A sales forecast in Power BI has to combine three things: pipeline data, historical conversion behaviour, and the judgement of the people who own the numbers.
1. Build the baseline
Start from actual revenue and pipeline by stage, product and region. Use DAX measures for weighted pipeline and run rate, and keep the grain consistent with how targets are set.
2. Add the human layer
Statistical projection is the starting point, not the answer. With accoPLANNING, regional leads adjust their own forecast lines in the report and the values write back to SQL immediately, so the consolidated number is always current.
The forecast improves when the people closest to the deal can change it without emailing a spreadsheet.
3. Run scenarios
Keep a named version per case, best, expected and worst, so the comparison is one slicer rather than three files. Version tags travel with every plan row.
4. Track accuracy
Store each submitted forecast version. Comparing forecast to actual by round is what turns forecasting from an exercise into a discipline, and it is only possible if history is retained.
Common pitfalls
- Mixing target and forecast in the same field.
- No version tag, so history is overwritten each round.
- Forecast held at a grain that cannot be reported against actuals.
- No place to record why a number changed.
