Embedding Intelligence

In a way, this is the second part of a piece on non-reporting tool BI.

Consider a customer contact centre - historically the realm of CRM applications. Customers interact with agents by voice telephony, text messaging, email, web portals, in writing, and there may even be face-to-face contact as in store or banks. Much of this communication is instigated by the customer but some is the result of call centre agents making contact. Integration of IP telephony, document management and plain old CRM (if it was ever plain) into a single system is achievable and gives many advantages (especially in these days of SOx compliance)

But just bringing back the customer's details or recent interaction history is a standard CRM feature so where does embed BI fit into the picture and how is it achieved?

One obvious and non-invasive way is to build reporting dashboards over the whole customer interaction application stack. Marketing managers can asses the effectiveness of campaigns, operation managers can monitor call centre agent throughput; in fact almost any performance metric can be captured and displayed - the art here is the choice of metrics and techniques needed to find the right data.

The less obvious technique is to embed intelligence services into the CRM workflow. These services either alert agents to particular conditions such as "be extra nice, this is an important customer" or, perhaps more sophisticatedly, change the workflow for a case based on a set of BI derived attributes. Here the trick is use fast metrics (or suitable aggregates) that provide some sort of predictive measure about a customer. For example the simplistic value of a customer might be their spend in the year, but at a more sophisticated level we would want to take in to account the cost of servicing that customer over the year (man hours spent by contact centre staff?) or we might want to rank them in some form of peer group.