Adoption Is Earned, Not Configured

I've led enough rollouts to notice the pattern in the failures before I noticed it in the successes. The failures all had a smooth launch. No complaints in the first week, glowing adoption metrics, a project team that moved on to the next initiative believing the hard part was over.

Then, months later, usage would quietly collapse. Not because the software broke. Because the silence in week one hadn't been agreement — it had been clinicians deciding, individually and without saying so, that raising concerns wasn't worth the friction. They complied instead of adopting, and compliance has a short half-life.

The rollouts that actually held up looked messier at the start. Loud pilot phases. Clinicians who pushed back hard, sometimes in front of the whole department, on decisions the project team thought were settled. It felt, in the moment, like resistance. In hindsight it was the only real signal we ever got about whether the system would survive contact with an actual shift.

What we changed, eventually, was structural: a mandatory period, before any network-wide rollout, where a working group of frontline clinicians had explicit authority to send a system back for changes — not just feedback that got logged and ignored, actual authority to block. It slowed every rollout down. It also meant that by the time something reached the full network, the people using it had already fought for the version that survived.

Vendors don't love this process. It's slower than a confident sales pitch prefers. But the alternative — a clean launch that quietly dies six months later — is slower still, and far more expensive, once you count what it costs to win back trust the second time.