West Corporation

Posted on April 10, 2013 by West Corporation 



What’s Too High? A Six Sigma Approach to Caller Behavior Analysis

Is 95 percent transfer rate too high? Is 10 minutes in IVR to authenticate too long? Anyone in the IVR business would respond with a resounding, “Yes!” But what about a 60 percent transfer rate? Or 20 seconds to authenticate? The idea is that it’s easy to use business experience to judge the obvious extremes. It is not so easy when the numbers are in the gray area. The outliers can hide just close enough to “normal” to go undetected by the human eye, yet they can be far enough away to cause a financial impact.

Luckily there is a solution: Six Sigma. This well-known technique statistically defines exactly what is normal and to identifies outliers falling outside of the normal range.

The idea behind Six Sigma is to track a particular metric (e.g., number of calls made by a customer) over time to generate a distribution or histogram of its acceptable values (those that are close to the average) and unacceptable values. (those beyond a certain distance from the average). Sigma, or σ, is a symbol used by statisticians to denote this distance and is also known as standard deviation. Six Sigma says that more than two-thirds of the values of a particular metric would fall within one standard deviation from the average, while nearly all values would fall within three standard deviations (a total of six) of the average. Anything outside of that is a serious outlier.

West uses this approach to establish normal boundaries for repeat calls made by a single customer in a particular month. This study was important because our clients’ customers make upward of 600,000 calls per day, and understanding repeat callers is the key to decreasing such calls. The resulting distribution was not exactly text-book “bell curve” due to the nature of data; however, it did turn out that roughly 10 percent of callers were outside of normal.

Zeroing in on particular extreme behaviors, certain questionable customer practices were identified. For example, one customer who made 200 calls per month (with an average customer making three calls during the same time frame), turned out to be a small-business owner who used his familiarity with IVR and agent negotiation to get his clients (other customers) discounts for a cool $30 per service, i.e., per call.

Leave a Reply

Your email address will not be published.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