Teams ship faster than ever now, and product adoption still fails at roughly the same rate it always did. People don’t refuse to use something that works because it’s broken. They refuse for reasons that were never functional in the first place, and that refusal has never clearly been anyone’s job to own. It is mine now, by default, because nobody else’s job description covers it either.
Delivery used to be the whole job
For most of my career, delivery absorbed the CPTO role almost completely: the roadmap, the sequencing, the trade-offs against a limited build capacity. Product and growth teams spent a decade building sophisticated instrumentation to get people to use what got shipped: activation funnels, onboarding flows, engagement and retention metrics. All of it assumed the hard part was getting the thing built and in front of people.
That assumption is breaking. AI has collapsed the cost and time of building almost anything, and once delivery stops eating the whole job, whatever was always the real job becomes visible. The evidence says that’s adoption, not ambition. Tara Seshan’s framing of steering versus rowing is about a different, earlier bottleneck: the will to decide what’s worth building at all. This is about what happens after that decision is right and the thing works and people still won’t use it. Steering picks the destination. Adoption is a separate craft, and it starts once the boat has already arrived.
Refusal is rarely functional
Jobs-to-be-Done theory has always said a job has three dimensions, not one. The Christensen Institute defines them plainly: functional, social, and emotional, and the theory exposes the functional, social, and emotional dimensions that explain why people make the choices they do. The social dimension is how others view someone using it; the emotional dimension is how the thing makes them feel. Bob Moesta, who co-developed the framework with Christensen, built his career on the same point: people aren’t hiring a product, they’re hiring progress against a job, and progress is never only mechanical.
Most product organisations still design and measure only the functional dimension. Someone completes onboarding, hits every activation event, has no complaint about whether the thing works, and quietly never makes it part of how they actually work. The funnel was never built to measure this cost: does using this make someone feel less competent, less senior, less themselves. Does it signal something to the people they work with that they’d rather not signal. Nobody put those questions on a dashboard, because dashboards measure what’s easy to instrument, not what’s true.
The instrumentation is watching the wrong dimension
The obvious objection is that adoption is already one of the most measured things in the industry. Awareness, onboarding, engagement, retention, advocacy: the whole funnel is instrumented, and activation-versus-adoption is a distinction growth teams argue about constantly. This isn’t a neglected problem.
It’s a precisely measured version of the wrong problem. Those instruments track whether someone started, whether they returned, whether they crossed an activation line. None of them were built to capture an emotional cost or a social one, because both are quiet and neither shows up as an event you can fire. A feature with healthy funnel numbers that still never gets truly adopted isn’t a counterexample to this. It’s the clearest evidence for it: the dashboard is green and the thing still isn’t real, because green was never the right colour for what actually mattered.
I’ve written before about a narrower version of this, in AI adoption specifically: the people who hang back longest usually aren’t protecting the old tool, they’re protecting the basis on which they understand their own contribution. That’s an identity cost, one instance of the emotional dimension. The pattern is bigger than AI tools. It’s every adoption problem that gets diagnosed as training, or communication, or change management, when the actual cause is that nobody addressed what the change costs someone socially or emotionally, only what it does functionally.
Affordances carry more than looks
Don Norman gave design the word affordance in 1988, in The Design of Everyday Things, borrowed from the perceptual psychologist James Gibson. A well-designed door handle tells you, without instruction, whether to push or pull. Norman is now re-examining that same framework against AI, and the re-examination matters here for a specific reason: an affordance was never only about what a button looks like. It’s about whether the thing in front of someone tells them what it costs to use it and what it’s for. A rollout that explains what a tool does but never addresses what adopting it signals is a door handle that shows you the mechanism and hides the room behind it.
This is also why the oft-repeated claim that seventy per cent of transformations fail is worth retiring. It isn’t cleanly traceable to the McKinsey research it gets attributed to. What that research actually shows is closer to only sixteen per cent of transformations both improving and sustaining performance against a strict bar. The real number is more damning than the myth, not less, because it isn’t describing bad execution. Most of those transformations were built correctly and delivered on schedule. What they didn’t do was get chosen, every day, by the people who were supposed to live inside them.
The only letter in RACI that was ever load-bearing
RACI charts assign an Informed role to almost everyone on a rollout, and almost everyone treats it as the cheapest one to fill. Tell people, tick the box, move on. I think it’s the opposite: being genuinely informed, in the sense of understanding why and feeling some ownership over the answer, is the actual mechanism by which the emotional and social costs of adopting something get addressed at all. Broadcasting a change is not the same act as designing how someone comes to feel it was theirs.
If you’re a CPTO, or you hold that job under a different title, and adoption keeps stalling despite a product that works and a funnel that looks healthy, the question worth asking has changed. Not did we build it right. Who actually feels informed and has ownership here, and what is adopting this costing them emotionally or socially. Design for that answer with the same rigour you’d demand of any affordance. That question is the job now, not a side quest from it.
The last mile has always been the letter I.
Leave a Reply