Dimensional modeling
A dimensional model separates measurable events from the descriptive context people use to analyse them.
Facts and dimensions
Orders or order lines can act as facts; customers and products can act as dimensions. Facts store measures and dimension keys. One clear fact-table grain is essential.
Star schemas and conformed dimensions
A star has a central fact connected to dimensions. A conformed dimension gives multiple facts the same definitions so measures can be compared consistently. Sharing a column name alone is not conformance.
Slowly changing dimensions
Type 1 overwrites history. Type 2 keeps multiple versions with effective dates and often surrogate keys. A fact must connect to the correct version, using non-overlapping validity ranges and a clear point in time.
Metric definitions
Revenue, units and customer counts have different aggregation behavior. Ratios and distinct counts usually cannot be safely added across arbitrary groups. Document grain, filters and additive behavior.