This is a ready-to-use checklist. Run it before every management review, one pass per reporting period, and retain the completed record. The purpose is simple and unforgiving: the number on the slide must equal an independent count from the system of record. The cheapest place to catch a mismatch is here; the most expensive is when an inspector catches it, because once one number is wrong, every number on the dashboard is suspect. Replace every <<FILL: ...>> placeholder. A filled specimen follows. This content is general educational reference, not legal or regulatory advice.
Header
| Field | Entry |
|---|---|
| Reporting period | <<FILL: e.g. Q2 2026>> |
| Dashboard version reviewed | <<FILL: version / run date>> |
| Reconciled by | <<FILL: name, role>> |
| Date performed | <<FILL: date>> |
| Management review date | <<FILL: date>> |
Part A: Per-metric reconciliation
For each metric on the dashboard, pull an independent count from the source system and compare. Record the two numbers, not just a tick.
| # | Metric | Source system | Dashboard value | Independent source count | Match? | Discrepancy resolved (ref) |
|---|---|---|---|---|---|---|
| 1 | <<FILL: metric>> | <<FILL: QMS / LIMS / LMS>> | <<FILL>> | <<FILL>> | Y / N | <<FILL>> |
| 2 | <<FILL>> | <<FILL>> | <<FILL>> | <<FILL>> | Y / N | <<FILL>> |
| 3 | <<FILL>> | <<FILL>> | <<FILL>> | <<FILL>> | Y / N | <<FILL>> |
| … |
Part B: Program integrity checks
| # | Check | Pass / Fail / NA | Evidence / comment |
|---|---|---|---|
| B1 | Each metric traces to a single validated system of record (no orphan spreadsheet as the sole source) | ||
| B2 | Every reported value uses the current approved metric definition sheet | ||
| B3 | Any definition change this period is annotated on the trend at the change point | ||
| B4 | No data point was excluded without a written, pre-defined rule | ||
| B5 | Volume-sensitive metrics are normalized (per batch, per million units), not raw counts | ||
| B6 | OOS is reported as three numbers (initial, confirmed, invalidated fraction), not one blended figure | ||
| B7 | Cycle-time metrics show the aging distribution or percent within target, not the mean alone | ||
| B8 | Each tile carries owner, definition reference, source system, target, threshold, and trend | ||
| B9 | The reporting tool’s audit trail shows no unexplained edits to pulled values | ||
| B10 | Prior-period actions are re-listed for the review with current status |
Part C: Signoff
| Role | Name | Signature | Date |
|---|---|---|---|
| Reconciled by | <<FILL>> | ||
| QA reviewer | <<FILL>> |
Acceptance criteria
Reconciliation is acceptable to present when: every metric in Part A matches its source count, or every mismatch has a resolution reference and the dashboard was corrected before the meeting; and every Part B check passes or has a documented, time-bound action. A single unresolved mismatch carried into the meeting fails the reconciliation.
What to do with a mismatch
- Do not silently overwrite the dashboard. Find the cause first: a late-entered record, a filter difference, a definition drift, a double count, or a genuine data-integrity concern.
- Correct the source of the error, not just the displayed number, so the same mismatch does not recur next period.
- If the cause is a definition or filter difference, update the metric definition sheet under change control and annotate the trend.
- If the cause could be a data-integrity issue in the source system, raise it under
<<FILL: SOP-ID for data integrity events>>.
Filled specimen
The following shows Part A completed for three metrics, with one mismatch caught and resolved before the meeting. Numbers are illustrative.
| # | Metric | Source system | Dashboard value | Independent source count | Match? | Discrepancy resolved (ref) |
|---|---|---|---|---|---|---|
| 1 | Deviation rate (/100 batches) | QMS deviation module | 7.5 (18/240) | 18 deviations, 240 batches | Y | n/a |
| 2 | Open CAPAs over 180 days | QMS CAPA module | 3 | 4 | N | Dashboard filter excluded one CAPA reopened on the last day of the period; filter corrected, value updated to 4 before review; note added to MDS-018 |
| 3 | Initial OOS rate | LIMS | 0.60% (24/4000) | 24 initial OOS, 4000 results | Y | n/a |
In this example the aged-CAPA count was understated by one because a reopened CAPA fell outside a date filter. Catching it here turned a potential “your slide says 3, the QMS says 4” inspection moment into a corrected number and a filter fix. That is the entire value of the checklist.
Common inspection findings this checklist prevents
- The slide says one thing and the QMS says another, found first by the inspector.
- A dashboard number that cannot be traced back to a system of record.
- Aged items smoothed out of the rollup because a filter quietly dropped them.
- Cycle time reported as a healthy mean while a tail of year-old investigations sits open.
- Undocumented exclusions discovered during data review.
How to adapt this checklist
- List your actual mandatory metric set in Part A.
- Point the escalation reference in the “What to do with a mismatch” section to your real data-integrity event procedure.
- Automate the independent counts where the source system allows a second, separate query, so the reconciliation itself is not just the same pull twice.
- File the completed checklist as an input to the management-review record.