Dimensions

Dimensions, in terms of Business Intelligence, are qualitative values that are used to describe and categorize Metrics which are quantitative.

Dimensions are used to 'slice and dice' metrics, and to 'drill down' to obtain more detail. For example, if a report highlights the number of sales per product, sales is the metric and it is sliced by product which is a dimension. If the report is further broken down to show state specific figures, state is another dimension in the report. Dimensions can also be hierarchical. The most common dimension in Business Intelligence reports is Time, which generally takes the form of a hierarchy such as Year, Quarter, Month and Day.

Contents

• Identifying Dimensions
• See Also
• Sources and further reading

Identifying Dimensions

The metrics and dimensions tracked by different organizations vary depending on the strategic goals and objectives of the organization. Many associations share a common goal of increasing membership over time. Some dimensions that may add value and assist in meeting that goal are explained below.

• Time: Time is the most commonly used dimension as it is a necessity for trend analysis and predictive forecasting. Time generally takes the form of a hierarchy such as Year,Quarter,Month,Day.
• Chapter or Affiliate: Membership is generally identified for each Chapter or Affiliate, and then aggregated at the high level.
• Region: Region is a valuable dimension as it helps to identify variances in metrics that are present due to regional factors. For example it may be discovered that individuals in the Northeast move more frequently than those in the Midwest, contributing to a higher churn percentage in the Northeast. Region is often expressed as a hierarchy so that data can be drilled into. A sample path of the hierarchy may be USA,Northeast,New England,Vermont. It may alternatively include a lower level of detail such as city, or be based on zip code.
• Tenure: Trends often appear when sliced by member tenure. For example, it may be found that members who have held membership for 10 years or longer are less likely to cease membership while those with a very short tenure of a year or less are the most likely not to renew next year.
• Age: Individuals of different ages tend to make decisions differently, therefore creating patterns in the data when sliced by age bracket. Marketing strategies often target the various age groups with different strategies depending on what the data reveals.
• Income Bracket: Stratifying metrics by Income Bracket often leads to the surfacing of previously unseen patterns. The findings can be used to determine how much of the marketing budget should be spent targeting particular brackets based on their historical and predicted added value.

The examples above highlight just a few possibilities chosen from thousands of available demographics. The art of Data Mining and Trend Analysis is finding the metrics and dimensions that provide the necessary insights to meet your organizations objectives.

See Also:

Metrics
Cubes
Analysis Services
Business Intelligence

Sources and further reading:

• Overview of Dimensions in SQL Analysis Services: http://technet.microsoft.com/en-us/library/ms174527.aspx
• Defining Cube Dimensions in SQL Analysis Services 2005 is explained here: http://technet.microsoft.com/en-us/library/ms174900.aspx

The core of this entry was written by:
Ann Wallinger
Solutions Architect
Susquehanna Technologies
e: annw@susqtech.com