Customer retention is an important business goal. Corporations design systems, purchase hardware and software, convert and upgrade applications, and deploy new telecom technology. They hire and train more staff, and they expand their managerial staff, too. All this in pursuit of better retention.

Sometimes, even with gigantic investments of time, money, and staff, improvements in retention prove elusive. Barring competitive issues, if we assume that the customer became a customer because she desired the value received, the retention process is not operating within design parameters or the design is flawed. How do we get to the truth so changes that result in improved customer retention can be made?

Some of the tools we use to evaluate processes are value mapping, face-to-face interviews, direct observation, inside and outside surveys, tracer objects, flowcharting, and simulation. Depending on the size and complexity of the process, the number of outsourced sub-processes, and the organizational span of the process, we may use any one or a combination of these tools to find truth. In some cases, we use the relatively new science of process mining in our discovery mode to diagnose automated processes, such as those defined and managed through automated workflow systems.

Process mining is a form of data mining applied to a specialized data set. Whereas data mining finds patterns in generic data, process mining finds patterns in data specific to processes. We can't identify market-related issues or product-acceptance issues with process mining, but we can tell if the activity in question is operating within specification or not. In cases where no formal design exists, process mining can be used to reverse engineer the design so management teams can consider changes and their potential impact on the activity in question and/or upstream and downstream processes.

For example, are customers receiving their policy before the start date? Why is it taking 45 days to underwrite a case when we hired up to achieve a 15-day turnaround? Where are the bottlenecks with our outsourced service providers? What is actually happening in our process compared to the original design document? Where is the best place to add a retention activity in our existing process?

Process mining is directed at finding unknown patterns from the data that are useful in decision-making. The data comes from the log files captured by Web sites, automated workflow systems, telephone systems, inquiry management systems, transaction capture, and other systems. If log files are not kept, tracer objects may be submitted with some form of data-capture mechanism attached. The results from these tracer items are used to create a pseudo log file that can serve as process mining input.

Most transaction systems use their log files to diagnose performance problems or to help rewind the system to a certain point in time after a system failure. Some log files are used by auditors when reviewing control or quality issues. Then, the log data is used to trace single instances of an activity to a single point in time or place in the organization. Process mining works differently: it uses log file data in a relational context by grouping similar objects by operator, activity, or location along a time dimension. From this, statistics are gathered, and inferences about the entire process can be made.

Process mining can differentiate types of activity by a specific identifier or by the activity's behavior compared to all other activities. For example, a request from a customer for a policy reprint might have a different routing than the same request from a broker. Process mining can also define social networks within an activity—who works together, who hands off work to whom, and so on. We can use this information to help guide interviewing and to determine the real-life activity within the process. Comparing this to the design or requirements of the process often reveals points for improvement.

In our business, results count more than concepts, and performance counts more than presentation. If you have a process that defies explanation or is not responding to management input, we can help uncover problems and implement corrective action. Is process mining the right tool for your company? We can help you find out and along the way, perhaps, find those change points in your process that will really improve your customer retention.