®

M A G A Z I N E

From the March/April 2005 Issue n Volume 1, Number 2

T H E F I N A L T O L L G AT E

Reducing Help Desk

Turnaround Time

An Individual Copy
This copyrighted reprint from iSixSigma Magazine is for individual use only.
It may not be republished, electronically or physically reproduced, changed,
posted to a website or otherwise distributed without written permission.
For reprint permissions, please contact iSixSigma Magazine
at [email protected] or 847-919-0922 x4.
This report and others can be ordered
at http://www.iSixSigma.com/pubs

$
995 EACH
 2

T H E F I N A L T O L L G AT E

Reducing Help Desk

Turnaround Time
B Y WAY N E S TA N S B U RY

A microchip manufacturing company found that the

slow response time to computer-related problems was
resulting in nearly three-quarters of a million dollars
annually in lost worker productivity and overtime
expenses. By identifying root causes and instituting
several changes in the existing process, the Black Belt
Six Sigma team succeeded in reducing the defect rate
from 88 percent to just 1.4 percent and saving the
company $586,000 annually.

The Final Tollgate features a Six Sigma project as it would presentation. It is assumed that the project leader has been
be presented to a panel of company executives at the final making regular presentations at each tollgate and that the
project review. The objectives of such a presentation are to executives in the audience have a basic understanding of
1) communicate significant results of the project, 2) share Six Sigma. The content for this project was assembled for
highlights of how results were achieved and 3) gain agree- illustration purposes. It is based on fabricated data from a
ment to close the project. The slides are the Black Belt’s fictional company. Any similarities to an actual project are
visual presentation and the accompanying text is the verbal coincidental.

Do you have an exemplary Six Sigma project to share? Would you like to see it here? Submit it to us at isixsigma.com/submit

Copyright ©2005 iSixSigma Magazine n No Reproduction Without Permission n www.isixsigma-magazine.com/reprints

3 THE FINAL TOLLGATE

DEFINE D M A I C

Reducing Help Desk TurnaroundFinal

TimeTollgate
Problem Statement: 88% of work req uests to the help Business Case: In the commodity business of
desk tak e longer than f ive hours to resolve, resulting in microchip manuf acturing, strong competition and low
productivity loss. The overall cost in lost productivity and prof it margins mak e it imperative to aggressively
overtime is $715,000. target areas of unnecessary cost.
Goal: Reduce work req uest turnaround time (TAT) such VOC: CTQ:
that 100% of req uests are completed in f ive hours or less. Employees have to Computer-related
Operational Definition of TAT: Elapsed time f rom when wait too long to have work requests
support is req uested by employee until issue is resolved computer-related completed in five
by IT technician and logged by help desk coordinator. problems solved. hours or less.

Suppliers Inputs Process Outputs Customers

Site inventory New / used Log problem Working computer Employees in the
computers and productive organiz ation
Software vendor Prioritize problem
employee
New / used • Engineers
Hardware vendor Dispatch
hardware Work ticket with
Troubleshoot • O perators
completed
Software on work request • S upport personnel,
Suppliers were the network Order materials
information including
observed to be a Install materials
SOPs Finance and H R
source of variation in
the process. . . to be Train
Voice/mail system Management
analyzed later Close work request

Define (Slide 1) sive timeline and achieve a significant reduction

For employees with computer problems, the first in defects.
line of defense is the help desk. The minimal
standard set by management for our business is In this project, the customer is internal: employees
to have all help desk work requests resolved of the manufacturing facility. Based on customer
within five hours. An analysis of historical data input, the CTQ was defined as quick response to
revealed that 88% of work requests were defective work requests, and the goal was to reduce the
(taking longer than five hours). average work request TAT by 50 percent. The
outcome would be twofold: increased customer
The result of this slow turnaround time (TAT) is satisfaction and reduced associated costs.
approximately $715,000 of annual expenses
attributed to productivity loss and increased As illustrated in the SIPOC, the process begins
overtime for IT technicians working on computer- when the help desk coordinator receives a phone
related problems. This figure was calculated call or an email from a customer with a comput-
using an average hourly rate of pay for employ- er-related issue to be resolved and ends when
ees and was verified by the Finance department. the work request is closed.

As you know, strong competition and low profit The team’s operational definition of TAT is the
margins are the nature of our business. We cannot elapsed time from when support from the help
afford to ignore or respond slowly to any desk is requested by an employee until the work
opportunity to reduce costs. With this motivation is completed by the IT technician and logged by
in mind, the team was asked to tackle this the help desk coordinator.
problem immediately and complete the project
within four months. By following the DMAIC In the course of completing the SIPOC and
process, we were able to accomplish the aggres- identifying suppliers, we realized that the site

Copyright ©2005 iSixSigma Magazine n No Reproduction Without Permission n www.isixsigma-magazine.com/reprints

 THE FINAL TOLLGATE 4

MEASURE D M A I C

Reducing Help Desk Turnaround Time

Unit Work request Baseline Cpk = -0.39 Measurement System Improved to
Defect TAT > 5 hours USL -1.18 Sigma <10% of Tolerance Via Automation

DPMO 881,361 Standard

Deviation 4.4 hours

In Spec Out of Spec

2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0 7.5 8.0

29.3
28.1
2.96
-7.1
-9.6
-8.3

-5.8
-4.6
-3.3
-2.1
-0.8
0.45
1.7

4.22

9.24
10.5
11.8
13
14.3
15.5
16.8
18
19.3
20.5
21.8
23.1
24.3
25.6
26.8
5.47
6.73
7.99

Measurement error (before) = 105%

Run Chart Showed No Discernable Pattern

Average TAT, Hours

16

12

4
2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0 7.5 8.0
July Aug Sept Oct Nov Dec Jan Feb Mar April May
Measurement error (after) = 9.5%
Month

Current Process Is Not Capable of Completing

a Work Request in Less Than Five Hours
inventory and outside vendors could have a (USL) of five hours, showed a defect rate of 88%
significant impact on the variation in TAT. For (DPMO = 881,361), with a Cpk, or process
example, unavailable parts (e.g., site stockroom capability index, of -0.39 and a Sigma of -1.18.
locked after hours) or having parts out of stock Because the mean was larger than the upper spec
could add many hours to the TAT. This would limit, a negative Cpk was calculated, meaning
be something to explore later in the project. that the process is totally incapable.

Measure (Slide 2) A run chart was developed on the TAT historical

The work request TAT data originated from data, and no discernable patterns were observed.
work request tickets stamped with the start and During the time the data was collected, both the
finish times. Also included on the ticket were number of work requests and the workforce
the IT technician’s name and a description of remained relatively constant.
what had been done.
Measurement system analysis was conducted
Based on the historical data, the average TAT was on a sample of work requests, which revealed
calculated to be 10.2 hours, with a standard devi- that there could be as much as a one-hour
ation of 2.1 hours. Therefore, to reach our goal of discrepancy between when the IT technicians
a 50 percent reduction in average TAT, the gap to began or finished working on a problem and
close was five hours. This allowed us to define a when the Help Desk coordinator recorded the
defect as any work request that was not completed time. To address this, the system was automated
within five hours. Although not explicitly such that the IT technicians would, through
mentioned, we also want to reduce variation wireless technology, enter the exact time that
(standard deviation) by at least 50 percent. they started and finished working on a problem.
This reduced the measurement error from 105%
Analysis of the data, using an upper spec limit to 9.5% of the tolerance.

Copyright ©2005 iSixSigma Magazine n No Reproduction Without Permission n www.isixsigma-magazine.com/reprints

 THE FINAL TOLLGATE 5

ANALYZE D M A I C

Reducing Help Desk Turnaround Time

14
12
10
One-Way ANOVA of Data Confirmed Printer and Hardware Problems
8 Have Statistically Higher TATs (p < .001) Than Other Problems
6
4
2
0
Printer Hardware Misc. Office Desktop Connectivity E-mail User
Software OS

Measurements Materials Methods

Increase in Server Users Install Root Causes of High
Hardware Software Not Proactive Printer Incorrectly
Requests Corrupt with Upgrades Printer and Hardware
No Troubleshooting SOP
Increase in Run Out
No Training More Research Needed
TAT Identified:
Software of Stock
Requests of Users
Availability Slow TAT for • Age of PCs and
of Hardware Requisitions Printer/
Hardware monitors
Office Area vs. Rolling Out TAT
Manufacturing s/w w/o Age of PC/ Printer • Requisition process
Floor Testing Monitors Incompatibility
Complexity Issues
Limited IT
Memory • Lack of SOPs
Operating Resources Hardware is
Upgrade
System Waiting on Needed Too Limiting • Printer incompatibility
Complexity Vendor issues
Mother Nature Manpower Machines

Analyze (Slide 3) causes fell into two main categories: Methods

From information on the work tickets, we catego- and Machines.
rized the work done into problem buckets and
performed a one-way ANOVA of this data. The From the CE analysis, the team identified four
analysis showed that hardware and printer inputs that were root causes having the greatest
troubles had a significantly higher TAT than the impact on TAT: the age of PC/monitors, requisi-
other problem categories (p < 0.001). Statistical tion process, lack of standard operating proce-
analysis also revealed that there was no difference dures (SOPs) and printer incompatibility issues.
in TAT between different IT technicians (p > 0.10).
Improve (Slide 4)
The statistical analysis was further verified The team developed corrective actions for each
through Pareto analysis of TAT and cost of poor of these causes and then assigned a member and
quality (COPQ), in which hardware and printer due date to each. Corrective actions included the
troubles accounted for 65% of the total TAT and following:
40% of the total COPQ.
1. Several PCs and monitors that constantly
At this point, although we had already identified caused problems and for which parts were
hardware and printer problems as leading con- harder to obtain were replaced. The asso-
tributors to the high TAT, the team continued to ciated cost was $23,000 in parts and
use the Analyze tools to identify the root cause(s) labor. The IT technicians also upgraded
of the hardware and printer problems themselves. standard PC configurations to include
more RAM.
The cause-and-effect (CE) diagram was used to
brainstorm the causes for the high TAT associat- 2. A blanket purchase order was set up for
ed with hardware and printer issues. Most of the miscellaneous hardware and software pur-

Copyright ©2005 iSixSigma Magazine n No Reproduction Without Permission n www.isixsigma-magazine.com/reprints

6 THE FINAL TOLLGATE

Improve D M A I C

Reducing Help Desk Turnaround Time

Solutions: Pilot Proves Solutions Improve TAT
1. Replace PCs/monitors and 18
Max
upgrade memory 16
Q3
2. Install blanket purchase order 14 Median
12 Q1
3. Institute tighter inventory control for Min

TAT, Days
spare parts 10

4. Upgrade client-server with 8

printing fixes 6

5. Develop SOPs with training 4 Hypothesis Test Results

– Troubleshooting SOP 2
Mean = 10.2 vs. 6.0, p < 0.001
Standard deviation = 2.10 vs. 1.59, p < 0.001
– Hardware and software SOPs 0
– Completing work request ticket SOP Before After

Results
• Within five months, TAT reduced to 3.2 hours and variation to 0.82 hours
• Reduction of defects yields $586,000 in annualized savings
• Implementation of solutions cuts risk in half: RPN reduced from 1282 to 619

chases, because ordering spare parts and of the Black Belt team. The accounting manager
waiting for material request approvals was was added to the team to help set up blanket
very time consuming. A tighter inventory purchase orders.
control for spare parts was established as
well. IT technicians also were allowed to A pilot test was conducted, in which corrective
keep an inventory of commonly used actions were put in place in a limited part of the
parts in their area. business to determine the impact on TAT. After
one month, a two-sample hypothesis test of the
3. The client-server was upgraded with print- mean and of variation was conducted on the
ing fixes, resulting in an improved print TAT data – both before and after the pilot. This
service that alleviated the problem of print- evaluation showed that implementation of the
ers losing connection to the network and corrective actions had a statistically significant
not distributing printers to the desktop. effect on the mean TAT (t = 7.7, p < 0.001) and
TAT standard deviation (F = 58.7, p < 0.001).
4. Several standard operating procedures
(SOPs) were developed: Based on the good results in the pilot, the cor-
a. Basic trouble shooting SOP to rective actions were institutionalized throughout
determine the root cause of the the organization, and the progress was followed
problem more rapidly for five additional months.
b. SOPs for installing hardware and
software After five months, the average TAT was reduced
c. SOP for inputting information onto from 10.2 hours before implementation to 3.2
the work request ticket hours after implementation of the corrective
actions (t = 22.1, p < 0.001). The standard devi-
Addressing the root causes affected the make-up ation was reduced from 2.1 hours to 0.82 hours

Copyright ©2005 iSixSigma Magazine n No Reproduction Without Permission n www.isixsigma-magazine.com/reprints

 THE FINAL TOLLGATE 7

Control D M A I C

Reducing Help Desk Turnaround Time

Control Plan
Defect Rate
• Process monitored daily through I-MR chart (see below)
• Out-of-control symptoms trigger action plan
Reduced from
– One data point outside control limits 88% to 1.4%
– Four of five data points greater than one sigma from the center line
• Process audited every quarter

Individuals-Moving Range (I-MR) Chart

Before After
20
18 UCL = 17.293
16
14
12
CEN = 10.2
10
8
6 UCL = 5.4677

4 CEN = 3.2333
2 LCL = 3.107
0 LCL = 0.9989
July Aug Sept Oct Nov Dec Jan Feb Mar April May June July Aug Sept Oct

(F = 489, p < 0.001). The defect rate decreased coordinator has been trained to monitor the
from 88 percent to 1.4 percent, resulting in an control chart and therefore follow the OCAP
annual savings of $586,000. The FMEA was when an assignable cause is observed.
revisited, and implementation of the corrective
actions resulted in a significant decrease in the The procedures that were developed in the
overall risk. Improve phase were institutionalized throughout
the site by training all members who touch the
Control (Slide 5) process and ensuring that the procedures are
The effect of implementing the corrective aligned with the site’s quality manual. The work-
actions was observed almost immediately and is station managers will conduct an audit of the
illustrated using an Individuals and Moving procedures and the control charts in another
Range (I-MR) chart. The work request process three months to ensure that the new procedures
from this point forward will be monitored daily are being used and the TAT level is being main-
using the I-MR chart. tained. u

Initially, only one out-of-control symptom will Your input is welcome. Please send comments to
be used for detection of assignable causes: one [email protected].
data point outside the control limits. As the IT
members become more sophisticated in the use
of the I-MR chart, a second out-of-control symp-
tom will be added: four of five data points Wayne Stansbury is a certified Master Black Belt.
greater than one sigma from the center line. He has taught Champion, Black Belt and Green
Belt courses across the United States, Mexico and
An out-of-control action plan (OCAP) has been Japan. Mr. Stansbury is currently a senior project
developed for the I-MR chart. The help desk manager at Harcourt Assessment.

Copyright ©2005 iSixSigma Magazine n No Reproduction Without Permission n www.isixsigma-magazine.com/reprints