Reading a finding. Every part of the card.

• The check that found it, in small capitals, such as Net hiding gross.
• Money at stake, computed by the check, with its arithmetic in the Test tab.
• How sure it is: a state, and for an assessed finding, its confidence.
• What changed since last night: New, Worse, Better, or Open with its days.
• A title that takes a position, one or two lines, and a chart drawn from the computed result. Missing-records findings list the accounts instead, because a customer who stops ordering leaves no rows to chart.
• Why you haven’t seen this: the reason your reports miss it.
• A question to ask, for one named person, that would settle it.

Measured

A computed fact: a value, a share, a gap, the size of a step. It says nothing about the cause.

Assessed · High, Medium or Low

The claim rests on a statistical test. High means very strong evidence and a large effect; Medium means it passed the test. Low is never shown in a brief. Confidence is the strength of the evidence, not how much it matters: the money says that.

Not determinable

The fact is measured, but the explanation you need can’t be settled from the data connected. The finding lists the possible reasons, which ones the data rules out, and what would settle the rest.

Needs confirmation

Two of your systems disagree about the same number, and only you can say which is right. It never becomes a brief item; it becomes a question on Home.

Written by rules from facts, never by a model. Usually one of four reasons: the number isn’t one any report tracks; it isn’t in the reports you uploaded; the total hides it; or the source has no status for it, so nothing records the change.

Keep an eye on it. Othor AI drafts what to measure, a threshold and an owner from the finding, and shows how that watch would have behaved over the last 90 days before you save it. Every night the run checks it. When it crosses the threshold, it shows on Home and as one line in every brief that covers the number. If it is already over the threshold when you set it, it starts in breach.

Not now. The finding is held for 7 days, and every count drops it at once. When the 7 days are up, it comes back with its first-seen date, so you can see how long it has really been open.

Stop telling me about this pattern. It holds this finding and any later one from the same check on the same number and the same accounts, with no end date. Muted findings are listed in Suppressed, never deleted.

Snooze and Mute each record who did it and when, and Undo ends them. A brief already sent doesn’t change, and the night’s stored run is never rewritten.

Pick a brief and see where the finding would rank tomorrow. Add it as an item, or watch it. A brief whose scope doesn’t cover the number can’t take it until its scope is widened.

Opens the finding’s Test tab: the test, its parameters and its result. The other tabs show the finding, the number’s definition, the exact SQL, and the run it came from. See How every number is made.