People ask me sometimes what the hardest part of moving an entire hospital’s IT infrastructure is.
The honest answer is: not the 12-hour cutover window. Not the war room. Not the 3 a.m. decision calls.
The hardest part is the 11 months of work you have to do correctly before any of that starts.
The Setup
When I joined the Ardent Health / SCL Health divestiture program, the ask was straightforward in concept and extraordinarily complex in execution: move St. Francis Hospital and all 13 of its associated clinics from one health system to another. Completely. Every system, every integration, every clinical workflow — on a defined deadline, with a hard ceiling on how long the hospital could operate in a split state between two owners.
There were two Epic instances involved. Two different versions of interface engines. Clinical teams that couldn’t go without system access. Financial systems that needed to stay accurate during transition. Security and compliance frameworks from two organizations that had never had to agree on anything before.
And everyone — both sides — had a different definition of the word “ready.”
The Dependency Map Was the Program
Before we touched a single system, I spent weeks building a dependency map that most people would have considered excessive. Every integration. Every vendor relationship. Every team that had to be available, in what sequence, with what authority to make decisions.
Not because I enjoy documentation for its own sake — I don’t. Because in a cutover of that complexity, what you don’t know in advance will find you at the worst possible moment.
We found a wireless infrastructure issue in one of the clinic wings during rehearsal. We found a vendor who had assumed someone else was managing a critical handoff. We found an Epic integration that behaved differently across the two engine versions in a way that nobody had flagged because nobody had thought to test it explicitly.
We found all of those things before the night of the migration. That’s why the migration worked.
The Night Of
The cutover itself was — and I say this with appropriate humility — about as clean as an enterprise hospital migration can be. Twelve hours of planned downtime. Both sides of the organization in the war room. Clear command structure. Pre-authorized decision makers who knew exactly what their rollback triggers were.
Things went sideways in a few places. They always do. But because we had rehearsed the hard parts, the sideway things were contained sideway things, not cascading sideway things. There’s a meaningful difference.
By the time the sun came up, the hospital was live on the new system. Clinicians had access. Billing was running. The 13 clinics were connected.
What I Actually Learned
Every complex cutover I’ve run since then has followed the same principle: the quality of the event is determined by the quality of the preparation, not the competence of the people in the room on the night.
You can have the best engineers in the world in that war room. If the rollback procedure hasn’t been tested, if the decision authorities aren’t established in advance, if the dependency map has a gap that nobody found until production exposed it — the competence in the room doesn’t save you.
Preparation saves you. Structure saves you. The willingness to spend three months being rigorous about things that might not matter — because you don’t know which ones won’t matter until it’s too late — is what makes the hard nights survivable.
The quality of the event is determined by the quality of the preparation.
The 12-hour window was the easy part. Everything before it was the job.