The very concept of AI and its impact on select industries is enough to cause dramatic swings in stock market sectors and wild future-of-work predictions. But when the world’s largest ad agency talks about changing its revenue model – and trots out a real-world client example to back it up – people pay attention.
In late February, WPP, citing Jaguar Land Rover as an example, signaled its intent to move to fee structures tied to measurable sales and brand performance as opposed to an hours-worked model. The general logic is: As AI enables exponentially more creative output and faster, more adaptive media strategies with less human intervention, the idea of charging clients for hours worked by talent falls by the wayside.
Performance over process
WPP’s bold stance is a twist on earlier statements coming from Martin Sorrell’s S4. By the end of 2026, its Monks agency will derive 25% of its revenues from a subscription model, via which clients will not pay by the hour for tasks but rather subscribe to a suite of solutions.
These solutions are aimed at solving problems, not delivering units of creative or hours of media strategy. While this isn’t quite pay-for-performance, “performance” is implied.
There are many barriers to clear before pay-for-performance (PFP) becomes the industry standard:
- The ability of corporate finance departments to budget fees in advance. In speaking confidentially with several large consumer brands, it’s clear that despite the appeal of skin-in-the-game incentives for their agency partners, the sheer logistics and corporate fiscal policy would make a clean swap to a PFP model difficult to pull off in the short term. B2B companies are more likely to be open to a PFP relationship. In some specific and contractually defined instances, some are already structuring deals on a cost-per-lead or cost-per-acquisition basis.
- The unknown – and potentially volatile – cost of AI compute. Early pushback on a more generalized PFP model, especially for consumer brands, echoes concerns around media arbitrage, or principal-based media buying. Much as historical media arbitrage raises questions regarding whether agencies are selling (or reselling) the best media to meet a client’s needs, the early complaint around AI compute stems from agencies buying bulk compute and reselling it at a premium. For agencies, the risk comes from not buying AI compute up front and learning later that their costs have gone through the roof.
But the big question will be around measurement.
If the difference between getting paid and getting fired comes down to how sales are measured and mapped back to media, agencies and brands have a lot to lose if the measurement goes wrong.
Most agency measurement solutions today fall back on platform-based ROAS numbers and “closed-loop” measurement that is only closed-loop if you ignore all the variables that contributed to sales beyond the advertising execution and outside an agency’s purview.
Despite brands’ persistent suspicion around agency- and platform-provided ROAS, the brands have less risk in a PFP economy than their agency counterparts do. Imagine a scenario where a brand and an agency enter into a PFP agreement and the agency sets about delivering world-class creative targeted with AI-enabled precision to the perfect audiences at the perfect time.
At the onset of this imagined campaign, sales take off and brand health metrics start climbing. But after a brief time, COGS issues force the brand to raise price, inflation skyrockets due to geopolitical issues beyond everyone’s control and a formerly exclusive retail partner adds a primary competitor to its shelves.
Without a robust measurement solution that both quantifies the effects of these events and delivers them in a manner all parties can believe, the agency will be wrongly punished for a brilliant campaign; brand executives will look at top-line sales and assume the worst. A potentially fruitful partnership will be dead within a year.
This can be avoided if all parties have a trusted referee that understands how to build measurement models that account for more than just media and has no vested interest for either positive or negative ROAS outcomes.
These referees already exist. Third-party marketing mix providers are equipped to act in this role today. Plus, compute power and modeling techniques have advanced such that they can provide comprehensive measurement with frequency and scale. The old days of once-per-year marketing mix models with outdated results and delayed strategy consultation are long behind us.
The days of an AI-driven PFP model for the ad industry may still be off in the distant future. But the referees are ready once they’re needed.
“Data-Driven Thinking” is written by members of the media community and contains fresh ideas on the digital revolution in media.
Follow Big Chalk Analytics and AdExchanger on LinkedIn.
For more articles featuring Rick Miller, click here.
