Sales data analysis · 2025
A sales review the client still opens
ShippedIndependent engineer
Seven months of sales records raised the same questions at every review: what drove revenue, where the gaps were, and who returned. The answers lived in a spreadsheet only one person trusted.
The constraint
The deliverable had to survive the meeting. The client already worked in Excel and Power BI, so the result had to fit those tools.
Built with
Power BISQLExcel
How it works
Records into a model. Excel for the working surface. Power BI for the view that is opened each quarter.
The calls that shaped it
Each decision with the pressure that forced it and the price it keeps costing.
Model first
The same questions came back at every review with no agreed definitions behind them, so two answers could disagree and both be defensible.
Drivers, regions, and returning-customer behaviour are defined in the model, then shown. Charts without that work are decoration.
The cost: Definitions are fixed before anything is drawn, so a new question means going back to the model.
Workbook and Power BI
The review already happened in Excel and Power BI. A new tool would have been a login nobody opened after the first month.
Both, because the review still happens in those tools. I did not wrap the data in a web app they would stop opening.
The cost: Two artefacts to keep in step, in tools I do not control.
The model is what survives the meeting. Whether the deliverable is a worksheet or a dashboard, the analysis is built first so the review still holds six months later.
Where it stands
Shipped. A model the client still opens each quarter, built so the review holds up long after we stop working together.
- Seven months of sales records modelled into drivers, regions, and returning-customer behaviour.
- Delivered as an Excel workbook and a Power BI view, both still opened at quarterly reviews.
- Shipped.
What was handed over
- The workbook
- The Power BI view
- Which questions the model actually answers