You Don't Have an AI Problem, You Have a Data Problem
Every marketing leader I talk to has an AI initiative. Fewer than half have a data foundation that can support it.
Adobe's 2026 AI and Digital Trends report found 96% of B2B marketers are using AI, but only 44% say their data is clean enough to run it on.1 That combination tells us that most of the industry is building on sand.
AI doesn't produce a bad campaign because of a lazy prompt, it produces a bad campaign because the CRM has three versions of the same contact, the funnel numbers don't match what sales sees, and the forecast hasn't been reconciled since Q2. Validity's latest CRM survey found only 26% of marketers rate their CRM data as more than three-quarters complete and accurate.2 Feed a tool half of the picture and it will still hand back a confident, albeit wrong, answer.
What does that look like in practice? Here's an example: last month, one of our client’s in-house marketers built a campaign brief, got sign-off from their VP, and felt good about it. Then, their CEO ran it through Claude and came back with nine pages of notes. But the notes recommended the wrong audience and messaging – an approach we all had ruled out weeks earlier. It didn’t line up with what the team already knew, because Claude didn't know the context and there wasn't data to draw from. They then spent two days figuring out how to respond to feedback from a tool that had none of the history they had.
Your marketing managers are now tasked with catching these kinds of issues. That's the job AI was supposed to eliminate, and instead it's causing more work. We see our clients struggle with it frequently. Their teams are over-generating instead of being selective, and in lieu of solid, data-backed strategy, they are churning out wildly over-engineered briefs that nobody asked for, and 15-slide decks that really just needed one clean visual.
Someone has to read all of this content, flag what's wrong, and explain what "good" looks like again. That's not time saved. That's a new kind of managing, and it's wearing your managers down.
We get it. No one has perfect data. Not the enterprise teams with a full RevOps function and definitely not the startup and SME teams running marketing with three people and a stack of tools nobody fully integrated. There's always going to be a "make it work" layer, gaps in attribution, contacts that never made it into the CRM, and forecasting data that's more art than science. That's normal. The mistake is treating those gaps as a reason to wait on AI, ignoring them and hoping the tool fills them in, or worst of all, not knowing where they exist. Your AI strategy doesn't need perfect data to be useful today. But it does need someone at the helm who knows how to work around the missing pieces now, while building the roadmap to close them for good.
You don’t have to pump the brakes on AI. But acquiring a flashy AI tech stack and calling the job done won’t help you either. The leaders who win this next stretch will be the ones who treat data hygiene like a permanent line item instead of a cleanup project that gets cut when resources tighten.
If your team is drowning in "just one more revision" on every AI-assisted deliverable, you don’t have a tool problem. You have a data problem, and it's fixable! But it takes someone intentionally paying attention to it.
What we see working best now is being specific. Define what AI is actually good for in your organization, what data you have to back that use case, and how that intel is going to evolve and scale your marketing operations. That's your roadmap.
What's your reality look like? Reply and tell us where your gaps are, we'll tell you what we'd tackle first.
Source:
