The Maze: AI in marketing is usually sold as a productivity story. Faster briefs. Cheaper assets. More versions. BCG's survey of senior marketing executives points to a bigger change: accountability is moving. Agencies are expected to own less of both media and creative execution, while technology and internal teams own more. That is not just a procurement shift. It changes who controls data, workflow design, measurement, and the final business outcome.
Agencies lose the cleanest handoff in media. In the MMA-BCG survey, agency accountability for media falls from 42% today to 27% in two to three years, a 14 percentage-point drop. Creative falls too, from 35% to 23%, down 11 points. The old model was comfortable: brand teams set direction, agencies executed, and vendors supplied tools. AI makes that separation harder. When planning, buying, creative testing, product-page content, and measurement start feeding the same decision loop, the owner of the system becomes the owner of the result.
Technology becomes more than plumbing. Technology accountability rises from 7% to 16% in creative and from 10% to 18% in media. That is a 9-point increase in both workflows. The implication is sharp: platforms stop being passive software and start becoming operating layers. They shape the brief, produce variants, read performance signals, and recommend the next action. CMOs can still outsource work. They just cannot outsource the logic of how the machine learns. If the operating system sits outside the company, so does part of the marketing brain.
Internal teams gain leverage, but also pressure. Marketing's internal accountability rises from 48% to 50% in creative and from 45% to 51% in media. That sounds modest. It is not. Internal teams are already the largest accountability holder, and AI gives them more direct control over workflows that used to require handoffs. The problem: control without a model becomes chaos. BCG says only 15% of surveyed companies have AI initiatives operating cross-functionally at scale for enterprise value. Everyone wants leverage. Few have the wiring.
Agentic AI turns process design into a CMO problem. BCG expects agentic AI to handle more than one-fifth of total marketing workload within two to three years. That pushes marketing leaders past the toy phase. A prompt library will not decide who approves creative, how value is shared with agencies, which data feeds the system, or how finance judges impact. Those are operating-model decisions. The winners will not be the teams with the most AI pilots. They will be the teams that know where human judgment belongs and where the machine should take over repeatable decisions.
Why it matters: Agencies are not dead. But the agency-client contract is being rewritten. The next fight is not over cheaper production. It is over outcome ownership. Retail media, ecommerce content, search, social, and programmatic are already signal-heavy workflows. AI compresses the distance between idea, placement, measurement, and iteration. That rewards CMOs who build internal muscles, make technology accountable, and pay partners for value rather than activity. The uncomfortable truth: automation is easy to buy. Accountability is the expensive part.
Sources: BCG | LinkedIn post
