Measures (Quantifiable Data)

Measures represent the quantifiable data in a cube. They are always numeric values that result from an aggregate calculation, such as a COUNT or SUM, on the fact data.

In an analysis, a measure is used to answer business questions such as 'how much?', 'how many?' or 'how long?'. Examples are in the following table.

One way to determine the measures of a dataset, is to look at the columns in a fact dataset, and ask yourself how that data might be measured. Choosing the right fields in a dataset to serve as measures, and the right aggregations to apply to those fields, depends on how you plan to analyze the data. For example, a product ID number might not make sense to sum or average, but it might make sense to count how many distinct product IDs were sold.

Dimensions (Categorical Data)

Dimensions organize the data and provide context for a cube's measures. Dimensions are logical collections of attributes, which are bound to one or more columns in a source dataset. A dimension attribute is data that answers business questions such as 'who?', 'what?', 'where?', or 'when?'. For example:

In BI client applications, the dimension attributes are used to GROUP and FILTER the measure data. When designing the dimensions of a cube, think about how you want to slice, dice, drill into, and filter the data in your analysis, then design your dimensions to support those operations.