90% adopted AI. 30% governed it. We need to talk

There are two conversations happening about AI in marketing right now.

The public one happens at conferences, on LinkedIn, in webinars, in vendor presentations. It's optimistic, polished, and full of case studies where everything worked. The message is consistent: AI is transforming marketing, the early adopters are winning, and everyone else needs to catch up.

The private one happens over coffee. After the session ends. In the group chat that doesn't include the vendors. Between peers who trust each other enough to say what's actually going on.

That conversation sounds nothing like the first one.

We know because we asked. Our 2026 AI Benchmark report surveyed marketing operations professionals across North America, EMEA, and beyond. The findings confirmed what the private conversations have been saying for months: AI adoption is accelerating and the operational foundations underneath it aren't keeping up.

Some of the numbers are striking. Others are uncomfortable. A few should concern anyone responsible for how AI operates inside their marketing stack.

We're not going to walk through the full findings here. That conversation is happening in a room, not in an article. But we'll share enough to explain why we think this conversation needs to happen differently.

What the benchmark revealed

Nine in ten organizations expect to expand their use of AI materially over the next 12 months. That alone tells you the direction is set. AI in Marketing Operations isn't an experiment anymore. It's becoming part of the operating model.

But here's where it gets interesting.

Only around three in ten have an implemented written AI policy specifically for Marketing Operations. Roughly eight in ten have no clear owner when AI causes harm. And when we asked about data sent to AI tools, the majority either don't know how it's protected or consider the question not applicable.

These aren't numbers from organizations that haven't adopted AI. These are numbers from organizations that are already using it across content, scoring, segmentation, routing, and journey orchestration. The adoption happened. The infrastructure around it didn't.

We found that data quality was cited as the number one blocker to safe AI scaling, ahead of legal uncertainty, budget constraints, and skills gaps. Seven in ten describe key-field completeness across their core records as mixed. The same percentage say duplicate or identity issues affect their reporting at least sometimes.

And when we asked whether teams could trace why an individual received a particular message, from data through logic to send, only four in ten said they could always do it.

These findings raise questions that don't have simple answers.

The questions behind the numbers

The numbers tell you where the industry is. They don't tell you what to do about it.

That's because the right response depends on context. Your organization's risk tolerance. Your data maturity. Your team's capability. Your regulatory exposure. Your leadership's expectations. The same benchmark finding means something different for a 500-person SaaS company and a 20,000-person healthcare enterprise.

The questions that follow from the findings are judgement questions, not knowledge questions. They can't be answered by a report or a webinar. They need a conversation between people who are living the same reality and are willing to share what they've learned.

Three of them are on our minds right now.

Are we moving faster with AI than we're becoming ready for it? The benchmark shows a clear gap between adoption speed and governance maturity. But the pressure to keep moving is real, and telling leadership "we need to slow down" isn't a conversation most MOPs leaders want to have. The question is how to build readiness without losing momentum, and whether that's even possible for teams that are already in production with AI features they haven't fully governed.

What's really holding back AI in Marketing Operations? The benchmark says data quality. But data quality has been a problem for decades. What's different now is that AI amplifies whatever state the data is in, good or bad. The question is whether teams should fix the data before expanding AI, or whether AI itself can help fix the data, and how you decide which approach fits your situation.

How much control do we need before we're comfortable giving AI more responsibility? AI agents are arriving. They can act, not just recommend. The benchmark shows that most teams don't yet have the traceability, accountability, or rollback capability to support autonomous AI. But waiting until everything is perfect means waiting indefinitely. The question is where the line sits between "ready enough" and "not ready," and how other organizations are drawing it.

These aren't questions we can answer in an article. They're questions that deserve a room.

Why this conversation needs a different format

The standard formats for industry conversations about AI aren't designed for honesty. Conferences reward polish. Webinars reward certainty. LinkedIn rewards confidence. None of them reward the kind of candour that produces useful learning.

The marketing leader who admits in a room of 12 peers that their AI governance is a single paragraph in a planning document nobody references is being honest in a way that produces value for everyone present. The same admission on a conference stage would be career risk.

The leader who says "we activated predictive scoring eight months ago and I genuinely don't know whether it's improved anything" is sharing the kind of insight that helps ten other people in the room calibrate their own experience. That sentence doesn't survive the transition to a public format.

The most useful conversations about AI in marketing operations right now are the ones where people say what's actually happening rather than what's supposed to be happening. Those conversations need a room that's small enough for honesty, senior enough for relevance, and private enough for candour.

MPE: the room where this conversation happens

We organized the Marketing Playbook Exchange because the most important conversation about AI in marketing operations isn't happening publicly, and most leaders don't have a structured opportunity to have it privately.

MPE is a closed-door roundtable on Wednesday 16 September at The Ivy Tower Bridge in London, 9:30 to 11:30. Breakfast, coffee, and a conversation grounded in what the 2026 AI Benchmark report actually found.

Not the summary. Not the highlights. The full findings and what they mean when you sit with them honestly.

From there, the conversation belongs to the room. Nothing said inside it is reported outside it.

Twelve people. Senior marketing, technology, and strategy leaders. Small enough that everyone contributes. Private enough that nobody performs. Two hours of the conversation that matters most right now, with the people best positioned to have it.