Skip to content

Lesson 15 — One channel cannot balance

Some source systems deliver only one material-flow channel — I/O reconciliation isn’t an analytical choice, it’s structurally not computable

The I/O coefficient — material movements in (warehouse-to-ward consumption) versus material billed out (charge events) — is one of the most powerful signals in the pack. It surfaces structural revenue leakage, documentation gaps, and operational compromises that other signals cannot.

It also requires two data channels: a consumption channel and a billing channel, each emitted by an independent source system. When the source system delivers only one of the two, the I/O coefficient is structurally not computable, no matter how clever the analyst.

This Wisdom is a property of the source system, not a property of the analyst’s work.

In a hospital with OPALE-style data delivery, the I/O coefficient works because:

  • Channel A — consumption. The clinical information system records material pulls from satellite stocks, posted overnight, surfaced in gold_material_movements.
  • Channel B — billing. The patient administration system records charge events tied to encounters, surfaced in gold_billing_events.

The two are independent: same material can appear in either or both, and the reconciliation surfaces the gap. The thesis layer reads the gap honestly.

In a hospital with SAP IS-H-style delivery, the picture changes. SAP IS-H produces a unified material-and-billing stream where both consumption and billing flow through the same posting chain. The data shape looks the same in the gold layer — there’s a gold_material_movements table and a gold_billing_events table — but they are projections of the same underlying ledger, not two independent channels. Asking I/O reconciliation on this data is asking a single source to disagree with itself.

The two-channel property is invisible from the data shape alone. Both tenants present the same table names, the same columns, the same row volumes. The structural difference lives in the source system’s architecture, which is a fact the analyst has to know, not see.

The Wisdom motivated a clean fix:

  1. Per-source-system metadata declares each pack tenant’s I/O reconciliation capability. SAP IS-H tenants are flagged as single-channel; OPALE tenants as two-channel.
  2. signal_io_coefficient and thesis_revenue_leakage_unbilled skip on single-channel tenants — explicitly, not silently. The Explorer surfaces the skip with a note: “I/O reconciliation is structurally N/A for this tenant’s source system.”
  3. The Net Exposure framework (gold_net_exposure) reads the structural-N/A signal honestly. The headline number for a single-channel tenant is N/A, not zero. The difference matters — zero implies no leakage detected; N/A implies cannot be assessed with this source.

A consequential side effect: the pack documentation now explicitly enumerates which signals require which source-system capabilities. A future tenant on a new source system can be checked against the matrix before any analysis runs.

  • A two-channel tenant whose I/O coefficient converges to 100% is unusual and worth investigating. Most hospitals show structural gaps (ward base consumables, documentation lag, billing-cycle effects). A perfect coefficient often indicates that the two channels are not in fact independent — that someone has wired billing to follow consumption automatically, defeating the reconciliation.
  • A single-channel tenant migrating to a two-channel source (re-platforming) suddenly gains I/O reconciliation capability. The transition’s first quarter of data is worth investigating as a baseline-establishing event.

Because “signal not applicable on this tenant” is one of the hardest things for an analytical system to say honestly. The temptation is always to emit something — a zero, a placeholder, a best-effort estimate. None of those reads correctly to a CFO or auditor. The Wisdom anchored the discipline of N/A as a first-class value, distinct from zero and from missing data. The Lesson is that some pack tenants have analytical capabilities others structurally cannot, and the system’s job is to name the difference, not to paper over it.

  • Wisdom: smebit_sap_ish_single_channel.yaml
  • Provider: a hospital IT integration lead with SAP IS-H deployment experience
  • Anchor signal: signal_io_coefficient
  • Anchor thesis: thesis_revenue_leakage_unbilled
  • Date Wisdom captured: 2026-05-12

This is the last entry in the current nuMetrix Lessons series. Lessons 01–15 cover the fifteen Wisdoms in the pack’s SMEbits directory as of the 2026-05-12 snapshot. The series will extend as new Wisdom gets added.

jazzisnow jinflow is a jazzisnow product
v0.64.7 · built 2026-09-20 19:48 UTC