Data mart designers are often asked whether the data mart will be for decision support or "dynamic" in nature. In this context, dynamic means that the data mart has several attributes (e.g., frequent updates, recalculation of data fields) which are thought to conflict with decision support requirements. The reality, however, is that the requirements of the data mart include dynamic characteristics if the business process supported by the data mart is fully considered. Therefore, the real decision to be considered is how to support this requirement. To fully explore this issue, let's consider a marketing data mart for a telecommunications firm. At first glance, the requirements for the telecommunications marketing data mart include ad hoc decision support, analytic capabilities, list extraction capabilities (e.g., pull lists of names for direct mail) and reporting, to name a few. The resulting solution would provide marketing users with the ability to analyze customer trends, target customers using expected response/value models, pull the names for the mail shop or telemarketing operation and analyze results on a campaign or customer segmentation basis. At this point, all the elements seem to be integrated in a solution that supports the direct marketing business process.
If we investigate further, however, we learn that nothing can be further from the truth. To demonstrate, let's focus on just one component of the business process--the analysis step. This step of the process provides the marketing user and organization with marketing promotion results information. To effectively provide this, the system needs to have all promotion-related information. At the most basic level, it must provide the answer to the question: Did person X respond to promotion Y? This requires the data to provide knowledge on who received promotion Y and what purchases or transactions are associated with person X. To provide this information in the database, dynamic characteristics must be supported!
First, the system must have the capability to "backfeed" promotional information into the data mart. This requires the feeding of the mailing list back into the data mart so the data mart can flag the customer as having received the promotion. Furthermore, these flags, often need to be updated later since additional decision criteria are added to the list by the mail house or downstream system prior to the customer actually receiving the promotion. In these cases, an updated list of customers who either passed or didn't pass this downstream filter must also be fed back into the data mart.
If these requirements are fulfilled, then all transactions coming into the data mart need to be checked against this promotion information to provide response indicators on a customer-by-customer basis. Promotion response can be indicated by the purchase of a product (e.g., a cell phone) or a change in the customer's transaction behavior (e.g., a 20 percent increase in long distance calling). As a result, all transactions coming into the data mart need to be checked against the appropriate response logic so the response indicators can be set during the update process.
Both the flagging of customers and the response attribution process performed during the data mart update need to be somewhat real time. Feeding transactions to the data mart which may have occurred as a result of a promotion without the aforementioned promotion flagging will result in inaccurate promotion and customer measurement. Although the real-time aspect needs to be considered as part of the overall requirements and business process, increasing use of the Internet and direct links to telemarketing and customer service systems require that data be updated more often than the monthly or weekly expectations set for early adopters of decision support systems.
If we explore all other aspects of the system, we will find many similar areas that require dynamic characteristics. Although some innovative solutions have been found in the past, most projects ultimately provide support for some, if not all, of these dynamic characteristics as part of the design and development of the data mart or data warehouse.
Several key points must be addressed early on in the planning of a data warehouse or data mart. First, the business process to be supported by the data mart needs to be well understood at both a strategic and detail level. Second, the requirements of the data mart need to be reconciled with the business process. Third, the relevancy of information and the interdependencies of the data warehouse/data mart components need to be considered in their entirety. A breakdown in any of these three areas will not only result in a less than optimal technical solution but the business value promised from the resulting capability will not be returned.
Register or login for access to this item and much more
All Information Management content is archived after seven days.
Community members receive:
- All recent and archived articles
- Conference offers and updates
- A full menu of enewsletter options
- Web seminars, white papers, ebooks
Already have an account? Log In
Don't have an account? Register for Free Unlimited Access