Risograph print inspired by Agnieszka, Warsaw
94 percent of what?
The headline number on our org’s AI adoption dashboard reads 94 percent weekly active utilization across Product Management, Product Design, and Program Management. It’s a real number. It’s also a number that, on its own, tells you almost nothing about what’s actually happening with AI in our company.
I took on the work of producing that dashboard a few months ago. It wasn’t formally assigned to me. AI tool adoption measurement across three disciplines sits in a gap: too cross-functional for any single discipline to own, too operationally specific for the CPO, too design-adjacent for the CTO. I picked it up because nobody else was going to, and because the connective-tissue work around AI (the UX patterns we’re building for the product, the eval definitions we’re aligning with engineering, the agentic playbook in progress) all sits closer to design than anywhere else.
I want to talk about the work, not because I think dashboards are the point, but because the act of building one taught me something about where the real story lives.
The 94 percent number is technically accurate and operationally misleading. It says that of the 68 named users across the three disciplines, 64 are weekly active across at least one of the three tools we provision. That’s a real signal. It tells you the floor is high. It tells you that experimentation is no longer the right framing for what’s happening.
The three patterns underneath the headline are where the real story is.
The first pattern is champion-led depth. 62 percent of the credits on our primary design tool sit with the top five users. That concentration is not a problem. It’s a Q3 harvest. Those five users are the people building the workflows the rest of the team will inherit. The leadership job is making sure those workflows leave the champions’ heads and become shared playbooks before the champions become a single point of failure or move on to a different company.
The second pattern is broad and even. 31 percent of usage on our second tool sits with the top five. The widest footprint, the least concentration. That tool is where org-wide rituals can anchor. Status reports, meeting notes, light synthesis. It’s the tool that gets traction with the people who aren’t champions, which makes it the tool to standardize on for the practices everyone needs to share.
The third pattern is the one that taught me the most about why this work matters. There are four active users on our third tool, a heavyweight enterprise-grade option, and the top two of those four hold 94 percent of the credits. The same 94 percent headline that frames adoption as universal also describes the concentration of the most powerful tool we offer. The people who would benefit most from access to it don’t have it, because the seat allocation is constrained by cost. They either slow down or, as a peer of mine flagged in a recent conversation, find their way to it on personal accounts with company information, which is a different kind of cost that doesn’t appear in any budget line.
That third pattern is the article in miniature. The headline says everyone has it. The seat data says four people do. The same metric tells two opposite stories depending on whether you read it as adoption or access. If nobody owns the measurement work, the headline wins by default and the gap gets bigger.
The reason I took the work is that the gap was hiding inside the headline, and the gap was going to land somewhere whether or not we tracked it. Without the dashboard, the people slowed down by the access disparity would have been invisible. The leadership move was to absorb the measurement work into a scope that wasn’t formally mine, so the patterns could become legible.
The horizon framing I use to keep this honest is three waves of roughly six months each. Foundation through end of Q3: standardize the tools, the workflows, the metrics. Move from “people are experimenting” to “we know what AI saves us, by job, by discipline.” Scale and embed through Q4 and Q1: operationalize the workflows that proved out in the foundation phase. The design system grows into a product development system with code contracts that AI can read. Collapsed triad in 2027 and beyond: the sharp lines between PM, design, and engineering blur further, full-stack product owners take on work that used to require formal handoffs.
The horizons keep me honest about what’s real this quarter versus what’s direction-setting for next year. Without them, every monthly metric becomes a debate about whether 2027 is happening yet. With them, the dashboard is anchored to a story about where we are in a multi-year arc.
The pattern I want to leave you with is the leadership move underneath all of it. If you wait for AI transformation measurement to be formally assigned to someone, the work won’t happen. The cross-functional surface area is too wide for any single discipline to claim and too specific for any executive to pick up directly. Design leadership is positioned to absorb it because design already operates across disciplines, already deals with ambiguity, and already cares about the user-facing implications of what the tools do.
Taking it on means the patterns become legible. The access gap becomes visible. The horizon framing becomes the operating rhythm. The board sees a real story instead of a headline number.
That’s the displacement frame in its constructive form. Somebody absorbs the work. Choosing to absorb it on purpose, with eyes open, is different from having it land on you. The first version is leadership. The second version is what happens when leadership defaults.