Every recurring KPI review meeting has a failure mode that experienced managers know by heart: the meeting spends its first twenty minutes debating whether the numbers are right instead of deciding what to do about them. The cause is almost never bad data — it is unexamined data, walked into the room without a freshness time, an owner, or a stated list of known caveats. The data source check described here is the twenty-minute ritual, run the day before the meeting, that moves those twenty minutes out of the room. It operationalizes the standing check from KPI data source check before the review at meeting frequency, feeds the question discipline of dashboard review questions before the weekly meeting, and leans on the KPI owner definition when the check finds a metric nobody owns.
The check has four moves, and its output is a one-page brief the meeting opens with — thirty seconds to read, and the whole debate about data quality is settled before it can start. Small-business managers running their first review cadence will find the broader measurement framing in the SBA’s guide to managing a business useful context for why disciplined numbers review belongs alongside cash-flow and operations review rather than after them.
Why Numbers Get Debated Instead Of Used
Understand the mechanism and the fix designs itself. When a number appears on a dashboard without provenance, every person in the room silently applies their own guess about how current it is and how it was calculated — and the guesses differ, so the number means different things to different people and the meeting argues about the meaning instead of the decision. Freshness is the most common silent disagreement: one person assumes the revenue figure includes this week, another knows the export runs Sunday night, and neither says so until the number contradicts someone’s expectation. Ownership is the second: when a metric drifts or breaks, a meeting without named owners discusses the anomaly abstractly and assigns nothing. The data source check exists because all three of these failure modes are cheap to prevent a day early and expensive to repair in the room — the check is, in the strictest sense, twenty minutes spent to buy back the meeting’s entire attention.
The Four Moves The Day Before
| Move | What goes in it | Minutes |
|---|---|---|
| Freshness pass | Confirm every headline metric’s data is current as of an agreed cutoff; mark anything stale and say when it ends | 5 |
| Owner pass | Every metric on the agenda has a named owner present or represented; gaps get assigned before the meeting | 5 |
| Caveat pass | List known data quirks affecting interpretation — tracking gaps, definitional changes, one-off events | 5 |
| Broken-source pass | For any source that failed or is late: decide now whether the meeting proceeds on last-known values or drops the metric | 5 |
The broken-source pass deserves emphasis because it is the one teams skip and regret. When a source is broken, the room without a prior decision will either pretend the stale number is fine or stall the agenda improvising. Deciding the day before — proceed on last-known with the staleness stated, or drop the metric and reschedule its discussion — turns a meeting crisis into a footnote. Either decision is defensible; making it in the room under time pressure is where the damage happens.
What The Twenty Minutes Actually Look Like
The check is more concrete when walked through once. Open the dashboard and read each metric’s source line top to bottom — for a twelve-metric dashboard this takes about eight minutes. Against each source, answer three questions in the margin: when did this data last refresh, who confirmed it is calculating as defined, and is there anything known to be wrong with it right now. Most metrics answer all three cleanly and take thirty seconds each; the time goes to the two or three that do not — the revenue export whose automation silently failed last weekend, the traffic figure that still counts the old domain, the support metric whose owner left and was never reassigned. Finishing pass, the facilitator writes the one-pager, emails it, and the check is done for the week. The first run takes forty minutes because the sources must be located and documented; by the third week it is reliably twenty, because the documentation from the first run is the map every later run follows.
The One-Page Brief The Meeting Opens With
The check’s output is deliberately small: one page, three sections, read in thirty seconds. The first section states the data cutoff — as of Friday 18:00, all figures current except where noted — which instantly synchronizes every reader’s assumption about freshness. The second lists the exceptions: two metrics are stale with their cutoff dates, one has a known tracking gap since the site change, one source failed and its metric is dropped from this agenda. The third names the owners present, so when a number surprises the room, the meeting knows who can answer in one question rather than three. This page is not bureaucracy; it is the difference between a meeting that opens on decisions and one that opens on archaeology. Teams that adopt it consistently report the same shift — the data-quality debate does not get resolved, it simply stops being necessary, because the answers arrived before the questions could.
Who Runs The Check, And When
Twenty minutes of preparation needs an owner, and the natural one is whoever facilitates the review meeting — the operations lead, the office manager, the founder in a small company. Assigning it to the facilitator rather than to each metric’s owner matters: individual owners will check their own numbers generously, while the facilitator checks all of them against the same standard, which is the point of a standard. Timing has one rule: the day before, never the morning of. A check run the morning of the meeting can find a broken source but leaves no time to decide what to do about it, converting the ritual into a smaller version of the crisis it was designed to prevent. The day-before rhythm also lets the facilitator circulate the one-page brief by email in advance, so the thirty-second read happens before anyone enters the room and the meeting opens directly on decisions. In companies where the facilitator role rotates, the check rotates with it; what must not rotate is the habit itself, because the second consecutive skipped check is exactly when the old data debates creep back in.
When A Metric Fails The Check Repeatedly
The check has a long memory worth using. A metric that arrives stale or caveated three meetings running is telling the team something the meeting keeps working around: the source pipeline is unreliable, the definition is unstable, or the metric is not worth its extraction cost. The standing rule: the third consecutive failure triggers a fifteen-minute agenda item outside the review — fix the pipeline, simplify the definition, or retire the metric. This rule matters because metrics that fail quietly accumulate interpretive debt; every stale week adds another caveat the room must remember, until the metric exists as a ritual number everyone discounts and no one retires. Retirement, it should be said, is a legitimate and underused outcome: a dashboard that drops its least trustworthy metric gains credibility for the ones that remain. The data source check is not only hygiene, then — it is the mechanism by which a measurement system stays honest about what it actually knows.