Adoption was never the hard part
Every business has lived through a version of this before AI arrived. A company makes a serious technology investment. It spends years in gating and governance and writes a business case built on lower cost, greater agility and less rework. Then the project closes, the technology goes live, and nobody is left whose job it is to extract the value the case promised. The benefits that justified the spend sit badly under-used, quarter after quarter, and the organisation moves on.
AI inherits that pattern and removes the one thing that used to slow it down, which is friction. The tools are cheap to acquire, so procurement barely notices. IT has a say on security and data sovereignty, then steps back, and the tools spread through a team in weeks rather than years. Individuals get faster at their own jobs almost immediately.
Whether the business is any better off is a much harder case to make.
What the numbers actually show
Nearly 60% of marketers now report using AI multiple times a week. Only 28% say their company is pursuing a fundamental rewiring of its marketing teams and workflows (McKinsey Global Marketer Survey, 521 respondents, published June 2026). The adoption curve and the redesign curve have come apart, and most functions sit entirely on the first one.
Adoption and redesign have come apart
Share of marketers
That is the trap. Widespread AI use, with the operating model and the org chart left exactly as they were, produces outcomes that look much like the ones before it. It is natural to conclude that AI projects underdeliver, or that the benefits were overstated from the start. We read it differently: the capability was never installed, so there was no value to capture.
New capability has to be named, owned and resourced inside the operating model, not left to accumulate informally in whoever happens to be curious.
The gap at the top
The same pattern shows up at board level. In EY-Parthenon’s October 2026 CEO Outlook, 34% of the 100 UK CEOs surveyed named legacy technology and operating models as the biggest obstacle to converting productivity gains into measurable business value. EY-Parthenon’s CEO Outlook also found almost half of CEOs (47%) say their organisation is not developing skills quickly enough to keep pace with AI innovation. Everyone has access to the same models, so the difference is what has been built around them.
Marketing’s own leaders show the gap most clearly. Gartner’s CMO survey found 65% of CMOs expect AI to dramatically change their role within two years, yet only 32% say significant changes are needed to the CMO profile and skill set (Gartner, 402 senior marketing leaders in North America and Europe, published 23 February 2026). Most expect the change, and far fewer have planned for it.
What the redesign actually requires
Capturing the value isn’t a training problem or a budget problem. It is a structural one. New capability has to be named, owned and resourced inside the operating model, not left to accumulate informally in whoever happens to be curious.
A simple diagnostic exposes it. How many marketing functions have an AI Orchestration Lead, the named role that directs and quality-assures the agentic layer, rather than leaving it to whoever on the team got there first? And how many have recruited a CMO against the operating model AI now makes possible, rather than against the job description written for the one it replaced?
We ran into this ourselves. We have each led marketing organisations of more than 100 people and built new functions that, between them, removed roughly £15m in avoidable spend, which was reinvested in market-facing activity. We also remodelled client engagement to generate opportunities qualified to mid-funnel stage without additional business development spend. Recent as those achievements are, both were built around the operating model of their era. Neither would survive contact with what AI can now do. Redesigning the model properly is the job. We built Crew & Engine to do it the second time round.
Adoption isn’t transformation. The operating model is the real bottleneck, not the tool.
We’ve written elsewhere about marketing’s habit of bolting new technology onto old ways of working. This is the evidence for why it keeps happening, and what it costs the functions that let it.