Power BI star schemas: connecting sales tables correctly
The foundations of reliable Power BI reporting: fact and dimension tables, unique keys and filter direction.
Sample period date: Aug 8, 2015. These essays revisit earlier technologies from today's editorial perspective; they are not represented as posts originally published in those years.
Separate facts from dimensions
Sales rows represent measurable events such as transactions, dates and amounts. Product, customer and calendar dimensions provide descriptive context. Combining everything into one enormous spreadsheet can feel convenient initially but repeats the same customer and product details many times. A star schema makes relationships and calculation grain easier to understand.
Look for key uniqueness problems
If the product dimension contains the same product code twice, totals can become ambiguous or relationships may not behave as intended. Check uniqueness, missing keys and row counts before building visuals. Do not enable bidirectional filters merely to make a report look correct; define the intended direction of filter propagation.
Measures and columns have different roles
A calculated column is evaluated at row level while a measure is recalculated under the visual's filter context. Differences across categories can arise from filtering rather than a broken formula. Define a reusable base sales measure and test it with known examples before building a report of many visuals.
Do not project today's features onto 2015
This is a present-day retrospective on the early public Power BI era. It would be inaccurate to attribute modern AI features to the product as it existed in 2015. The durable lessons of that period were stable keys and well-defined measures. Tools have changed, but the underlying model disciplines remain essential.
Comments
0 real commentsNo real reader comments yet.
✎ Write a comment