Database Design is Difficult

Customer intelligence

Information Management Magazine, January 2008

Larry Goldman

I regularly hear the complaint, “I can’t get basic information” from the business side - particularly from sales and marketing individuals. Often, the phrase is distorted. Sales departments typically have access but haven’t spent the time to learn how to download the report or use the cube. Often, they have forgotten their training, don’t remember where or how to access the reports and have given up. Marketing departments can usually get the basic information, but “simple” or “basic” is often defined differently between the customer intelligence team and marketing.

Customer analysis is very difficult. Many relational databases and database designs are not able to handle some of the difficult concepts supported by customer analysis. Time series, product correlation and the nonexistence/existence of activities make analysis very difficult. Many times these requirements come across as ad hoc requirements that may be ignored in order to optimize for scheduled management reports.

To avoid being blindsided by these requirements and requests after implementation, those responsible for marketing databases should consider the following types of analysis:

• Multiproduct or service relationship to the customer.
• Identifying specific events that have occurred over time.
• The nonexistence of events, transactions or behavior within the database.

Product Correlations

Though all sales and marketing databases are customer-centric at their core, the first question from a customer analyst is, “What did they buy?” This simple question is usually easy to address, as is the management report that shows sales, quantities and profitability broken down by product category or division. The more difficult questions include:

• How many product categories do specific customers use or purchase?
• Which product categories do specific customers use or purchase?
• What combination of product categories do specific customers use or purchase?
• How many customers use a specific combination of products?
• What type of customers use specific combinations of products?

Many organizations focus on the breadth of their product lines. Typically, there is a huge opportunity to drive more usage across product lines rather than push single products to individual customers. This is true for telecommunications companies who want their customers to use landline, wireless and DSL or technology companies who want to sell computers, printers, modems and other accessories.

Time-series events and time series help us understand how fast customers adopt or respond to new marketing pitches. Common questions include:

• How fast did the customer get to a certain revenue level?
• How long did it take the customer to buy their second product?
• How long has it been since the customer’s last purchase?
• What is the average time between purchases?
• After introducing a marketing program into the field, how long did it take customers to adopt?

Page 1 of 2.