Two computations that should agree give different answers.
Reconciliation
Finance has one number for the month, risk has another, and both come from the same feed. Every month we reconcile them by hand, and nobody can say why they differ.
The operations lead
Where do these two computations first diverge?
The product at work
Finance's month-end total and risk's, side by side, and they differ.
Both computations traced back from the figure, line by line, from the same feed.
Each matching step tied off: the same source extract, the same report step.
The first step that differs, ringed: one divides by ten, the other by one hundred.
What usually happens
Two analysts walk the two pipelines side by side with a spreadsheet, column by column, until the week runs out. The gap gets an adjustment line and a name, and next month it is back.
With a reading
With a reading, both computations are read from their code and the first place they part company is found, down to the step, the condition and the line on each side, and the argument about which is right can start there.
The earliest structural divergence the supplied materials establish, with the relevant source spans and conditions retained for review.
Both computations were traced through the code behind them. Whether that step is also why the figures differed on the day is confirmed against execution evidence, when it is supplied.
Who it is for
For the operations lead, finance manager or risk officer who signs an adjustment every month without knowing why it exists, until the Lineage record and Finding register show it.
What to bring
The code, configuration and schedules behind both computations, and the shared inputs they start from.