The holding-company CEO said it out loud. “By 2028, we’ll double profits and halve the people.” Forrester printed the quote in its 2026 agency predictions. Then Forrester ran the numbers and made the call: 15% of agency jobs gone in 2026. Not eventually. This year.
You read that right. A decade of “consolidation” talk just landed on the desk of every media buyer, creative director and CEO in a glass building in New York, London and Tokyo.
And they have no one to blame but themselves.
The shot everyone heard
Omnicom closed its $9 billion merger with IPG on November 26, 2025. By December, the new Omnicom had shuttered FCB, DDB and MullenLowe and cut about 10,000 jobs. The CEO called parts of it “shocking.” The press called it rationalization.
The market called it the beginning of the end.
Publicis now sits at a market cap of $26.3 billion. The combined Omnicom at $25 billion. WPP at $4.8 billion below where it was a decade ago, and recently bounced out of the FTSE 100. Havas weighs in at $1.8 billion. Forrester’s principal analyst Jay Pattisall now calls it “the big three: Omnicom Group, Publicis Groupe and a WPP variant.”
“Variant” is a polite word.
The new buyer doesn’t want you
For decades, the agency contract was a marriage. Now it’s a TaskRabbit gig.
In the same Forrester prediction cycle, 85% of US B2C marketers said they planned to review their media agencies in 2026. For WPP, that review happened in real time. Coca-Cola invited Publicis to pitch the $4 billion global media account WPP has held since 2021 a relationship that employs around 5,000 WPP staffers and was supposed to be “the catalyst in the transformation of marketing effectiveness.” Omnicom just stole Adidas’ $512 million global media account from WPP’s incumbent EssenceMediacom.
The market hasn’t lost trust in agencies. It lost the patience to pay for them.
Your client already moved on
Look at the math a CMO is staring at right now. It’s ugly for the agency CFO.
88% of marketers now use AI daily. 93% say they create content faster because of it. The average payback period for AI investment has collapsed from 7.8 months in 2024 to 4.2 months by the end of 2025. The average marketer now reclaims 6.1 hours per week.
Do the math on what that means for a $400K-a-year retained agency.
And while you’re doing it, 39% of CMOs have already told Gartner they plan to cut agency budgets in the next planning cycle. More than a third. In one quarter.
WPP’s top 25 client spend was down 9.4% year-on-year in Q1 2026. North America revenue fell 7.8%. The UK fell 6.6%. Headline revenue dropped 6.6% to £3.03 billion. And the CFO, asked if clients were buying Elevate28, wouldn’t give a number on gross wins she just pointed at “Net New Business momentum.”
The trend line is the pitch deck.
The solo founder is the punchline nobody prepared for
In April 2026, the New York Times ran the headline that marketing’s biggest agencies didn’t want you to read: “A $1.8 billion company with just two employees? In the age of AI, it’s increasingly possible.” The company was MEDVi, the telehealth GLP-1 startup.
Two employees. $1.8 billion in revenue. AI agents doing what 200 marketers used to.
Then the FDA sent a Warning Letter. The Decoder reported MEDVi had used AI to “create ethically questionable advertising, fake doctor profiles on social media, fabricated videos, and generated before-and-after comparisons.” So the unicorn turned out to be ugly.
But the structural argument survived the fraud case. Sam Altman had predicted the one-person $1B company in 2024. Two years later, it’s no longer a punchline. It’s a template.
Gartner now predicts AI coding costs will surpass the average developer’s salary by 2028 as token consumption surges. Gartner separately put $234 billion in enterprise application software spend “at risk from agentic AI.” The AI coding line items are the agency retainers of 2028.
What the founder actually does at 2 AM
Here’s the part the trade press doesn’t tell you.
Rory McEntee is the CMO of GymNation. A year ago he wrote a piece saying AI could replace his creative agency. Adland wanted him tarred and feathered. A copywriter canceled his GymNation membership “and was sleeping better than ever.”
McEntee just wrote the follow-up. His headline: “I was wrong. I didn’t go far enough.”
Twelve months in, his marketing stack publishes SEO content daily across multiple markets without a brief or a retainer. Generates and tests hundreds of paid-ad creative variants simultaneously across every platform. Runs a press office that drafts in his voice, references recent bylines, and pitches specific journalists. He rebuilt the function with a leaner, AI-augmented team. Down at the junior level. Up at the strategic level.
His most recent campaign a Protein Shisha Café concept generated 836 press mentions across 23 countries, 150 influencers organically, and 16.5 million impressions in five days. The original idea came from an agency partner. The execution came from a stack McEntee owns.
That’s the play. Senior strategy in. AI scale out. Agency as occasional sparring partner, never as full-time vendor.
The holdcos are selling you the panic
Watch what the survivors do when their own clients vanish.
Publicis just acquired LiveRamp for $2.55 billion in a bet that “agentic AI” is the future of marketing data. WPP launched WPP Enterprise Solutions already a $1.8 billion business, 13% of group net revenue explicitly to compete with Accenture and Deloitte on AI transformation consulting. Omnicom is busy axing brands nobody wanted to see die.
David Droga, who just stepped back as CEO of Accenture Song, put the diagnosis on the record: “The talent that resides in the marketing and advertising world is second to none, but the business model is broken.” His prescription for WPP: take it private. “I think you take it private so you can actually do what needs to be done.”
Read that again. The man who built Accenture Song into the third force in marketing just told the industry’s most storied holding company to break itself.
The strategist who runs the new ADWEEK House panel on agency execs “leaving holdcos” Tombras’ Jeff Benjamin, Known’s Kern Schireson, Gut’s Anselmo Ramos, nice&frank’s Laura Petruccelli told the audience they found “freedom” outside the holding companies.
The talent isn’t leaving the industry. The talent is leaving the structure.
The in-house wave is now a tsunami
Publicis’s content engine Unilever has 300,000 creators on its books now up from 10,000 two years ago. It’s spending 50% of its $9 billion marketing budget on social channels and wants a brand influencer in every one of India’s 19,000 zip codes.
The creator economy globally is now projected at $480 billion by 2027, per Goldman Sachs’ 2023 baseline that has only been revised upward since. After Unilever’s announcement, creator rates jumped 30% in micro-influencer brackets in a matter of months. Billion Dollar Boy saw a 22% rise in spending from existing clients between May and July 2025 versus the prior year. Influencer marketing hit 6% of marketing budget at the typical Gartner-tracked CMO in 2024 and the survey data captured before the pivot.
That’s not an “in-housing trend.” That’s the agency layer being routed.
Forrester says 78% of the top 80 digital media agencies now have PE or VC money on their balance sheet. Total PE investment in marketing businesses rose 21% in the last year, with digital/social/influencer agencies taking 50% of all PE dollars three times last year’s share. For the first time, bolt-on acquisitions fell by 92%. Read that again. PE stopped paying for agency roll-ups. It’s paying for tech-leveraged, AI-native specialist shops.
The “biggest buyer” in this market isn’t a brand anymore. It’s a buyer whose job is to flip the thing it bought before the AI curve broke.
The four archetypes (and why none are agencies)
In its 2026 predictions, Forrester named the four new models replacing the agency-as-agent:
Vendors. Executing programs. Merchants. Reselling software and media. Affiliates. Plugging into a bigger matrix. Partners. Doing strategy and client-centric work.
Notice what’s missing. None of those four is a creative department you sit with for six weeks. None is a media planning team that lives in your Slack. All four push the agency to a transactional seat at a table where the AI is now the default author, the creator is the default channel, the platform is the default data steward, and the brand is the default strategist.
The exit door was visible three years ago. Bose’s CMO hasn’t used a creative agency in five years. When one of the most premium consumer brands on earth told Adweek it doesn’t hire agencies anymore, Adweek put it under a “Briefing” header. They should have put it on the cover.
What “recovery” actually looks like
There is one version of recovery. It isn’t pretty.
It looks like Publicis: leaner, data-led, AI-orchestrated, integrated. The only holding company still winning new business at scale. Publicis grew organic revenue ahead of plan in 2025, citing “booming demand for AI,” then bought Lotame for an identity-based data moat, then LiveRamp for another $2.2–$2.55 billion. It is becoming a tech company with an ad agency attached. Whether that survives the next AI wave is a question Publicis’s Arthur Sadoun cannot answer in public.
It looks like WPP, betting the ranch on AI: Bulchandani publicly says “our level of confidence has never been stronger” while the share price slumps toward a multi-decade low. That is not confidence. That is a leader performing conviction under existential stress.
It looks like a constellation of indie, AI-native, senior-only shops Pendulum, nice&frank, Tombras, Gut bagging Cannes Lions against the holdcos while running on a tenth of the headcount. The hybrid model Mark Listes described at Cannes “AI handles research, synthesis and formatting while humans focus on original analysis and brand voice” is now the only one gaining market share.
It looks like McEntee’s GymNation stack. Half a dozen senior marketers, agent-led execution, agencies retained as occasional creative ignition sources. That is the prototype of the surviving agency relationship: the “agency as idea spark, in-house as everything else” model that wins when 88% of marketers are already using AI daily.
It doesn’t look like the agency business you signed up for. Not even close.
The thing nobody inside the building can say
Open this.
HubSpot’s 2026 State of Marketing report finds 61% of marketers believe marketing is experiencing its biggest disruption in 20 years and 80% now use AI for content creation. Salesforce’s 10th edition State of Marketing, surveying 4,500 marketing leaders, says 83% recognize the shift to personalized, two-way messaging but only one in four are satisfied with how they use data to power it. Forrester’s B2C CMO pulse survey for Q3 2025 shows 64% of marketing executives expect 2026 to be more volatile than 2025.
That’s the environment in which your agency is asking you to re-sign for three years.
Forrester’s prediction: three AI-native B2C marketing technology applications will enter the market in 2026, with one major vendor and two startups expected to debut next-generation AI-native solutions. Confidence in marketing measurement will fall another 7% as AI-driven data transparency erodes trust in the legacy models agencies rely on. By Forrester’s count, 79% of B2C marketing leaders felt confident in 2025 and that number is heading lower.
When the buyer’s measurement goes away, the seller’s billable hour has nothing to hang on.
What the solo founder actually does to you
Look. There’s a version of this story that’s a tech-bro fantasy about two-person unicorns. Forrester’s J.P. Gownder already wrote the takedown AI “miracles” that turn out to be AI-generated fakery don’t count.
But that takedown missed the real point. The solo founder killing the agency isn’t the one-person billion-dollar company.
It’s the 50-person SaaS founder with $14K/month in AI tools, three retained contractor specialists, and zero in-house marketing leadership. It’s the seed-stage founder who used to spend $180K/year on a fractional CMO + agency combo and now spends $36K/year on Claude, a fractional growth advisor, and a part-time BDR. It’s the bootstrapped founder shipping five product updates a week, then firing GPT-4o-class agents at LinkedIn, blog SEO, and lifecycle email without a soul in the middle.
These founders don’t write press releases about it. They just quietly stop sending the agency RFP.
The agency revenue doesn’t decline with a press cycle. It declines with an attrition cycle. And by the time you notice it on the macro chart, it has been gone for six quarters.
Dentsu tried to sell its international arm last year. No one bought it. Bain Capital was reportedly the last suitor standing and even they walked. The FT quoted both strategic and PE bidders walking away “from serious talks.” When the international arm of a 110-year-old Japanese holding company can’t find a buyer, that is not a pricing problem. That is a category in liquidation.
The honest 18-month forecast
Here is what I think actually happens between July 2026 and December 2027.
-
Headcount cut compounds. Forrester’s 15% headline gets revised to 22% by Q4 2026 as AI absorbs tasks agencies still bill for. The “halve the people” holdco CEO looks prescient.
-
One more megamerger tries to hide the bleed. Rumored WPP-to-Accenture or WPP-to-private-equity deal circulated all Cannes Droga called it plausible only if WPP goes private first.
-
The agency-of-record gets unbundled aggressively. Private equity buys 10 creative shops in 2026, per Forrester. Then it sells them again as components the in-house teams at Unilever, Coca-Cola, Bose and Nike want to buy outright.
-
Principal media goes mainstream. Forrester says principal agencies selling media as principals rather than agents will hit almost a third of all agency billings in 2026. 81% of marketers are increasing principal-media spend. That’s not a margin improvement. That’s the agency surrendering its agent status.
-
The CME-CMTO brand collapses into one. Roughly half the CMO function gets merged into a CTO/CMO hybrid role, with AI fluency as the deciding skill not “the big idea.” Forrester predicts the confidence in marketing measurement falls another 7%. The new seniority is whoever owns the model that proves pipeline.
-
The indie AI-native agency becomes the winning template. Senior, small, AI-amplified, creator-orchestrated. Outside the holding companies. Charging premium. Out-delivering the holdcos on the work that wins awards.
Forrester’s 75%-of-agencies-absorbing-AI-costs-and-only-6%-monetizing-AI-work stat is the single most telling number in this cycle. Agencies are losing margin on AI even while their clients use it to fire them.
The verdict nobody wants to hear
There is no recovery for the agency as it was.
The 12-person account team. The quarterly planning cycle. The retainer that escalates 4% every year. The “junior + senior + EVP” pyramid. The “media-neutral agnostic partner” pitch. The “we know your brand” defense. The “we’ve always done it this way” culture. All of it is being repriced down to a transaction that AI does better, faster and cheaper.
What survives is a sliver. Two things, specifically.
The senior strategic idea. The original creative ignition. The connection to a real human truth that an LLM trained on Reddit and Stack Overflow will not surface. The “I had an idea in the shower and it changed the brief” moment.
And the senior strategic execution of that idea at the places AI can’t yet reach high-stakes celebrity deals, in-the-room cultural intelligence, brand-positioning arbitration with the board.
The rest is now either a feature in a SaaS stack, a freelancer on a retainer, or a contractor on a fractional CMO’s roster.
The solo SaaS founder didn’t end the agency model out of malice. They undercut it by accident, with a $20/month AI subscription, a fractional advisor, and the same brand instinct that used to belong to the agency. They just stopped noticing the agency was there.
The agency looking at itself in the mirror in 2026 has a choice. Pretend the old retainer model still works and ride compounding price pressure, or become the smaller, more senior, AI-native version the next 24 months actually rewards.
Most holding-company structures will struggle with the second path. Their cost base depends on not making it. That is a business-model problem, not a moral one, and recovery means reinvention, not a return to 2019.
