Migrating Off an Incumbent

The migration is where the argument is settled. An organization that cannot reconcile its own history against the incumbent has learned something about its records, not about the new system.

The sequence is unglamorous: import patients, reconcile historical activity, match active programs, map care plans, validate time and documentation, run alerts in parallel, and compare billing-support output against what the incumbent produced.

The design decision that matters is what happens to the residue. In the reference implementation, historical activities that cannot be reconciled and care plans with no machine-readable equivalent are listed individually and hold the readiness score down, rather than being counted as migrated. Those are not import failures. The source records do not contain what the incumbent's interface displayed, and that discovery is one of the more valuable outputs of the whole exercise.

Parallel running produces the second discovery. Shadow-mode disagreement is rarely uniform: it concentrates in identifiable contexts, and those clusters are a map of where the new system is not ready and where the operation's own practice is undocumented. Cutting over before that map is read is how migrations acquire their reputation.

That migration risk is the strongest reason to stay; that parallel running doubles operating cost for a quarter; and that a readiness score is a vendor's own mark of its own homework.

All three are fair. The third is answerable only by making the score's inputs inspectable, which is why the unreconciled records are listed rather than summarized.

Post-cutover measures with pre-cutover baselines: reconciliation completeness, billing continuity through the first two closes, the number of workflows that had to be rolled back, and staff hours during the parallel period. A migration nobody measured cannot inform the next one.

How long a parallel run must be to be informative rather than merely expensive. Whether disagreement clusters are portable between organizations or specific to local practice. What is owed to records that can never be reconciled, and who decides.