I had the opportunity to work on business process improvement efforts with a call center a few years back. The call center employees were feeling a great deal of stress due to expected response times and resolution of customer inquiries or complaints. Management collected a lot of data regarding hold times, the number of representatives available to answer calls and average call handling time. To improve these measures, management posted live updates on a board to promote a little competition (Figure 1).

Figure 1: Call Center Scorecard

The competition did work to an extent. Agents did their best to answer calls and get the customer off the phone as fast as they could. Hold times and call duration were dramatically decreased. The catch was that many of the calls were ended without resolutions. Anybody associated with a call center would know that Resolution is the  key customer satisfaction area for any call center. In addition, call volumes began to increase dramatically. CDR data eventually showed that often the same caller was calling back multiple times because of non resolution, thus increasing call volume. The manager was planning no data driven resolution, as almost always. This is when i was called in.

My first observation was - they were not collecting data relative to the company goals of resolving 75% of the inquiries the first time around (first-call resolution). In addition, they gathered data on the callers who were unable to get answers to their questions. The team then incorporated Lean and Six Sigma to focus on the goals associated with improving performance.

Call Center DMAIC Findings

Defining the customer, their critical to quality (CTQ) issues and the core business process involved:

  • Defining who the customers are, what their requirements are for products and services, and what their expectations are.
  • Defining project boundaries, including the stop and start of the process.
  • Defining the process to be improved by mapping the process flow.

Common sense dictated that call center agents must be available to answer calls to keep hold times down; however, voice-of-customer (VOC) data revealed that providing issue resolution or answering questions was, by far, more important than hold time. Although Team A and Team B utilised the same process, process maps for both teams were created in hopes of finding a quick win, given that the data collected showed Team B outperformed Team A. Early analysis proved this was merely due to call volume because Team A serviced more heavily populated areas.

Measuring the performance of the core business process involved:

  • Developing a data collection plan for the process.
  • Collecting data from various sources to determine types of defects and metrics.
  • Comparing performance to c-sat results to determine shortfall.

Baseline performance had previously been measured on hold, call abandonment and duration times. With the emphasis now placed on call resolution time, we established the following baselines: First-call resolution = 35 percent and three-day resolution = 45 percent. Specification limits were validated and communicated. In preparation for Analyse, the team gathered additional data, including call volume and cost per call. This data was stratified by region, employee, month, time of day, and number of callers who had to make multiple calls for the same issue or question.

While Analysing the data collected and the process map to determine root causes of defects and opportunities for improvement, we:

  • Identified gaps between current performance and goal performance.
  • Prioritized opportunities to improve.
  • Identified sources of variation.

During the Analyse phase, we learned:

  • The 30 agents, divided between two regions, were handling different call types. Although they followed the same procedures, there was variation across all 30 agents – each agent did things differently. There was variability within the agents themselves because many handled the same call type differently each time. Thus, the call center did not have one process to improve – it had hundreds, just for one simple call type.
  • Although all agents were thought to handle multiple call types, certain agents were more proficient at or enjoyed certain call types more than others. Internally, agents were directing specific types of calls to people within their region. The key finding was that there was, in fact, a need for cross-training.
  • Most of the calls could not be resolved to the customer’s satisfaction on the first call and required additional research or escalation to Tier 2 and 3.
  • Call center agents performance was measured on their availability to answer the phones, not answering the questions. Obviously, this decreased the time they could devote to research issues. The outcome was that most calls that could not be resolved right away.
  • Customers whose inquiries were not answered would call back. This increased the call volume, inflated the number of calls that could not be resolved on the first call and led to multiple entries in the data-tracking application for the same inquiry.

Improving the target process by designing creative solutions to fix and prevent problems involved:

  • Creating innovate solutions using technology and discipline.
  • Developing and deploying an implementation plan.

In order to achieve company goals, the following solutions were agreed upon:

  • Divide the team into separate functions: answering the call and resolving the issues
  • Rotate call center agents between the separate functions in order to transfer knowledge
  • Assign any calls that were not resolved within two days to management for action

To reduce variation, call center agents needed to have a process to follow (standard operating procedures). The agents should follow that exact process because it allows them to deliver quality service and because it makes their work life easier.

However, standardising the process did not mean the process was optimised. We eliminated several kinds of waste in the call center: waste from having the people waiting around, waste from excessive steps or movement of call, and waste from rework.

Controlling the improvements to keep the process on the new course involved:

  • Preventing agents from going back to the “old way.”
  • Development, documentation and implementation of an ongoing monitoring plan.
  • Institutionalizing the improvements through the modification of systems and structures (staffing, training and incentives).

In less than eight weeks, baseline goals rose from 35 percent for first-call resolution to 80 percent and from 45 percent for escalations to 90 percent. In addition, calls that never received an answer were all but eliminated. Figure 2 provides an example of the revised scoreboard displayed for call agents to better monitor progress toward goals.

Figure 2: Revised Call Center Scoreboard

The benefits of applying Lean and Six Sigma process improvement methodologies to call center processes can be rewarding. Needless to say, customer delight rose significantly.