One margin number: getting finance and merchandising to agree
When two departments bring different margin figures to the same meeting, the problem is rarely the maths. It is the data model. A practical path to a shared definition.
Every retailer has sat in the meeting. Finance says margin was 41.2%. Merchandising says 43.8%. Both are right, by their own definitions, and the next twenty minutes are spent arguing about which definition is correct instead of what to do about the number.
Why they differ
The usual suspects: whether supplier rebates are included, how markdowns are timed, whether shrink is allocated to product or to store, and which exchange rate was used for imported lines. None of these is wrong. The problem is that each department computes them in its own spreadsheet, from its own extract, on its own schedule.
The lakehouse is the meeting room
A shared data platform — Microsoft Fabric in our practice — does not resolve the definitional argument by itself. What it does is force the argument to happen once, in the semantic model, and then be over. Both teams read the same measure. If merchandising needs a pre-rebate view, it is a second measure with a clear name, not a second spreadsheet.
A four-week path
Land the source data. Sit both teams down with a whiteboard and write the definitions. Encode them as measures. Retire the spreadsheets. It is less a technology project than a diplomatic one, but it only holds if the technology makes the shared definition the easy path.