Every few years, marketing goes through a version of the same reckoning. Cloud. Digital. ABM. Something shifts, everyone runs toward it, and then at some point someone in the room asks: “What are we actually getting for this?” We’re at that moment with AI. And this time, the question is harder to dodge.
Earlier this year, I participated in Callan Consulting’s State of AI in Technology Marketing 2026 report: 19 marketing professionals, good companies, serious leaders. We all basically said the same thing: AI isn’t experimental anymore. It’s how the work gets done. Then the researchers asked us to quantify it.
And none of us could. Not really. We had feelings about it, directional estimates, anecdotes. I told them we’ve seen campaign performance increases of 4x to 100x depending on audience and channel. That’s real. But could I isolate AI’s contribution from everything else happening in the business at the same time? No.
Two-thirds of us rated AI’s impact as “strong” or “very strong.” Not one of us could prove it on a spreadsheet. That gap is not an AI problem. It’s a data problem, and it’s one marketing needs to stop outsourcing to someone else to solve.
The Measurement Problem Is Upstream
Here’s what I’ve come to believe: you can’t measure AI’s contribution when your customer data lives in a dozen places that don’t talk to each other. You can’t trust the attribution model when the data feeding it isn’t governed. In this situation, the value of AI shows up as a feeling: we move faster, we do more with less. But the foundation underneath isn’t built to make it visible.
Built in, not bolted on. That’s how we talk about data infrastructure at NetApp, and it’s the frame that keeps proving itself out. When AI gets layered onto a broken data environment, the outputs might look impressive in isolation. But you can’t measure them with any rigor, and you can’t optimize them over time. The ROI question doesn’t get answered at the AI layer. It gets answered two levels down.
Overreliance Has a Second Problem Nobody’s Talking About
The Callan Consulting study also raises the overreliance risk: marketers anchoring to AI’s output instead of doing the harder creative thinking first. That is valid. But there’s a version of this that concerns me more than content quality.
Marketers are trusting AI’s conclusions about what customers want, what campaigns are doing, and what competitors are up to. We’re not asking where those conclusions come from. If the underlying data is stale or inconsistent, the AI sounds confident anyway. It always sounds confident. And we’re moving fast enough that we’re not stopping to check.
When I educate my team on AI, the conversation isn’t about prompting. It’s about judgment. Knowing when to trust the output and when to push back. That’s the skill that actually matters.
GEO Raises the Stakes on Content Quality
One of the bigger shifts captured in the study is how fast Generative Engine Optimization (GEO) has moved from future-state to right now. Half the CMOs involved in the study have formal GEO efforts underway.
GEO has higher data quality requirements than SEO ever did. Search engines crawl what you publish. LLMs synthesize from it. That means your thought leadership, your technical content, and your documentation all have to be accurate and consistent enough to survive that synthesis. If noise goes in, noise comes out. And unlike a bad search ranking, you might not catch it for a while.
Your organization’s thought leadership strategy must be built around this. LLMs amplify what’s already in the market. If you don’t have a specific strategy focused on what to publish for humans vs. bots for your messaging and defining your business, someone else will, and it will likely be your competitors. If you’ve been putting out a strong, consistent signal, that amplification works in your favor. If you haven’t, you’re about to find out.
This Is Marketing’s Conversation to Lead
The study introduces a useful frame: “Born in AI” companies, meaning organizations founded after generative AI emerged, where using it is just assumed. No tiger teams. No change management. It’s wired in.
Those companies didn’t bolt AI onto existing infrastructure. They built around it. Everyone else is dealing with a foundation gap. They are retrofitting, and the distance is widening.
Marketing should be the function pushing hardest to close the gap. Not because we own the infrastructure, but because we need it most. Personalization at scale, real-time optimization, content that holds up when an LLM synthesizes it: none of that works if the plumbing isn’t built for it. This puts the brand and the funnel at risk.
Someone in the enterprise is going to own the AI infrastructure conversation. If it’s not marketing, we can’t be surprised when what gets built doesn’t serve what we need.
