The standard sequence gets it backwards
The typical approach to AI transformation follows a familiar order: assess where you are, design where you want to be, build a roadmap to get there. It looks rigorous. It produces documents that look like strategy. And it regularly produces transformations that go sideways, because the assessment step almost never captures what it needs to capture.
What standard assessments miss
Most assessment tools measure what organizations say about themselves. Survey scores on AI readiness. Self-reported capability levels. Stated priorities from leadership interviews. What they miss is the gap between what people say and what they actually do. The cultural undercurrents that don't surface in official channels. The resistance that exists but hasn't yet found a voice. The managers who are nominally supportive and quietly undermining. I've seen organizations score highly on AI readiness assessments and then watch the transformation collapse in the first quarter of implementation, because the assessment described the organization people wished they worked in, not the one they actually worked in.
Reading the system rather than surveying it
What I mean by Signal is the ongoing capacity to read the organization accurately, not just at the start of a project, but continuously. It's the difference between running a survey and knowing how to listen. It means noticing that the head of finance hasn't said anything critical about the AI initiative in three months but also hasn't championed it once. It means understanding that the resistance in the marketing team traces back to a specific incident six months ago that nobody mentioned in the kickoff workshop. These things don't show up in readiness scores. They show up when you pay close enough attention.
Why structural design without signal fails
When you design structural changes, new governance models, new accountability frameworks, new collaboration protocols, without an accurate read of the actual culture, you design for an organization that doesn't exist. The structures are imposed on a different reality than the one they were built for. They create friction that wasn't anticipated, workarounds that weren't planned for, and eventually they get abandoned or quietly ignored. The design was right for the wrong organization.
The practical implication
Before you design anything structural, spend more time on Signal than feels comfortable. Talk to the people who aren't in the room. Ask the questions that might get awkward answers. Try to understand the gap between what the organization publicly commits to and what it actually rewards. Design for that organization. It takes longer. It's less tidy. And the transformations built on it actually hold.