: are they paying for our time or our outcomes? If the honest answer is time, you have a project shop with a retainer wrapper, not a growth infrastructure business. The first move is picking one client vertical and building a documented delivery playbook for it . defined inputs, defined outputs, measurable milestones, AI-powered execution. Run one client through it. Document the results. That one case study is the proof point that starts the model shift. The doctrine is simple: competence beats credentials. Show what the system produces, not what your team can do.
Doctrine Connection: Competence beats credentials. The agencies winning in 2026 are not the ones with the most creative awards or the longest pitch decks. They are the ones who can show, with documented systems and compounding client results, that they run growth better than anyone else. That competence . built into a repeatable engine . is what gets acquired. Everything else gets cut.
*Jeff Barnes is the founder of DEMG.ai and has been building marketing systems for owner-operators since 2023. He has no commercial relationship with any tool or platform named in this article unless explicitly stated. This content is educational, not professional advice. Your results depend on your execution.*