Four CMO conversations on what changes when the constraint moves from producing creative assets to routing the right one to the right customer.
Four conversations, one common thread. Over the last quarter we spoke with CMOs at a global streaming platform, a European telecom operator, a subscription commerce brand and a retail loyalty program. Different categories, similar scale, wildly different tenures. They converged on the same observation without prompting: the constraint on their function has moved. The old constraint was producing enough creative to feed the calendar. The new constraint is deciding, per customer and per moment, which of the many assets they already have should ship.
For most of the last twenty years the marketing function was organized around asset scarcity. A campaign meant a shoot, a brief, a review cycle, a legal check and a launch. The rate-limiting step was the production line. Everything downstream, media planning, targeting, measurement, was arranged to make the most of a small number of expensive artifacts.
Generative tooling has, over the last two years, collapsed that production line. All four CMOs used some version of the phrase “creative is basically free now,” and none of them meant it as a boast. They meant it as a problem. Their teams can produce hundreds of viable variants per week. Their systems cannot decide which of those variants should ship, to whom, in what sequence, with what guardrails. The constraint moved. The org chart did not.
Decision volume is not a synonym for message frequency. Nobody is arguing for sending more email. Decision volume is the count of distinct, causally read routing choices the system makes across the customer base per unit time. A team that ships one message per customer per week can still be making millions of decisions per week if each of those messages is chosen from many candidates on a per-customer basis, with a preserved holdout.
Every CMO independently described the traditional creative brief as an artifact whose value had collapsed. A brief was useful when it disciplined a scarce resource. When the resource is abundant, the brief becomes a delay. Three of the four had already restructured their creative teams around hypothesis-led production rather than brief-led production.
The consensus was blunt. The competitive advantage in the next cycle will not accrue to whoever produces the best assets. It will accrue to whoever can causally measure which of their many assets is producing incremental ARPU on which segment. The moat is the readout, not the render.
A generative production line can produce ten thousand variants in the time it took to legal-review one. The only durable answer is to move brand and legal review upstream, into the guardrails on generation itself. Every CMO named this as an unfinished project and none of them had cleanly solved it.
Two of the four had already created a role variously titled Head of Decisioning, Head of Growth Systems or Head of Revenue Intelligence. The remit is identical across the two: own the operating model that connects opportunity to experiment to shipped variant, and defend the incrementality report against everything the rest of the organization wants it to be.
The interesting divergence was on where the decisioning layer should sit in the org. Two argued it belongs in marketing, because the artifacts and the customer relationships live there. Two argued it belongs in product or data, because the systems and the measurement discipline live there. Both camps were confident. Neither camp had seen the other model work at scale.
Our observation, offered without a strong view, is that the reporting line matters less than the fact that the role exists at all and reports high enough to defend the measurement discipline against short-term revenue pressure.
The most memorable line from the four conversations came from the retail loyalty CMO. She said the change felt less like a technology upgrade and more like a discipline upgrade. Her team no longer argued about whether a campaign was any good. They argued about whether the experiment behind it had been fair, whether the holdout had been protected and whether the readout could survive finance. Everyone found the second conversation more useful.
The operating model these CMOs described is the same one we wrote about in From campaign calendars to continuous decisioning. The team shape it implies is the one we wrote about in What a top data science team looks like in an agentic era.
Frequently asked
See it in the product
The same loops this note describes run 24/7 against your customer base. Watch the workspace decide, experiment and execute 1:1.
