A common problem encountered in Lean, Six Sigma and TQM implementation is performance deterioration after the completion and handover of a breakthrough improvement. This can happen despite installation of control mechanisms to monitor ongoing performance.

In this case study, a financial services company with a national sales network had already begun addressing concerns about its level of internal customer service. The company, which services the IT problems of internal users through a centralized help desk, had dramatically reduced help desk response times in Phase 1 of its improvement project. Response time had dropped 85 percent in the most frequently recurring call category. A control chart showed that 99.7 percent of these calls (average +3 sigma) were “closed” within the 30-minute service standard. But the issue of sustaining the changes suddenly was complicated by the simultaneous resignation or transfer of key team members.

The improvement project therefore continued with two objectives:

  • Train new members in improvement methodologies
  • Build a framework for sustaining the change

Sustaining Improvement: The Problem

Converting individual project successes into sustained performance by operating personnel is critical to the development of any change initiative. This requires creating a quality mindset – a far more challenging and time-consuming task than achieving the breakthrough improvement.

What is a quality mindset? Using firefight as an example, the typical response to a fire is to fight it until it is extinguished. Then be ready to fight the next fire when it happens. Having a quality mindset also means responding to a fire by fighting it until it is extinguished. However, instead of waiting until the next fire, a quality mindset means finding the root cause of the first fire and working to prevent any other fires stemming from that cause.

Companies with a quality mindset recognize “If you do not improve, you deteriorate” and demand sustained, continuous improvement efforts by operating teams. In institutionalizing this approach, everyone involved has a role to play:

  • Senior management – Regular critical-to-quality (CTQ) trend reviews with performance rewards/penalties and raising standards.
  • Line personnel – Regular plotting/review of control charts; killing new (or old) defects.
  • Change agents – Institutionalization of the process.

Sustaining Improvement: The Program

Before the next phase (Phase 2) of the improvement project was launched, new team members were exposed to a two-day quality mindset program. Four key elements for sustaining the improvements were outlined:

  1. Daily control chart plotting by the operating personnel
  2. Analysis of the three worst calls of the previous day
  3. Killing their root causes
  4. Regular periodic review by senior managers

This simple process ensures continuous improvement, however, to make it a “way of life” takes a lot of “grinding in.” During Phase 2 every project meeting was therefore begun with a review of these activities. It took three months for the process to become routine and another three months before the benefits of the discipline began to be felt by the operating personnel.

Phase 2 of the Improvement Project

The objective of Phase 2 was to achieve 99.7 percent response time within service standards for all calls. The project team used a seven-step process for problem-solving.

Step 1. Selecting CTQ: This phase had the same CTQ as Phase 1, improved customer service.

Step 2. Defining the Problem: Problem equals desired state minus current state.

Selecting the metric: The metric chosen was D = actual – standard response time. For 99.7 percent achievement of standard (average + 3 sigma) of D < 0. Analysis of one month of data revealed that (average + 3 sigma) of D was 439 minutes – a surprisingly high figure.

The problem: Reduce (average + 3 sigma) of D from 439 to < 0.

Step 3. Finding Root Causes – Why? Why? Why? Why? Why?: Standard maximum call response time categories were: 30, 60, 120, 240 and 480 minutes.

Finding the vital causes: A Pareto chart revealed surprisingly high deviations in the 480-minute category. This category involved local power supply outages. Because this was an external condition and not controllable internally, these calls were placed in the 480-minute category. But generally the calls were resolved in less than 10 minutes. Eliminating such calls from the reckoning reduced the average +3 sigma of the deviation (D240) of the remaining calls (referred to as T240) dramatically.

D240 = 82 minutes

The calls remaining constituted 93 percent of all calls. The target was to reduce D240 by 50 percent.

A two-pronged approach for finding root causes was used:

  • Tracing a few of the worst calls every day and recording and eliminating the causes of delay (Method A)
  • Categorizing calls by problem, eliminating them by killing their root causes (Method B)

Method A is quick, easy and effective. It gears the team to resolve problems regularly and reinforces the quality mindset for sustaining improvement. The method consisted of brainstorming possible root causes and recording live data for reasons of delay when it occurs.

This procedure resulted in a wide variety of causes being suggested. The most frequently mentioned included engineer busy, wrong classification, user wanted delay, customer served by call not closed, nobody available, call referred to vendor – customer problem solved by not entered, approval delayed and hardware not available. “Decentralized” work to kill the delay in the three worst calls of the previous day was begun

Method B prevents problems from occurring and yields dramatically better results in the longer term. From the above categorization, two problems were taken up for elimination.

Problem 1 –

  • Calls took too long
  • Why? Engineer did not close the calls
  • Why? Outstation calls take too long to close
  • Why?

Using the dictum “What cannot be measured cannot be improved,” questions posed were:

  • What is the standard time to close a call?
  • What is the process?
  • What are the expected times for each step?
Location

Steps

 

1

2

3

4

5

6

7

8

9

1

n

y

n

n

y

y

y

y

y

2

y

n

n

n

y

y

n

y

y

3

n

n

n

n

y

 

 

n

n

4

y

y

n

n

n

n

n

n

n

5

y

y

 

y

y

y

y

y

n

6

y

y

n

y

y

n

y

y

n

7

n

y

y

y

y

y

n

n

n

8

n

n

n

y

n

n

y

y

n

The ensuing discussion clearly showed that no standard operating procedure (SOP) existed. A draft process was developed for testing by some branches of the company with a request to identify the problem steps (y). The chart to the right was the result:

Three points became evident: 

  • Every engineer had a problem in one step or another.
  • Every step had at least one engineer not having a problem.
  • Standardizing a best practice for each step would therefore yield the best process.

A test with best practices yielded the following results: 

Open calls: 100 seconds
Close calls: 75 seconds

Considered adequate at this stage, the process was tested at the two branch offices with the greatest problems. 

 

Branch 1

Branch 2

Standard
Open Call

137

356

100
Close Call

82

444

75

Branch 1 was delighted with the result. Branch 2 had to face the next “why?” It was revealed that the communication link was slow. This was scheduled for upgrading. The project team was elated. A team member volunteered to replicate the results at all branches. The faith in data-based problem-solving and pride in their contribution strengthened the quality mindset. 

Cause

Frequency

Pct.

Cum. Pct.
Loose Connection

14

30

30
Data Purging

9

19

49
Connect PC

8

17

66
Restart PC

6

13

79
Maintenance

4

9

87
Application

2

4

92
Virus

1

2

94
Display Setting

1

2

96
Power On/Off

1

2

98
Operating System

1

2

100
Totals

47

 

100

Problem 2 – Personal computer problems created eight percent of all calls. Data analysis revealed the causes in the table to the left. 

The first cause, loose connections, exposed the problem that the wiring was disorganized. The wiring of one of the most troublesome user stations was made neater, and for 10 weeks no complaints were received. Another group member was made responsible for replicating this housekeeping operation at all stations during the next three months. The quality mindset had become stronger. 

It was now proposed to develop a solution for each call category and hand over the implementation to different members of the project team. In the meantime, Method A was producing startling results. 

Steps 4 and 5. Check Results: D240 was improving substantially. The results achieved were: 

Pre-project D240 = 82 minutes
After six weeks D240 = 39 minutes (best month)
D240 = 17 minutes (best fortnight)

How had the target of 50 percent reduction been achieved? The root cause was found to be that standards existed but were not used. The countermeasure was to begin regular reviews and thus create a sense of urgency. 

Step 6. Standardize Controls: A standard operating procedure was drawn up for daily plotting of the control chart and reviewing the three worst calls. The process was firmly established. 

Step 7. Quality Improvement Story: A quality improvement report was prepared and presented to senior management. 

Conclusion: Creating the Quality Culture

Every quality project needs to have both tangible and intangible gains. In this project the tangible gains were improved response time and faster customer service. The intangible gain was the quality mindset: 

  • Establishing regular control chart plotting, and thus data-based problem-killing
  • Developing team work
  • Learning that what cannot be measured cannot be improved.

The tangible gains took six months in each phase of the project. The intangible gains took a year and could be lost even now if not reinforced by senior management continuously. Building the quality mindset depends on senior management. 

Despite the company’s success in making improvements and sustaining them, it realizes there is room for further improvement. The company is moving ahead in a third phase of its improvement project. Phase 3 aims to cut the average +3 sigma delay level by 50 percent from 42 to 21 minutes for the tightened service standards.

About the Author

Niraj Goyal