The blog

Industry

96% of CMOs say AI is transforming marketing. The missing hires are two owners

The AI gap in marketing is an ownership gap. BCG's 2026 CMO survey, published June 15, found that 96% of CMOs say AI is driving end-to-end transformation of their function while only about a third have done the work. BCG's own summary: the gap between what CMOs claim and what they have built is the story of 2026.

Budget does not close that gap. Two roles do: an AI product owner who runs agents like products, and an AI governance owner who can stop them. Sprinklr's practitioner survey, reported by PPC Land, shows how rare that structure is: just 16.9% of respondents had a named owner for AI business outcomes.

The claim-versus-build gap in numbers

BCG sorts CMOs into three tiers: 32% Leaders, 26% Followers and 42% At-Risk. Only about 8% are starting to run campaigns in which multiple agents operate autonomously, and 42% still use generative AI only to assist humans with discrete tasks. Meanwhile 43% report marketing AI investments above $15 million, up from 28% a year earlier, and 94% say CEO expectations of marketing have risen significantly over two years.

Those are large-enterprise figures, but the pattern travels down-market. Spending is ahead of structure. Gartner's 2026 CMO Spend Survey reached the same conclusion from the budget side: CMOs allocate 15.3% of marketing budgets to AI, yet only 30% are ready to scale it. Money is not the constraint in either dataset.

What BCG says to build

BCG's recommendations read like an org chart rather than a tool list. It calls for AI product owners, AI governance owners, AEO and GEO teams, and expanded marketing science and AI engineering capability. It wants smaller, AI-enabled teams organized around business objectives or customer segments rather than channels. About 80% of CMOs in its survey are making significant upskilling investments and 75% are rolling out agile pods to address human bottlenecks.

Two of those roles do the work that decides whether an agent program survives. The rest support them.

The AI product owner

This person treats each agent as a product with a user, a metric and a backlog. They define the outcome the agent is accountable for, set the baseline, review performance on a schedule and decide what to improve next. They answer the question Sprinklr's survey found most teams cannot: what, precisely, did this agent deliver? In that survey only 27.2% of practitioners could quantify a result precisely, even though 91.8% of those running AI in live operation said it met or exceeded expectations.

The AI governance owner

This person owns the rules agents run under: what data they can touch, which actions need human approval, how brand and legal checks apply, what the spend limits are and what happens when something goes wrong. Crucially, they have the authority to pause an agent. The product owner is incentivized to ship and improve; the governance owner is incentivized to limit damage. Separating the two is the point, because the person who built the agent is the person least likely to switch it off.

What this means below enterprise scale

A company with a five-person marketing team does not need two new hires. It needs two named hats. The head of marketing or a senior operator can be the product owner. The governance hat should sit with someone who is not accountable for the agent's growth numbers, such as an operations lead, a finance partner or the founder. Where nobody fits, an outside partner can hold one of the hats on a fixed scope. What cannot work is leaving both blank, which is what the survey data suggests most teams do.

There is also a leadership angle. BCG finds about half of CMOs say marketing leads AI investment decisions, while 72% of CEOs see themselves as the main AI decision maker. When decision rights are contested, ownership at the working level is the first thing to go missing. Naming the two roles settles who decides, at least for marketing's agents.

What to do about it

  1. Name a product owner for every agent in production. One person per agent, written into your agent inventory.
  2. Name a governance owner who is not the product owner, and give them written authority to pause any agent.
  3. Give each agent one outcome metric and a baseline before the next budget conversation.
  4. Write the approval and spend rules down in one page that the governance owner maintains.
  5. Review both roles quarterly, and spend the next AI dollar on whichever one is still unstaffed before adding another tool.

Want this working in your business?

Book a discovery call