Few people compile a complete view of the information, from promotions to final collections, because data must be pulled from multiple sources. The process becomes more complicated if each system stores data in unique ways. For example, marketing refers to the customer by name, order entry is organized by purchase order number, and accounts receivable is keyed to the customer’s account number.
To overcome these difficulties, many eto erp companies have copied the data decision makers need from the legacy systems to a data repository. Special software automates this repetitious task. During the transfer, the data is “cleansed.” Cleansing reconciles the dissimilarities, handles missing values, and integrates the data into the repository in a consistent fashion. In the previous example, one key would be used to identify each customer in the data repository.
Several different names are commonly used for data repositories, depending upon the type of data stored and their intended use. For example, a data warehouse is a repository that draws data from many systems that will be shared by many users. A data mart contains data drawn from fewer systems that will be used by one or a few functional units. A multidimensional database provides different views of financial data: by product, channel, or time of sale, to name a few.