Why your reports contradict each other, and how to fix it
Two reports meant to measure the same thing sometimes give different numbers. The problem does not necessarily come from the calculation itself. It often comes from the definition: two people, two teams or two tools are not measuring exactly the same reality, even when they use the same word to describe it.
Where these contradictions come from
The same concept can be measured differently depending on the source. Monthly revenue might include or exclude VAT depending on the system calculating it. The number of active customers might be based on an order in the last thirty days, or in the last twelve months, depending on the definition used for the report.
Data can also be extracted at different points in time. A report frozen on the first day of the month and another updated in real time can give different results for the same period. This is not necessarily an error: it may simply be a timing difference that has not been made visible.
Filters may also be applied without being sufficiently visible. A report might exclude certain customer categories, order types or periods without the person reading it knowing exactly what was filtered out. The figure appears to represent the full dataset when it actually covers only part of it.
Finally, systems are not always synchronised at the same time. A cancelled order may take several hours or days to appear in another system, temporarily creating a gap between two reports viewed at the same moment.
Why this costs more than it seems
Every contradiction that is discovered gradually reduces teams' trust in the numbers, well beyond the report in question. Once someone has seen two different figures for the same question, they tend to check the data manually more often. Over time, this significantly reduces the expected benefit of automated reporting.
Meetings can then turn into debates about whether the numbers are reliable, rather than discussions about the decisions that need to be made. Time spent understanding why two numbers differ is time not spent analysing what they reveal or acting on it.
The problem becomes particularly costly when several teams build their own indicators from the same data. Each team may have a coherent definition internally, but the organisation ends up with several figures that are all treated as the "official" number for the same concept.
How to fix it
Document precisely what each important indicator means, not just its name. Specify what it includes and excludes, which period it covers, which source is used, any relevant filters and, where necessary, how the calculation is performed.
Designate a reference source for each key data element. When several systems contain similar information, it should be clear which one is authoritative and in which situations the others may be used. This prevents each team from rebuilding its own version of the same figure.
Systematically show the date and time of the last update on reports. Users should be able to see quickly whether two different figures simply come from two different states of the data.
Centralise the calculation of shared indicators wherever possible. If revenue, the number of active customers or another indicator is used by several teams, its calculation should ideally be defined and maintained in one shared place rather than recreated across different files or tools.
Finally, when a discrepancy appears, it should be possible to trace it back to its source: the indicator definition, the data used, the period, the filters and the last synchronisation. The goal is not only to correct today's number, but to understand why the discrepancy appeared so that it does not happen again.
The takeaway
Reports that contradict each other do not necessarily indicate a calculation error. They often reveal a different definition, a different source, a different state of the data or a calculation performed in several places. Documenting what each indicator represents, designating reference sources and centralising shared calculations makes the numbers more consistent and, above all, easier to understand and verify.
Two different numbers do not necessarily mean that one of the calculations is wrong. If your teams regularly debate which numbers can be trusted instead of what those numbers mean, I can help establish shared definitions and reliable sources.