Device Shipping — Operations Dashboard

Aggregates only, no PHI. Recompute (Jun 2025+, maturity-gated). Shipped unit = outbound package_ledger rows. Every number traces to a live read-only query against ansiblehealth_analysis.

Scope: Jun 2025–present · per-package ledger · see notes below for reconciliation.

1 · Packages shipped vs patients newly active — Jun 2025–present

package_ledger outbound monthly vs first came-alive per patient. Full history (Jun 2025+); months marked * are partial. Routers excluded from "active" (no patient key).

2 · Lifecycle — what happens to a shipped device (matured ≥180d)

Full history (Jun 2025+) matured cohort (first device-shipment per patient ≥180d old). Buckets sum to shipped.

3 · Where carrier exceptions go (Jun 2025+)

Classified from carrier timeline.details (the substatus code field is null) — a different basis than the prior substatus-code classification; the conclusion (exceptions mostly deliver) is robust. Exception detail is ~100% populated across the full ledger history, so this covers the full range (Jun 2025+). Flag categories overlap, so flows are per-category, not additive.

4 · Carrier exception flags → ultimate status (Jun 2025+)

Each flag's ultimate ledger status. Flags overlap (a package can carry several); rows do not sum across.

5 · Reships by device family (Jun 2025+)

Reship = a later separate package containing a device family the patient already received (any gap). Gross hardware reproduces the prior published basis.

6 · Recovery — it works at the item level

packiyo_returns + packiyo_return_items.quantity_received. Per-family split is approximate (shared SKUs); the received % total is exact.

7 · The verified bleed — delivered but never received

fh_compute_e: b1 (carrier-visible loss) + b2 (delivered then reshipped). The ambiguous non-engagement tail is excluded from the dollar figure by policy.

The story in three beats

Cost bases — what each backed number measures

Different metrics on different cohorts — NOT additive, NOT one headline. Excluded by policy: offboarded upper bound, non-engagement tail, came-alive difference-in-differences.

Data sources & method