The False Security of the Retainer Model

Sir Martin Sorrell recently suggested that the primary barrier to outcome-based agency pricing isn't technical capability, but fear. Specifically, he argues that Chief Marketing Officers are reluctant to abandon traditional billing models because they are terrified of procurement departments accusing them of "overpaying" if an agency performs too well.

This is a critical insight into the psychology of enterprise capital allocation. While the marketing industry obsesses over AI transformation and efficiency, the financial structures governing ad spend remain deeply conservative. Sorrell points to a stabilizing economy and rising corporate earnings as a counter-intuitive brake on innovation. When the P&L looks healthy, the appetite for structural risk evaporates.

For founders and growth leads, this signals a dangerous disconnect. The technology driving media buying is becoming exponentially more autonomous and outcome-oriented, yet the contracts governing these relationships are stuck in a labor-based past. Understanding why this friction exists—and when it will break—is essential for anyone managing a significant media budget in 2026.

The Procurement Paradox

The logic Sorrell outlines is perverse but rational within the confines of a Fortune 500 organization. In a stable economic environment—with S&P 500 earnings per share up and forecasts predicting double-digit acceleration—predictability is valued over efficiency. A CMO is safer paying a flat retainer for guaranteed mediocrity than signing a performance deal where the agency might earn a massive upside commission.

If an agency crushes a performance target and sends an invoice for double the expected amount, the CMO has to explain that variance to a CFO. Even if the ROI is positive, the variance itself is treated as a budgeting failure. This creates a perverse incentive structure where clients implicitly cap their own upside to avoid uncomfortable conversations with procurement.

This is why the death of the billable hour has been predicted for a decade yet never arrives. The billable hour is a hedge against risk. It turns marketing into a fixed operational expense rather than a variable cost of goods sold. For agile founders and aggressive growth teams, however, this mindset is fatal. You should want your agency to make a fortune, provided that fortune is a fraction of the net new revenue they generated for you.

The Inference Cost Time Bomb

While human psychology delays the shift to outcome pricing, technical reality is forcing the issue. S4 Capital executives have rightly identified "cost of inference" as the coming battleground. As agencies integrate agentic workflows, the primary cost driver shifts from human labor hours to compute costs paid to providers like Nvidia or OpenAI.

This breaks the traditional agency margin model. In the past, agencies marked up human time. Now, they are facing a future where they must mark up compute cycles. If an autonomous agent runs thousands of iterations to optimize a campaign, billing that time as "hours" is fraud. But absorbing the compute cost without charging for it destroys the agency's bottom line.

We are approaching a tipping point where the cost of running AI models at scale becomes significant enough that it cannot be hidden in a general overhead fee. When that happens, agencies will be forced to bill based on outcomes or output, simply because the inputs (compute) are too volatile to predict in a fixed retainer. The stability that CMOs currently prize is an illusion that will shatter once inference costs hit the agency P&L.

Winners, Losers, and Capex Moats

The broader implications here favor the platforms over the intermediaries. Tech giants are deploying over $530 billion in capital expenditure to build the data centers and infrastructure that power these AI models. They own the rails. Agencies are merely renting space on them.

This structural disadvantage for agencies is compounded by consolidation. As holding companies merge to protect margins, they create confusion and bloat. This leaves a massive opening for independent, agile shops that don't have to protect legacy headcount. The winners will be agencies that can operationalize high-cost compute to deliver outsized returns, effectively acting as arbitrageurs between platform capability and client revenue.

The losers will be the "safe" CMOs who cling to headcount-based retainers while their competitors adopt performance models that incentivize aggressive scaling. In a stable economy, you can hide inefficiency. But the moment the market turns, or the moment a competitor figures out how to weaponize agentic workflows, the safe route becomes the high-risk route.

Aragil POV: Strategic Implications

If we were advising a client based on this landscape today, we would immediately audit the incentive structure of their agency contracts. If you are paying a flat fee regardless of performance, you are incentivizing your partners to minimize their effort, not maximize your growth.

We would push for a hybrid model immediately. Cover the agency's base costs—including the rising costs of software and compute—but tie the profit margin entirely to hitting CPA or ROAS targets. This aligns the "fear" Sorrell speaks of. If the agency over-performs, the "overpayment" is funded directly by the excess revenue generated.

We are also monitoring the ratio of strategy to execution in billing. As execution becomes automated and commoditized, the value of the agency is no longer in pulling levers, but in designing the machine. We expect to see smart founders demanding transparency on "inference" costs versus "talent" costs within the next six months. Do not accept a black box fee structure.

The most common mistake teams will make is assuming that AI efficiencies should simply lower their agency fees. This is short-sighted. The goal of automation isn't to make marketing cheaper; it's to make it scalable. The money you save on human hours should be reinvested into compute and media spend, not pocketed as savings.

Conclusion

Sorrell's comments highlight a cultural lag in the industry. The technology for outcome-based marketing is here, but the corporate immune system is rejecting it. This creates an arbitrage opportunity for smaller, more aggressive companies.

While the enterprise giants worry about procurement rules, you have the freedom to align incentives perfectly. Pay for results. Let the agency worry about the cost of inference. If you get the economics right, you should be happy to write the biggest check of the month to your growth partner.