Ounjian is using Oracle databases on the front end to reassemble and synchronize back-end data so that consumers find consistent, timely information regardless of whether they use the Web channel or the call center channel. Ounjian says he's using different vendors for the various components of the Web self-service architecture because there's no one vendor that supplies everything one needs for an integrated CRM system. He wanted the flexibility to layer different applications and functionality from different vendors on top of his infrastructure.

Making Data Sense

Ounjian's biggest headache was devising a tactical strategy for moving data and transactions from the front end to the back end and vice versa. Making millions of bytes of back-end data available and understandable to users on the front end is one of the biggest challenges for any successful CRM project, regardless of industry. If you can't get accurate information to your customers in a format that they understand, they won't use your system. The number of records (100 million) that Ounjian and his staff had to migrate made the task even more daunting.

Explanations of benefits, for example, were stored as codes that were meaningless to consumers. Those codes needed to be put in terms end users could understand. Ounjian says his staff built a data dictionary for all the codes that included their equivalent English definitions. It essentially matches up raw back-end data with the appropriate translation for consumers on the front end.

In the past, when mail the insurer had sent to customers was returned because the address it had in its files was no longer correct, BCBS of Minnesota actually entered "bad address" into the address files on its legacy systems and would fill the ZIP code field with 9s to denote that the address was no longer correct. But now that its back-end data is customer-facing with the new customer self-service system, BCBS of Minnesota can't risk having a customer see "bad address" or 99999 as his ZIP code when he pulls up his file. So these days, whenever the company has mail returned, it immediately corrects the address in its legacy systems.

In addition, ZIP codes stored in back-end databases had been compressed from nine digits to eight in order to save money on storage. Ounjian and his staff had to expand those ZIP codes back into their nine-digit form for consumers on the front end.