LEARN. PRACTICE. GROW.Concept and mechanism
An analytical model gives meaning to tables, relationships, and measures. In a star schema, facts represent occurrences or observations and dimensions support filtering and grouping by attributes. Grain should state what each row represents: payment, daily account, and monthly total are different units. Mixing units or relating facts to duplicate keys can multiply amounts. Before building charts, confirm keys, cardinality, filters, and measure definitions. A correct visualization of a wrong total still misleads decisions. Review the semantic model with people who understand the process and business rules.
Guided application
Power BI can use imported data or source queries among other options. Import retains a copy needing refresh; opening the report today does not make data current. DirectQuery queries the source during interactions but requires assessing performance, limitations, security, and caches without promising universally instantaneous updates. In a fictional committee, show the effective data period and update status. For daily evolution, use an appropriate time-oriented visual; bars can clarify category comparisons; a card shows one value. Test totals with small known examples, including duplicated dimensions. Define who investigates loading, refresh, and formulas when reporting differs from the source.
Last refresh yesterday at 18:00: a report opened at 09:00 can still represent the previous day.
Common pitfalls
Browser time as freshness; duplicate key as harmless; chart as data correction; DirectQuery as an unassessed fix.
Related topics: Data, workloads, and responsibilities · Relational modeling, integrity, and SQL · Relational services and compatibility
Explain meaning, grain, and freshness before asking for a decision.
Reference: Fact dimension grain and relationships · DP-900 skills measured 2026-07-21