Synthetic demonstration data — State of Cascadia is fictional. Methodology is real.
Screen 4 · how the real engagement works

The dashboard is the reporting layer. The reconciliation is the work.

What this demo simulates

The State of Cascadia is fictional, but its conditions are not: an MMIS vendor transition in flight, county-administered eligibility, and a monthly Medicare COB exchange. Those are the exact conditions under which the real nine-figure breaks were found — a tens of thousands-member enrollment gap discovered mid-transition, and a nine-figure federal disallowance resolved when the reporting was called impossible.

Every number on these screens was computed, not invented: a generation script builds three internally consistent systems (~150,000 members, ~863,000 enrollment spans), plants six classes of cross-system discrepancy, and the reconciliation engine recovers them — validated to the member against the planted truth, with zero false positives. The methodology you are looking at is the methodology that runs against real extracts.

How the real diagnostic runs

  • Read-only extracts. Eligibility spans, enrollment and capitation records, and the COB/buy-in file. No system access, no writes, no PHI leaving your environment — the analysis runs on your infrastructure.
  • Two to four weeks. Week one: load, profile, and verify each system is internally consistent — the same ✓ line you see on Screen 1. Weeks two to three: cross-system reconciliation across the six break classes. Final week: member-level validation of every finding.
  • A findings memo, not a slide deck. Member counts, dollar exposure with the math shown, example member histories, and a federal-exposure tag per class — the same anatomy as Screen 2.
  • CMS-proactive framing. Findings you take to CMS before an auditor takes them to you. Self-identified issues are negotiated; discovered ones are disallowed.

What is synthetic here, and what is real

  • Synthetic: the state, the member records, the county names, the plans, the dollar figures. Seeded and reproducible; no real data of any kind.
  • Real: the six break classes, the detection logic, the exposure math, the vendor-transition and county-administration failure modes, and the experience they are drawn from — a career spent inside state Medicaid systems, on the state side of the table.
Book the diagnostic

If your state is mid-transition, county-administered, or carrying a COB backlog, the seams are already there. Two to four weeks of read-only extracts will tell you what they hold.

jeff@provenanceadvisorsllc.comprovenanceadvisorsllc.comLinkedIn