By David Loshin

In the past few entries in this series we have basically been looking at an approach to understanding customer behavior at particular contextual interactions that are informed by information pulled from customer profiles.

But if the focal point is the knowledge from the profile that influences behavior, you must be able to recognize the individual, rapidly access that individual’s profile, and then feed the data from the profile into the right analytical models that can help increase value.

The biggest issue is the natural variance in customer data collected at different touch points in different processes for different business functions. A search for the exact representation provided may not always result in a match, and at worse, may lead to the creation of a new record for the same individual, even one or potentially more records already exist.

In the best scenario, the ability to rapidly access the customer’s profile is enabled through the combination of smart matching routines that are tolerant to some variance along with the creation of a master index.

That master index contains the right amount of identifying information about customers to be able to link two similar records together when they can be determined to represent the same individual while differentiating records that do not represent the same individual.

Once the right record is found in the index, a pointer can be followed to the data warehouse that contains the customer profile information.

This approach is often called master data management (MDM), and the technology behind it is called identity resolution. Despite the relative newness of MDM, much of the capability has been available for many years in data quality and data cleansing tools, particularly those suited to customer data integration for direct marketing, mergers, acquisitions, data warehousing, and other cross-enterprise consolidation.

In other words, customer profiles and integrated analytics builds on a level of master data competency that is likely to already be established within the organization.





Leave a Reply

Your email address will not be published. Required fields are marked *