A Variance Is a Question, Not an Answer

A reconciliation can look deceptively complete. Two columns are compared, a difference is calculated, and every nonzero result is marked as a problem. The arithmetic may be correct, but the analysis has only started.

The same variance can come from very different causes: a legitimate price change, a timing difference, a duplicate record, a unit-of-measure mismatch, an incorrect mapping, or a transaction that never reached the second system. Treating all of those as one issue produces a long exception list without telling anyone what to do next.

I saw this clearly during a post-acquisition platform transition. I reconciled roughly 200 supplier transactions, identified pricing discrepancies, and prepared cleaned analysis for leadership review. The useful output was not the count of mismatches. It was the separation of explainable differences from records that required a decision or correction.

A strong reconciliation starts by defining what should match. Which source owns the supplier, item, quantity, price, and effective date? Which timing differences are acceptable? What tolerance is meaningful? Without those definitions, a clean-looking comparison can hide inconsistent assumptions.

Then each material exception needs a disposition. Confirmed, corrected, expected, escalated, or still under investigation are more useful states than open and closed. They preserve the reason behind the result and make it possible to see whether the same failure is repeating.

The final step is to connect the variance to a response. A pricing discrepancy may change a purchasing decision. A mapping error may require a system fix. A timing difference may need no intervention at all. The analysis is complete when the business can distinguish those cases and act proportionally.

I trust a reconciliation when it shows the expected relationship, the material variance, the evidence behind the explanation, the owner, and the disposition. The difference is the signal. The explanation and response are the work.