This section provides an overview of database management systems, in particular the relational database management systems (RDBMS) that are a component of most data warehousing systems, the technology that enables vast amounts of customer data to be stored, some of which will ultimately end up on the CSR's screen in a call center. This overview is not intended to be a definitive or detailed analysis of database technology; however, it does provide some selection criteria for and characteristics of this technology. The material is included in this book to illustrate its importance in the overall process of providing customer information to call centers.
Data mining, the process of extracting customer data from the data warehouse, is also reviewed and described, as is the importance of ensuring that only "clean" data are provided to the data warehouse.
Database management software is the technology that manages the data stored in the data warehouse and provides the tools for accessing and querying the data. In combination with the data warehouse, the repository of customer transaction data, this technology enables organizations to store, access, and manipulate customer data and to provide call center CSRs access to the data. (see Figure 2.11)
Figure 2.11: Data mining tools.
There are several viable database management systems used in data warehousing. However, as is typical of the IT sector, vendors often offer products that are in the final development stages and ready for first release. There are, therefore, usually implementation glitches and code that doesn't work in these products.
Determining database requirements is one of the critical areas of data warehousing, and the impact of their selection will filter down to the call center, one way or another. Organizations often tend to select a database with the rationale that it is the "company standard," because it is expedient and it eliminates the need for support IS staff to learn another database. However, the selection of database products should follow the same rigorous evaluation process as for any other IT product. Database management software should be selected on its own merits, that is, because it meets the objectives of the type of data warehouse to be implemented—operational data warehouse or informational data warehouse—and for its contribution to the corporate CRM strategy.
Most RDBMSs are based on on-line transaction processing. These products can handle operational data warehouses and have short but high transactional volumes, a response time requirement, and a very limited amount of historical data. These characteristics contrast very clearly with database requirements for the informational data warehouse, which has low transactional volumes, no real response-time requirement, and a large amount of historical data. The access characteristics of these two data warehouse environments are completely different. Database management systems need to differentiate between these two types of data warehouses, so it is important when selecting the RDBMS to be aware of its architecture for providing effective data access to either or even both data warehouse configurations.
Data mining and analytical tools, in combination with the data warehouse and database management technology, assist in increasing the return on investment (ROI) on stored customer data. In addition, they allow organizations to understand customer behavior patterns, rather than just grouping or segmenting them according to products they buy, age, or other personal characteristics, and highlight cross-selling opportunities and pinpoint the most profitable client profiles. These characteristics of the RDBMS are important to call center CSRs because they determine the ease of access and the usefulness of the customer data they will use in their day-to-day activities.
The information that can be gathered from the data warehouse and the RDBMS should form a ready source of customer data for the call center as well as provide information to marketing and salesforce automation programs. Conversely, customer information obtained in the call center should be continually fed back into the data repository. The more integrated the process, the closer the organization is to achieving one of the key objectives of a CRM strategy: a single view of the customer throughout the organization.
Standards are necessary for the data stored in the data warehouse—consistent formatting reduces complications for data extraction. Ensuring that the highest quality of data is provided at the input stage promotes acceptance of the data and develops a high degree of confidence in it. Many corporate CRM strategies are thwarted by faulty, inconsistent data that prevent users from having a clear, unified profile of each customer. Disjointed data, blanks in some of the critical fields, and broken business rules are a few of the ways in which data can be corrupted, resulting in data integrity problems.
Integrating customer datasets is a challenge for any organization that wants to achieve a single view of the customer. Various departments—call centers, ordering, shipping, manufacturing, sales, and marketing—have customer contact and therefore customer information to contribute to the database. In a typical financial institution or insurance company, for example, there could be 50 to 150 different systems containing customer data. To have a single view of each customer to establish value levels and to meet customer needs, this data must be combined and integrated. Combining and integrating data to obtain a complete, current customer profile requires assembling different data stores, with data of varying ages, on different databases, and usually involving multiple programming languages and data formats. Vendor software is available that can assist in assembling and profiling data, as well as analyzing data before it gets stored in the data warehouse. Typically, these products locate different relationships in customer data from multiple sources, irrespective of source code and documentation, and provide information on how to clean and restructure the data.
Cluster analysis is an exploratory data analysis tool that uses statistical algorithms to identify distinct groups of customers that may not traditionally group together. It is used in segmentation not only to independently validate business assumptions but also to discover new interrelationships between variables that were previously not associated. This technique may be useful in call centers that have an outbound call requirement that targets certain demographics in a customer population.