Prototype Databases: Try Before You Buy!

“I’M SORRY, I know you need this customer profile report, but it requires adding a new data source. The marketing database design specs were frozen in order to make the system delivery date. If we add this new requirement, the system will be delayed for two more months. You want to explain that to the boss?”

How many of us have heard these words? Most companies implementing a new marketing database will end up doing a major redesign, sometimes called Phase 2, within the first 12 months… and they’re the lucky ones. Many other firms can’t afford the budget or staff time to do a Phase 2, so their marketers have to live with the system as is.

Sometimes the problem is that the development schedule didn’t include sufficient time to collect end-user requirements and reconcile them into a comprehensive set of system requirements. But usually the marketing and implementation teams have an insufficient base of experience to draw upon. They don’t fully understand what type of data is available, how that data would be combined and updated, what new kinds of marketing programs the data would support, how they would go about putting those programs into effect and how best to track and evaluate results. In this situation, defining requirements is much more of a process than an event. The more that new capabilities are considered, the more opportunities are seen-and this never happens all at once.

In setting up marketing databases, there is usually a push by senior management to get the database running ASAP so the system can start generating a return on its investment. As a result, many companies fall victim to the old problem where there is not enough time to do it right the first time, but there is enough time to do it twice. Shifting to a two-step strategy where a prototype database is built before the final system can help alleviate a lot of these issues.

A prototype database is essentially a “throwaway” system that is built quickly, supports most of the required database functions, doesn’t have to be efficient and represents a short-term investment. A prototype provides immediate operational benefits while helping to build a base of knowledge to use in defining requirements for the final system.

Depending on the data that’s available, a prototype database can provide information to support strategic planning through use of demographic or purchase activity profiles, customer value, customer segmentation and the like. The system can also be used tactically to conduct promotions, track promotion history and results and do back-end response analysis.

A prototype offers many significant advantages:

* Fast start-up. Since the goal is to get a quick “70% solution,” not to implement a complete production system, the prototype does not need all of the same operational processes as the final system.

* Flexible implementation. Prototype databases use ad hoc processes to make them flexible, since efficiency is not a key issue. Therefore, it is much easier to include new sources of data, add newly calculated fields, process list selection instructions and produce reports.

* Hands-on experience with the data. Building the prototype gives a preview of the data available from all internal and external sources and how it can be combined. This involves investigating file-matching criteria (customer number, telephone number, name and address, etc.), determining how data fields from different sources should be combined or reconciled against each other and understanding how each field is coded and what the codes mean. During this process, each data field is profiled to review data formats, identify data errors and define options for resolving errors.

* Experience using a marketing database. Besides the production-oriented issues identified, end users in the marketing department will start gaining hands-on experience of a different nature. Once the database is built, the marketing staff will begin to understand what types of data are available. They will also gain considerable experience using the data effectively to run their business. Very often, the new range of possibilities will force a rethinking of how marketing does its job. Through a combination of planning and trial and error, new customer segments will be created, new list selection criteria will be put into effect, new reports will be created and marketing program success will be defined differently.

We have to remember that the speed and flexibility of implementing prototype databases comes at a price. These systems rely on a number of ad hoc processes, manual intervention and outside services. Prototype databases are often based on batch processing and having an experienced IT staff member write programs for each update step, list selection and requested report. This results in slower than desired turnaround times. Depending on the type of software used to set up the prototype database, the system may or may not support direct access by end users along with the ability to directly use marketing-oriented software tools. Finally, depending on the ease of gaining access to data sources, the prototype may not include all of the data desired by the marketing staff.