Skip to content

Coverage Dashboard

Monitor — instantiates Target-Complete Mapping Design

A live surface that shows current coverage against the required targets, weighted by priority, and lights up the moment a newly-added target has no witness yet.

Coverage Dashboard is the live monitoring surface that renders current coverage against the required target set at a glance — not as a single vanity percentage but as a priority-weighted view that distinguishes "98% of targets" from "98% of what matters." Its defining move is continuous monitoring of a moving target set: as targets are added or retired, the dashboard recomputes and immediately flags any new target that has no witness yet, so drift-induced gaps surface in real time instead of at the next audit. It visualizes and watches; it does not itself find or fix gaps.

Example

A health department watches immunization coverage across districts and required childhood vaccines. The dashboard weights by priority — a measles gap outranks a low-severity one — and, crucially, it watches the target set itself: when a new birth cohort enters the denominator each month, any cohort-and-vaccine cell with no delivery yet flips red the same day. The headline "94% fully immunized" stays comfortably green, but the weighted, change-aware view shows one district's measles cell slipping below threshold the moment the new cohort lands — visible that day, not at the annual review when the gap would already be a season old. (Illustrative; not drawn from a specific program.)

How it works

What distinguishes it from a static report is that it is live, priority-weighted, and change-sensitive at once. It reads the coverage relation, applies target priority weights so critical gaps cannot be masked by a pile of easy wins, and monitors the target set for additions and retirements — surfacing both weighted coverage and freshly-uncovered targets continuously rather than at a point in time.

Tuning parameters

  • Refresh latency — real-time versus periodic snapshot. Real-time catches drift instantly but costs pipeline; snapshots are cheaper and laggier.
  • Weighting scheme — flat percentage versus priority- or risk-weighted. Weighting is what stops easy targets from drowning out a critical gap.
  • Denominator policy — which targets count. The honest choice keeps hard and excluded targets visible rather than dropping them (see below).
  • Alert threshold — how far coverage must fall, or how new a gap must be, before it flags. Tight thresholds catch faster but generate more noise.

When it helps, and when it misleads

Its strength is that it makes coverage a monitored, priority-aware, moving quantity rather than a point-in-time claim — new gaps announce themselves the day they open. Its weakness is that a dashboard shows only what it is fed: a wrong or gamed denominator makes a real red gap render green. The classic misuse is managing the metric instead of the coverage — quietly dropping the hard targets from the denominator so the number stays high. That is Goodhart's law in miniature: once the coverage figure becomes the goal, it stops measuring coverage.[1] The discipline that guards against it is to keep excluded targets visible on the board (in an exception register), weight by priority, and audit the denominator — so the number cannot be improved by hiding the difficult cases.

How it implements the components

Coverage Dashboard realizes the monitoring-and-display side of the archetype — the components a live surface can carry:

  • target_to_witness_coverage_matrix — the underlying coverage relation it renders, kept live rather than snapshotted.
  • target_space_change_monitor — it watches the target set and flags any newly-added target that still lacks a witness.
  • target_priority_weighting — it weights coverage so that critical gaps are never masked by a majority of easy wins.

It does not produce the underlying relation — that is the Bipartite Coverage Matrix — nor decide the fate of a flagged gap, which is Uncovered-Target Triage; the dashboard surfaces and monitors, and hands off.

Notes

A live green board can lull. A dashboard reflects its inputs continuously but never questions them, so a quietly wrong denominator or a stale feed can keep it green over a real gap. Pair it with Periodic Coverage Recertification: the dashboard is for fast signals between cycles, the recertification is the independent truth-check that the board itself is not lying.

References

[1] Goodhart's law — "when a measure becomes a target, it ceases to be a good measure." A coverage dashboard is squarely exposed to it: the cheapest way to raise the number is often to shrink what the number is measured against, which is why the denominator, not just the percentage, has to be kept honest and visible. ↩