Granularity refers to the level of detail at which data is stored and analyzed. It defines how fine or coarse a dataset is and directly impacts how metrics can be calculated and interpreted.

Examples of different levels of granularity include:

• Transaction-level data (one row per order)

• Session-level data (one row per user session)

• Daily aggregates (one row per day)

• Monthly summaries (one row per month)

Choosing the correct granularity is one of the most important decisions in data modeling. If data is stored at too high a level (over-aggregated), detailed analysis becomes impossible. If stored at too low a level, queries may become slow and complex.

From a BI perspective, granularity affects:

• Metric accuracy

• Performance of dashboards

• Ability to drill down

• Flexibility of analysis

For example, calculating average order value requires order-level granularity. Calculating revenue trends can work with daily or monthly granularity.

Granularity also affects joins. Joining tables with mismatched granularity can lead to duplicated rows and inflated metrics — a common analytics error.

Modern data stacks often store data at the lowest useful granularity in warehouses and create aggregated views for performance. This approach balances flexibility and efficiency.

Understanding granularity helps analysts ask better questions and avoid misleading conclusions.