Better Customer Treatment Decisions

I once heard customer relationship management (CRM) described as a "vertizontal," meaning CRM has some of the characteristics of a horizontal technology, like databases, and some of the characteristics of a vertical, like health care. Over the years, the words have changed. Customer experience management (CEM) is in vogue, while plain old CRM has gone somewhat out of style. Characteristics of the market have not changed - it is still a vertizontal. So, this month I discuss smarter decisions in the context of customer service, customer experience and customer relationships.

Customer experience is made up of lots of very concrete interactions, such as visits to a store or Web site, calls to a call center and mail or email received. The cumulative effect of these interactions is to create an overall experience. A key measure of success in those interactions is what got done and how quickly. It is not the only measure, but it is really hard for a company to deliver great experience in those interactions unless customers actually do what they wanted to get done.

We must consider that the customer experience is now multichannel - most customers use more than one channel at least some of the time - and that a channel is no longer a good predictor of the kind of customer we have. It used to be that good customers always came to the branch while less good customers chose to use the Web or ATMs. Now that linkage is broken. Each channel must handle the full range of customers - from profitable to loss-making - and all channels must contribute to the development and engagement of our customers.

Five characteristics of good interactions come to mind in this environment - personalization, self-service, first-call resolution, loyalty and best next action. Better customer treatment decisions must address all five if they are to add value to the overall customer experience.

Past attempts at personalization tended to be shallow, personalizing each channel based only on what happened in that channel, and either automated or passive - either the company did all the personalization (think Amazon.com recommendations) or the person did everything (think setting options and selecting preferences from a long list). What's needed now is a focus on decisions about what to display or offer to a customer so that each can be personalized. These personalization decisions are applied to every channel and use rules from the customer and from the company, analytics based on data from many channels and enterprise applications, and constant randomized testing of novel approaches - adaptive control.

Self-service is a growing phenomenon. Companies often start with self-service as a way to cut costs, but increasingly, customers see it as a benefit. Instead of contacting a call center or visiting a store or branch, customers want to go online and serve themselves. The days when passive presentation of information was enough for a self-service application are long gone. Customers expect to be able to do everything online. This forces companies to think about all the approval decisions, which used to be referred to a person, because now these must be automated. Add to that the need to market to and up sell customers using these self-service applications, and the number of decisions that must be automated climbs rapidly. Without decision automation, these self-service applications will not add value for customers or do much more than reduce a few costs for companies.

One important measure in call centers is first-call resolution - how often a customer gets his or her needs met the first time he or she calls in with no callbacks, no transfers to another number and no delays. Delivering on this requires many things, not least of which is the elimination of needless transfers for approval. The approval relies on information already collected, so automate it and let the first person who speaks with a customer take action. This is a critical step in boosting first-call resolution, and approvals are driven by regulations, policies and risk. Business rules and predictive analytics models are by far the most effective way to implement approval decisions as a result.

Maintaining and increasing customer loyalty is a must-do item, especially in a tightening economy. Loyalty, like self-service and personalization, is a cross-channel phenomenon and must be managed across the company. Understanding which customers are loyal, which might become loyal and which will not be loyal no matter what requires analytics. Predicting who will be positively impacted by which actions also takes analytics. The rules of a loyalty program must be explicit and applied consistently across systems and business processes. Novel approaches should be tried periodically to make sure that there are not simple changes to the program that would boost results. Encapsulating all this into loyalty program decisions - how much a given action earns and what offers to make - is the only way to deliver a modern loyalty program.

The end result of all these steps is a drive toward delivering the best next action each time a customer interacts with a company. A company must understand its customers, predict their needs and likely future actions as best it can, decide what the best action is to take with each customer the next time it has an opportunity and deliver this best next action through whichever channel the customer uses. A singular cross-channel focus on the customer treatment decisions involved is needed. Treating decisions about customers as corporate actions is critical.