BCG's 2026 CMO survey and the MMA-BCG study show AI agents moving from pilots to real marketing work. What it means for brands and the agencies that serve them.
Most marketing leaders agree AI will reshape their function. The more useful question in 2026 is who is actually rebuilding how work gets done. Two recent studies from Boston Consulting Group give the clearest picture so far, and they contain a pointed message for agencies.
What the research found
BCG's 2026 CMO survey spoke to 300 chief marketing officers worldwide, across B2B and B2C, and added interviews with 50 of them. Ad Age's coverage of the study reports that roughly a third of CMOs, about 32%, count themselves leaders in agentic marketing, meaning AI agents that plan and execute campaign decisions rather than simply draft copy. The same coverage makes clear that many of those leaders are still learning how to scale the new processes.
A second study, run by the Marketing + Media Alliance with BCG between May and July 2026, asked about 60 global CMOs how AI is changing the marketing operating model. The headline gap is between ambition and execution. CMOs rated their AI vision at about six out of ten, while technology, governance, talent, agency partnerships and workflows all scored around five.
Three other findings from the MMA-BCG work stand out:
- Agentic AI accounts for about 7% of marketing work today, and respondents expect that to reach roughly 27% within two to three years.
- Close to half of AI investment in marketing, 48%, now comes from outside the marketing budget, which tells you AI is being funded as business transformation rather than a campaign tool.
- Around three quarters of CMOs say their creative agencies are not ready to scale AI. For media agencies the figure is 73%.
Why the agency finding matters
That last number is the one every agency should read twice. Clients are not asking whether their partners use AI. They are saying most partners cannot yet run it at scale: with clean data, governance, measurement and people who know how to supervise agents.
It also explains why the big holding companies are spending so heavily on platforms. Publicis has described AI built into planning, targeting and activation, and WPP has pushed its Open platform. But platform spend alone does not close the readiness gap. Clients still need a partner who can connect agents to their own data, set guardrails, and prove the result in revenue terms.
What good looks like in practice
Agentic marketing is not a chatbot added to an existing process. The pattern we see working has four parts.
Start from a revenue question, not a tool. Pick one measurable outcome, such as cost per qualified lead or repeat purchase rate, and ask which decisions in that chain an agent could make or speed up.
Fix the data layer first. Agents make many small decisions, and they are only as good as the signals behind them. For most Indian brands that means server-side tracking, a clean CRM and consistent conversion events before any automation goes live.
Keep humans on strategy and judgement. The MMA-BCG finding that agentic AI pushes human work upstream matches our experience. People set positioning, creative direction and risk limits. Agents handle the volume: variant testing, reporting, lead response, budget shifts.
Measure against a control. Every automated system should run against a holdout or a clear baseline. If you cannot tell what the agent changed, you cannot trust the improvement.
What this means for Indian brands
India's marketing teams are lean and move fast, which is an advantage here. A brand with a tidy funnel and one or two well-chosen workflows can get agentic systems into production faster than a global enterprise with fifteen approval layers. The risk is the opposite one: buying tools before the data and measurement exist, then wondering why results do not move.
If you are choosing a marketing partner this year, ask four things. Which of your decisions will be automated, and which stay human? How will you measure the difference against a control? What data do you need from us? And who is accountable when an agent gets it wrong?
Our AI marketing and agents service is built around those questions. A partner who answers those clearly is ready for the next three years. One who leads with a tool list is not.