Agents Are Commoditizing Lead Gen. Revenue Architecture Shows Marketing Where to Go Next.

Seventy percent of CMOs now call becoming an AI leader a critical goal for the year, but only 30% report the internal maturity to scale it. What that number hides is a prior question. Most marketing organizations cannot say which work they should hand to agents in the first place, so the ambition to deploy AI arrives with nowhere specific to put it.

Sascha M. Kimmel runs strategy and planning at WPP and says the map already exists. He points to revenue architecture, the framework built by Jacco van der Kooij of Winning by Design, which lays out the full commercial journey as a bow tie rather than a funnel that ends at the sale. The tension it exposes is specific: AI automates the execution-heavy, attribution-friendly layer that the function spent years building around, leaving teams a choice: defend work that platforms now perform cheaply, or use the freed capacity to own more of the customer journey after acquisition. Acquisition sits on the left; onboarding, retention, and expansion form an equally instrumented right side. That shape, for Kimmel, is where AI actually pays off.

He frames it less as a marketing diagram than a growth one. "How can marketing actually enable such a structured approach to scaling a business," he says, "to organizing a revenue engine?"

AI is automating marketing's most measurable work

Marketing narrowed itself over the past decade, Kimmel says, collapsing a wide remit into lead generation because that was the part with a clean number attached to it. The manual optimization work the function organized itself around turned out to be the work agents commoditize first. Once the data infrastructure is built, agentic flows can feed signals to the ad platforms and adjust messaging with less hands-on management than a performance team used to supply. What that automates is the executional layer, not the judgment around it: someone still has to set the goal, decide whether the creative is any good, and interrogate what the measurement is actually telling them.

The clearest sign of the shift is who does the manual optimizing. "You don't really need that much of a performance marketing manager," Kimmel says, describing platforms like Meta absorbing the campaign optimization that once occupied large performance teams. He sees a parallel shift through AI search, where the old lead-generation techniques matter less than getting credible brand content cited across third-party platforms. The part of marketing with the cleanest attribution is becoming the part platforms can run with the least day-to-day human involvement, which is exactly why it went first.

The bow tie reopens the revenue cycle

As agents absorb more of the acquisition work, the question becomes what the freed capacity is for, and the bow tie's right side is where Kimmel sees the largest opening. The stages after a customer commits, from onboarding through upsell, are where he wants automation pointed next. He is precise about aiming it at the accounts worth the spend rather than everything that moves. "You might not want to expand all your agentic effort on low-value traffic," he says, "because you're also paying for tokens." An agent watching account behavior can catch an expansion signal and trigger a bespoke offer tied to a customer milestone, the kind of ongoing engagement work human teams rarely had the bandwidth to run at scale. The right side of the journey, long a blind spot after the signature, gets the same stages and conversion rates that acquisition always had.

That is not an argument for a cheaper output. "I wouldn't use AI to make outputs cheaper. I would use it to scale outputs," Kimmel says. The point is to free people for the creative and strategic work headcount limits used to make impossible, and to do more at each touchpoint rather than the same amount for less. AI gives marketing the opening to move past lead generation toward what moves the needle more, including brand and retention, the wider remit the function narrowed away from when the clean number took over.

The org has to change shape to absorb it

The capacity only matters if the team is built to use it, and the teams built as lead-generation engines are the least ready for it. Those teams will have to reconfigure to do the full bow tie justice, because the manual work they optimized around is precisely the work agents and platforms now take on. The resistance, as Kimmel reads it, is less about discomfort with AI and more about teams defending the function they spent years narrowing into. He is frank that this is marketing's own long-running problem surfacing again. "Marketing obviously has that issue with justifying their contribution," he says, and he sees the current moment as the opening to answer it with structure rather than argument.

The reconfiguration he describes is a change of shape. Hierarchical teams heavy on execution layers give way to more strategic and creative resources, with an orchestration layer sitting atop the agentic work. In that model, the experienced operator becomes more valuable, not less, because the scarce skill is no longer running the execution but knowing where, when, and why to deploy it. The teams that make that move get the rest of the revenue cycle to work with. The ones that resist are left defending the performance layer, the one stretch of the journey platforms are increasingly capable of optimizing without them.