I hit a breaking point after adding locations, not because anything “broke,” but because scheduling behavior stopped being consistent enough to rely on.
If you’ve ever synced appointment data across multiple dental locations, you’ve probably noticed the API isn’t wrong, the data isn’t missing.
The problem is that “appointment” doesn’t mean the same thing everywhere once you cross locations.
At one location, a reschedule mutates the original record.
At another, it creates a new record and retires the old. Sometimes behavior changes based on timing or role or really nothing. Offices can have multiple statuses that all mean an appointment is cancelled.
From the PMS perspective, this is valid operational behavior. From an integration perspective, it means appointment is no longer a stable object.
I tried to normalize this by inferring intent from timestamps, treating status changes as authoritative or reconciling after the fact.
It usually feels fine, until another location gets added.
_
At what point did you stop treating appointments as stable objects? Did you normalize early, or accept drift and contain it?_
Top comments (0)