A young, rapidly expanding financial services company located in India runs an in-house help desk for responding to customer queries via email or telephone. Help desk response time is tracked using software that calculates the time between registering and closing calls. A project was instituted to train employees in the relevant tools and techniques of Six Sigma and Lean methodologies in order to achieve dramatic improvements in the help desk operations.

The case study demonstrates the complementary nature of Lean and Six Sigma methods and tools, and that a strategic combined application of both methods is more effective than applying either method in isolation.

The Six Sigma DMAIC methodology was structured into seven implementation steps.

Step 1 – Define and Measure
Step 2 – Analyze
Step 3 – Improve/Generate and test countermeasure ideas
Step 4 – Improve/Implement the ideas
Step 5 – Improve/Check the results
Step 6 – Control/Standardize results. Grind practices in
Step 7 – Control/Record the improvement story

Step 1. Defining the Problem

Theme: The company’s senior management team brainstormed more than 25 problems. Using the weighted average tool, the team agreed on a theme of “Build Quality Capacity.” Within this theme, the critical-to-quality characteristic selected was “Improvement in Customer Service.”

Project Selection: One of the projects selected called for a dramatic reduction in the response time of the customer care help desk. A cross-functional team was selected to define the problem as the gap between customer wants and actual results.

The voice of the customer was characterized into three major elements:

  • Closing calls could mean resolving the customer issue or just responding to the customer query with a promise to follow up appropriately.
  • Different types of calls require varying amounts of time to close and have different service standards.
  • The help desk is dependent upon other departments which may or may not cooperate in closing calls promptly.

Call Response Versus Resolution: After comparing the two ideas about closing calls on a 10-point scale, a discussion concluded that resolution was more important than response. However, resolution involved departments that were external to the group and were not involved in the project. The team therefore opted to address call response time in Phase 1 of the project. Events later demonstrated that this approach was valid as back office problems were found to be the root causes of some of the delays and were resolved within the project scope.

Different Call Types: Three call categories with specific service standards were identified as Category A (less than two days), Category B (less than five days) and Category C (less than seven days).

Call Categories
Call Categories

At this point, the team was introduced to Six Sigma concepts, and the Pareto chart to begin to prioritize and decide the order of attack:

Category A constituted 98 percent of calls and was chosen as the scope of this project. Response time data for Category A for a one-month period was analyzed and defined the problem as follows:
Current status: Average + 6 sigma = 12.7 days
Desired status: Less than 2 days

The problem therefore was to reduce response time by 10.7 days, or 85 percent.

Step 2. Analysis of Root Causes…Why?

The discussion was started by asking what information customers want from the help desk. Customer requirements were summarized as follows:

  • When will something get done or will I receive something?
  • What has happened? What is the problem?
  • When will the problem be resolved?

Generally a customer was satisfied if the call desk could tell him when a problem would be solved or when the help desk could respond with an answer – provided this time commitment was met. The customer’s overwhelming need was “when,” rather than “what.”

The team was asked: If the answer to “when” was available, how long should call closure take? Individual opinions ranged from 3 to 60 minutes. A consensus closure time of 20 minutes emerged if information was available from the help desk.

The answer to the question of “why” was the average + 6 sigma equal to 12 days and was hypothesized as: “Information is not available with us; we depend on others who respond at will.”

Step 3. Generate and Test Countermeasure Ideas

Lean methods were invoked to cut the waiting time. Response time in Lean methods based on the Queuing Theory consisted of three major elements:

  • Arrival pattern of inputs (customer calls)
  • Waiting time (for information)
  • Process time (time to respond)

Lean clearly recognizes that waiting time has a significant impact on cycle time (response, in this case) and leads to the following method of reducing it:

  1. Map existing process (information mapping)
  2. Reduce stages required (create ideal process map)
  3. Change processes from batch to flow (process change)
  4. Good housekeeping (reduce inventory)
    – Spring clean once/twice (sweep the backlog of difficult calls)
    – Clean regularly thereafter (sweep difficult calls regularly)

The current process map was a confusing tangle of lines of communication between the customer, the help desk, the customer care supervisor, operating personnel who could solve customer problems (doers) and even the customer care supervisor’s boss. In fact, the map noted that one month the customer care supervisor’s boss had to close 425 calls himself, or about 15 percent of the total calls during that period. In striking contrast, a map of the ideal process involved only the customer, a help desk person and the doer.

In developing the ideal process map, the team identified two basic customer query metrics:

  • Proportion of queries the call desk handle can handle independently
  • Confidence in the resolution of calls to a standard time

All calls received during one month were categorized depending on who closed them:
Category 1 – Information available from help desk (67% of calls)
Category 2 – Information/action by operating personnel (33% of calls)

The following plan for improvement was tested with call desk personnel for Category 1 calls:

  1. Measure average + 6 sigma response time
  2. Select two days – employees sweep backlog of difficult calls
  3. Try answering calls in the queue on a first-in-first-out approach
  4. Sweep difficult calls twice daily at fixed times
  5. Analyze root causes daily of calls taking more than two days
  6. Assign Category 2 calls to relevant operating personnel in flow

Step 4. Implement the Countermeasures

The above plan was implemented.

Step 5. Check the Result

The chronology of progress was as follows:
Week 1 and 2 – Hesitant start/system cleaned
Week 3 and 4 – Dramatic change (66 percent reduction)
Week 5 and 6 – Six Sigma results achieved (85 percent reduction)

The team’s feelings: “It all looks so easy now.”

Step 6. Standardize the Result

An x-bar control chart was introduced with daily root cause analysis and killing out the worst three calls of the previous day. The average + 6 sigma continued to fall.

For calls directly responded to by the call desk:
Sub-category 1 – Average + 6 sigma = 0.58 days
Sub-category 2 – Average + 6 sigma = 1.52 days

Energized by this success, the team took on the much more difficult task of improving the response of Category 2 calls.

From Response to Resolution: There were four sub-categories in Category 2. They were named 21, 22, 23 and 24. The process of Step 3 was repeated for each sub-category involving the relevant operating section head.

Sub-categories 21 and 22 went through a similar improvement cycle as Category 1, that is average + 6 sigma reduced to less than two days within six weeks. The objective of average + 6 sigma of all calls with service standard of less than two days for resolution/response had been achieved.

Sub-category 23 had a response expectation of less than five days. Eighty percent of their calls using the Step 3 method rapidly came within the two-day limit. The overall average + 6 sigma was reduced from 12 to 4.5 days. Further reductions were not possible because they had to depend upon documents received from external agencies over which they had little control.

Sub-category 24, with a seven-day service standard expected, was particularly interesting. It involved a supply chain of two sections within the company, an external agency and the movement of physical documents and information. The process was mapped. From the existing process, the handovers and operations required to do the job were reduced by 33 percent. This led to a productivity gain of 30 percent, and the turnaround reduced for average + 6 sigma of 12 days to 5.8 days – less than the expected service standard of seven days.

Thus a 6-sigma performance for all calls in both response and resolution had been achieved.

Step 7. Record the Improvement Story

A quality improvement report was prepared and presented to management. The report confirmed the truth of the common assertion that improving the quality of a product or service, in this case response to customer queries, reduces costs. With the average + 6 sigma of customer response time reduced 85 percent, the same staff was able to handle 250 calls per day compared to 103 before the project started. That represents an increased productivity of staff and assets of more than 140 percent.

About the Author