I've sat with that number for a few days. It doesn't match what LinkedIn is telling you.
Scroll your feed. Every marketing team looks like it quietly became a self-driving revenue machine. Adoption is basically universal now. Confidence is not.
If you've looked at your own dashboard lately, you've probably felt it. Somewhere in the last 18 months, your team started using AI for something. Content. Campaigns. Maybe a chatbot handling inbound.
Ask yourself if any of it moved a number that matters. You'll probably pause before answering.
That pause is the whole story.
Here's what the research says.
Gartner surveyed 401 CMOs earlier this year. 70% said becoming an AI leader is critical for 2026. Only 30% said their organization is actually ready to scale it.
McKinsey found something sharper. Fewer than one in five organizations (19%) track KPIs specifically built for generative AI. Most of us can't say whether the tool worked, because we never built a way to check.
Jasper surveyed 1,400 marketers this year. Only 41% could demonstrate AI ROI when asked directly.
Four different research houses. Four different survey groups. All pointing at the same gap.
Adoption stopped being the story a while ago. Measurement is the story now.
The four numbers
I've watched this pattern up close. It isn't complicated once you see it.
Most of us adopted the tool before we built the scorecard.
The tool arrived fast. Someone in leadership asked for an AI strategy. Content started shipping faster. Campaigns multiplied. Few teams paused to define what "working" would look like before switching it on.
So we default to counting what's easy. Posts published. Campaigns launched. Emails sent.
None of that is revenue. It's just motion.
I call this the volume trap: mistaking activity for outcome, because activity is the only thing we built a dashboard for.
You can't out-produce a measurement problem.
Three moves, in order.
None of them need a new tool.
-
Pick one metric before you scale anything new
Not "engagement." Not "content velocity." Pick the single number tied to pipeline or revenue that this specific initiative is supposed to move. Write it down before the tool goes live, not after.
-
Set a baseline first
You can't prove lift without knowing where you started. It's tempting to skip because it feels slow. It's actually the fastest step in the whole process. It just has to happen first.
-
Treat "we don't know if this worked" as a failed answer
If no one on the team can trace an AI initiative to a specific outcome within 30 days, it probably wasn't designed to be measured. That's worth fixing before it scales.
None of this is complicated. It's just not fast, and fast felt like progress.
The 95% adoption number will keep climbing. That was never going to be the hard part.
The 39% only moves if we decide what to measure before the next tool arrives, not after.
That's the real story under the AI headlines. Not whether the tools work, but whether we decided what "working" means.
Navneet Kumar
Building Neolism in the open.