2004 Feb 01
Real-Time Customer Views
David M. Raab
DM Review
February, 2004

In the world of marketing systems, no cliche is more popular than the “360 degree view of the customer”. This refers to assembling a complete set of data from all systems that record customer interactions. The underlying assumption is this comprehensive picture will allow highly tailored sales and service treatments that ultimately return higher profits.

Of course, reality is a bit more complicated: all data are not equally valuable or equally accessible. So determining exactly which data to gather requires balancing business, technical and political factors. The result is something less than a complete view of the customer, although hopefully still broad enough to be useful.

Traditionally this merged data has been placed in a central repository such as a data warehouse, where it is used for analysis and reporting. Feedback to operational systems is mostly through batch transfers such as lists of customers with segment codes or recommended marketing treatments. When fresher data is needed–in particular to react in real time or near real time to customer activities–a layer like an operational data store is added to consolidate new transactions as they occur and, in the most advanced systems, to perform analytics and select appropriate reactions.

Yet these real time layers are still basically appendages of the underlying data warehouse. Given the difficulty of merging data from different sources, it makes sense for the real time system to leverage the consolidation functions already built for the warehouse, rather than recreating them separately.

But not every firm has a data warehouse and not every warehouse can be effectively adapted to support real time interactions. At the same time, providing operational systems with a shared (if not necessarily complete) customer view often has tangible benefits that companies are not willing to forego. As a result, many firms find themselves looking for a solution that produces this shared view without the foundation of a traditional data warehouse. Naturally, where such a demand exists, software vendors will follow.

Companies that focus specifically on providing a real-time view of current, consolidated customer data include DWL, Journee, Nimaya and Siperian. Since sharing customer data is technically similar sharing other types of data, these firms also compete with less specialized data integration and synchronization vendors like GoldenGate, MetaMatrix, Ascential and DataMirror.

The general approach of the customer data sharing systems is to present a single resource that operational systems can access when looking for customer data. This resource may be a physical data store (DWL and Siperian) or it may be a data model that is mapped directly to the actual source systems (Nimaya and Journee).

The obvious advantage of building a physical record is that it provides quick, consistent access times: the data is already assembled in one place, so there are no problems with slow or inconsistent response times from the original sources. This approach also makes it easier to deploy sophisticated matching and reconciliation schemes, since these processes take place while data is being loaded in the background rather than when an operational system is impatiently waiting for a response. Updating the customer record whenever source data changes also lets these systems generate alerts or kick off business processes in response to specified events. This allows proactive customer treatment rules to be built directly into the data sharing system rather than requiring a separate process that scans for significant changes.

The mapping approach, which lets operational systems read customer data directly from source systems, has its own advantages. Reading the source systems clearly ensures that the most current data is presented. This is particularly important when managing real time interactions, where knowledge of the customer’s latest activity is essential. The on-demand approach may also reduce the total amount of processing, since source data is extracted only when needed. Thus information that changes frequently but is used rarely–say, a bank account balance–is posted much less often. Still, the actual benefit will depend on the situation: it’s possible to imagine cases where a system makes repeated queries against data that hasn’t changed, and actually increases the total processing level. As a practical matter, both Journee and Nimaya do include options to store customer data internally, providing an alternative when on-demand access is inappropriate.

Which architecture makes the most sense will naturally depend on the circumstances. The fundamental advantage of these products–making it easier for multiple systems to access unified customer data–is the same regardless. Compared with general purpose data integration tools, the specialized products do provide some features aimed at the particular problems of managing customer data, such as selecting among conflicting values for the same element and maintaining hierarchies of relationships among individuals, households and businesses. Perhaps surprisingly, none of these systems appear to have their own fuzzy matching engine to identify related customer records from different systems. It seems this is still a separate specialty.

The long-term prospects of specialized customer data sharing systems is unclear. Buyers may find their features worthwhile, or may choose to apply general purpose data integration products, or may even make one operational system the primary customer data repository. But for companies with specific requirements in the immediate future, these products provide an option that’s worth a look.

* * *

David M. Raab is a Principal at Raab Associates Inc., a consultancy specializing in marketing technology and analytics. He can be reached at draab@raabassociates.com.

Leave a Reply

You must be logged in to post a comment.