Measurement
The bucket labeled current is lying to you
On one engagement the past due balance was more than four times what the screen reported. Both figures were defensible. Only one of them was being acted on.
Two right answers
A receivables screen computed past due as everything outside the current bucket. A scorecard on the same data computed it from days past due directly. The second figure was more than four times the first.
Neither calculation was wrong in isolation. The problem was that a bucket labeled current zero to thirty contains invoices that are already one to thirty days late, and everyone reading the label assumed otherwise.
Why labels beat definitions
People act on the word at the top of the column, not on the definition in a data dictionary they have never opened. If the label implies a meaning the calculation does not have, the label wins every time.
This is why the first week of an engagement is spent agreeing what each measure means and naming a person who settles disputes about it. It looks like process overhead. It is the cheapest defect prevention available.
The fix is structural
The calculation moved into one place, so both surfaces read the same figure instead of each working it out again. Screens read totals now. They no longer recompute a headline from whatever rows they happened to load.
Once a measure is modeled once and read everywhere, two screens cannot quietly disagree. That property is worth more than any individual chart.
Want this against your own numbers?
A scoping call works out which measures are worth baselining and whether your systems can actually be read.
Let’s chat