Why householding is important in an MCIF.

• Author: Coffey, John J.
• Position: Database Marketing - Brief Article

The account-level data in your bank's processing systems can be compared to the pieces of a jigsaw puzzle in an unopened box. The pieces aren't arranged to make any sense yet. To make matters worse, obsolete fields and degraded data have put a layer of dirt and grime on the pieces, making the puzzle harder to assemble.

Clearly, there is a need to clean your bank's account-level data and consolidate it so you can know your customers well enough to present relevant product offers. Enter your MCIF! Your MCIF contains specialized programming called "matching algorithms" to help you automatically and accurately clean and assemble the pieces of the jigsaw puzzle so you can see the complete picture.

The matching algorithms in your MCIF produce at least two levels of consolidated data. The first level is the "individual" (or customer) level. For instance, Mr. Smith may have a checking account and a mortgage loan. Using the matching algorithms, Mr. Smith's accounts are linked together. The second level is the "household" level. In this case, Mr. Smith is married to Mrs. Smith, who has a credit card account at your bank. Using the matching algorithms, Mr. Smith's accounts and Mrs. Smith's accounts are linked.

How is householding accomplished?

On the surface, this householding process sounds simple to accomplish. However there are some obstacles. Most banks use multiple systems to process their customers' data. Because multiple systems are used, the data "feeds" use multiple formats and data layouts. The data format for each feed needs to be standardized, and the data layout from each feed needs to be mapped into a master file layout. Only after this has been accomplished can the matching algorithms be employed.

The matching algorithms in some MCIFs simply match the customer's last name, address, Social Security number, or any other linked field to develop a unique "household match key." The problem with this approach is that for it to work, the data in the customer's accounts needs to be an exact match. If a customer has two accounts and one account has his address listed as "123 Elm Street" and the other account has his address listed as "123 W. Elm St.," then the customer's two accounts will not match.

This...