DATA ANALYSIS TOOLS (STUDY OBJECTIVE 7)
The purpose of a data warehouse is to give managers a rich source of data that they can query and examine for trends and patterns. The data warehouse allows managers to examine important patterns or trends so that they can better plan and control the business. The data warehouse can help managers examine trends such as sales by product, region, or model over a long time frame. Data in the data warehouse are analyzed by data mining and analytical processing.
Various techniques and tools have been developed to analyze data in a data warehouse, and this analysis enhances the ability of the business to meet customer needs, improve strategic planning, and increase performance. While there are many data analysis tools and techniques, this section will describe one important category of techniques and tools. The general technique is called data mining. The tools used in data mining are generally called online analytical processing (OLAP). There are special variations of OLAP called relational online analytical processing (ROLAP), and multidimensional analytical processing (MOLAP; also called data cubes). But these special types of OLAP are much more technical topics that are beyond the scope of this chapter. The descriptions here will focus on the general characteristics of data mining and OLAP.
DATA MINING
Data mining is the process of searching for identifiable patterns in data that can be used to predict future behavior. Although there are many reasons to predict future behavior, the most popular use of data mining is to predict future buying behavior of customers. If businesses are able to more accurately predict customer buying behavior, they can plan appropriately to produce, distribute, and sell the right products to customers at the right time. Data mining techniques have considerable potential in a variety of areas. In the 1998 book Discovering Data Mining,1 the following are offered as examples of the questions that data mining can answer:
- What kind of behavior pattern does your customer emulate?
- How can the organization make more sales to existing customers?
- In the sales databases, are there hidden patterns of buying?
- Who are the better customers, and who are the high-risk customers?
- How can you maintain loyalty from current customers?
- How can you identify unknown buying habits and specifically market to those habits?
- What are customer perceptions of company products?
- How do you improve operational and strategic business plans based on data mining results?
Although it is difficult to describe the exact size of a data warehouse such as that used by Anheuser-Busch, certainly it must be very large, since it incorporates historical sales, shelf space information, and competitor information on a store-by-store basis. A data warehouse that large would require specific software tools to examine and analyze trends or patterns in the data. This is also true for any organization that maintains a large data warehouse. Software must be used to search for trends and patterns in the data. The general term for these software tools is online analytical processing, or OLAP, described in the following section.
Anheuser-Busch uses data mining to track and predict beer buying behavior. Using a combination of its own data, market data, and data from a third party, Anheuser-Busch can track its own sales and competitors' sales, revise marketing strategies, and design promotions targeted to ethnic groups. Anheuser-Busch has a name for its database and the process of using it: BudNet. The company attributes its market share growth to BudNet. Anheuser-Busch even maintains a website for sales reps and distributors to use in accessing and analyzing data. The website, www.budnet.com, is protected to allow use only by authorized parties.
Anheuser-Busch collects its own sales data by providing each of its salespeople with a handheld computer to use when they visit stores that stock Anheuser-Busch beer brands. Salespeople enter data such as new orders, shelf space devoted to Anheuser-Busch and competitor brands, and marketing promotions in use by competitors. The data are transmitted daily to regional Anheuser-Busch distributors and from the distributors to corporate headquarters. Brand managers then examine the data and provide sales and demand information and new promotion campaigns to distributors. Anheuser-Busch uses this data and computer technology to model and predict retail outlet buying patterns for the next 14–28 days. The model uses information such as sales history, price-to-consumer, holidays, special events, daily temperature, and forecasted data such as anticipated temperature, to create forecasts of sales by store and by product. Data are used by salespeople and distributors to rearrange displays, rotate stock, and inform stores of promotion campaigns.
In addition to using the internal data, Anheuser-Busch contracts with Information Resources, Inc. (IRI), to collect market sales data. IRI tracks every bar-coded purchase of beer at convenience stores and liquor stores. IRI also conducts consumer surveys of beer buyers. Using these buying trends, Anheuser-Busch creates promotional campaigns, new products, and local or ethnic targeting of markets. For example, more beer is sold by the can in blue-collar neighborhoods, whereas more bottles are sold in white-collar neighborhoods.2
OLAP
OLAP is a set of software tools that allow online analysis of the data within a data warehouse. The analytical methods in OLAP usually include the following:
- Drill down is the successive expansion of data into more detail, going from high-level data to lower levels of data. For example, if a person is examining sales, drill down would involve examining sales for the year, then by month, then by week or day. This examination of successive levels of detail is drill down.
- Consolidation, or roll-up, is the aggregation or collection of similar data. It is the opposite of drill down in that consolidation takes detailed data and summarizes it into larger groups.
- Pivoting, or rotating, data is examining data from different perspectives. As an example, sales of beer can be examined by time (months), by store type (convenience store or liquor store), by container type (cans or bottles), etc.
- Time series analysis is used to identify trends is the comparison of figures such as sales over several successive time periods.
- Exception reports present variances from expectations.
- What-if simulations present potential variations in conditions that are used to understand interactions between different parts of the business.
OLAP is the software tool that allows managers to access and analyze the data in the data warehouse. OLAP finds and highlights trends or patterns in the data.