Your SlideShare is downloading. ×
Help Desk Improves Service & Saves Money With Six Sigma
Upcoming SlideShare
Loading in...5
×

Thanks for flagging this SlideShare!

Oops! An error has occurred.

×
Saving this for later? Get the SlideShare app to save on your phone or tablet. Read anywhere, anytime – even offline.
Text the download link to your phone
Standard text messaging rates apply

Help Desk Improves Service & Saves Money With Six Sigma

3,202
views

Published on

Help Desk Improves Service & Saves Money With Six Sigma

Help Desk Improves Service & Saves Money With Six Sigma

Published in: Business

0 Comments
2 Likes
Statistics
Notes
  • Be the first to comment

No Downloads
Views
Total Views
3,202
On Slideshare
0
From Embeds
0
Number of Embeds
0
Actions
Shares
0
Downloads
377
Comments
0
Likes
2
Embeds 0
No embeds

Report content
Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

Cancel
No notes for slide

Transcript

  • 1. Making the Case for Quality Help Desk Improves Service and Saves Money With Six Sigma by Francisco Endara M. Organizations look to their help desks to provide expert problem-solving services. When users run into the unexpected, help desk staff intercede, bringing a systematic approach to make sense out of chaos. So when a help desk service supplied by Schlumberger faced a crisis of its own, marked by rises in customer complaints and losses in dollars and cycle times, it responded, in typical help desk fashion, by applying a methodology to improve the quality of its processes. Schlumberger is an international oilfield services provider headquartered in Houston, Texas. Through an outsourcing contract, we supply help desk services for a global telecom company that offers wire- line communications and integrated telecom services to more than 2 million cellular subscribers. This translates to 2 million potential customers for the help desk, which is located in Ecuador. When our telecom client decided to improve its business processes by using Six Sigma, an approach made popu- lar by General Electric, Schlumberger participated in the implementation, managing a project to address the help desk crisis. At the time, Schlumberger had extensive experience in Six Sigma applications in process manufactur- ing, and our Six Sigma efforts have since evolved to concentrate on Lean Six Sigma, which we describe as the combination of speed and improvement. However, this project for our telecom client was our first participation in a Six Sigma define-measure-analyze-improve-control (DMAIC) effort in the help desk environment. Define We defined a team of 30 people—the entire help desk staff—to participate in the improvement process. As help desk manager and a Quality MBA, I was project leader, and a dedicated Six Sigma At a Glance . . . Black Belt and Green Belt provided further expertise with the methodology. To define our quality objectives and project scope, we gathered information about our help desk • The help desk service for a customers’ needs and the issues they were experiencing. Results of customer surveys identified key telecom company launched opportunities and process gaps. Knowing we needed a measurable and challenging goal, we established a Six Sigma project to two objectives based on the actual situation: improve the quality of its processes. 1. The average solution time for issues reported to the help desk was 9.75 hours. We set a goal to • The project team reduced reduce the average solution time by 50%. the average solution time 2. The number of issues reported to the help desk had reached an average of 30,000 per month. for help desk customers Reducing the total number of issues reported to the help desk would allow us to address those from 9.75 hours to 1 hour. issues that hadn’t been resolved because of a lack of time, and to reduce the number of abandoned • Help desk operating costs also decreased by 69%. calls (calls not answered). We set a goal to identify preventable issues so that customers would not have to contact the help desk in the first place, and we set an ideal target of 15,000 issues. The American Society for Quality ■ www.asq.org Page 1 of 4
  • 2. To identify “non-quality” costs (costs associated with poor serv- requirements for development and production. We identified ice quality) and quantify the objective of the Six Sigma project, our customer requirements and key service factors based on we had to find the “hidden factory,” or the source of a lack of satisfaction surveys and continuous meetings. process control. We also had to quantify total costs and estimate Our QFD matrix, shown in Figure 2, revealed that the customer calculated non-quality costs. We calculated our costs per hour was always looking for: and costs per employee, based on the average desktop solution, and predicted that reaching our project goals could bring cost • The solution savings of 60% to the help desk. • A timely solution • Feedback Finally, we described the existing help desk process. Identifying all • Friendly attention inputs and outputs for our “as-is” process would allow us to streamline and systematize the process as part of our improvement. These customer requirements translated to the following process Measure specifications: • Process model Using Pareto diagrams, cause and effect diagrams, and a capabil- • Technical knowledge ity analysis, we took the following steps to capture our service • Solution experience measurements: • Staff self-confidence 1. Obtained data from the help desk ticketing system tool and • Ticket assignation classified all application issues by month • Tools 2. Transported the data to Minitab, a statistical software program that provides data analysis and graphics capabilities, to run By conducting an internal analysis, we identified performance Pareto diagrams for application function and solution time factors that were affecting the process and the output. Figure 3 3. Conducted Pareto analysis to understand that our problems presents the fishbone cause-effect diagram our team created. were rooted in the software applications used by the end users We next analyzed different hypotheses based on the data from 4. Measured and monitored our process through statistical the measurement stage. For example, noting that the average process control (SPC); used sample mean (X-bar) and range solution time for help desk technicians working at the call center (R) charts, shown in Figure 1, to detect when a process was seemed to be lower than the average for technicians working on going out of quot;controlquot; site with clients, we conducted a hypothesis test structured 5. Evaluated our “process capability,” the capability of our around the question, Is there a difference between having help process to perform within specification limits, and deter- desk employees working at an off-site facility rather than on site mined that our process was functioning at 1.5 sigma within the client’s main office? The null hypothesis (Ho) was that there was no significant difference; the alternative hypothe- Analyze sis (Ha) was that there was a significant difference. In the analyze stage we used quality function deployment (QFD), Figure 4 displays the results of our two-sample T-test, a test for a method for translating customer requirements into technical assessing whether the means of two groups—in our case, the call Figure 1 Sample Mean and Range Charts for Help Desk Solution Times Figure 2 QFD Matrix Note: These control charts are not from the measure stage but show process data from project beginning to end. Friendly attention Client’s feedback REQUIREMENTS 10 9.75 hrs Timely solution Individual Value Reduction of 89% 6.20 hrs UCL = 5.664 Solution CLIENT 5 Mean = 2.532 1.02 hrs 0 LCL = –0.59 Ranking client 8 10 7 5 1 2 3 4 5 6 7 Subgroup priority Mar-04 Apr-04 May-04 Jun-04 Jul-04 Aug-04 Sep-04 Months 5 1 PROCESS IDENTIFICATION Rank %Rank Moving Range 4 UCL = 3.842 Process model 9 10 8 7 263 18.10% 3 Self-confidence 5 10 7 9 234 16.10% 2 R = 1.176 1 Technical knowledge 8 10 6 7 241 16.59% 0 LCL = 0 Ticket assignation 8 10 2 9 223 15.35% The causes of variation that exceed the upper and lower control limits Solution experience 6 10 3 7 204 14.04% (UCL and LCL, respectively) must be eliminated in order to bring the process Tools 8 7 1 2 151 10.39% back into statistical control. The American Society for Quality ■ www.asq.org Page 2 of 4
  • 3. center and the help desk—are statistically different from each minutes. However, as it is more expensive to increase the number other. We found no advantage in keeping help desk employees of agents than to conduct training, we decided to implement a working at the call center. training program for help desk agents and end users. Once we saw these results, we moved our help desk agents to the Another successful experiment was to implement a patch to client’s main office area, allowing us to validate our hypothesis. improve the performance of software applications. The patch effectively completed internal system operations that our techni- As a result of our QFD analysis, cause-effect analysis, and cians used to perform manually while attending user calls. hypothesis testing, we identified the key variables that influenced Ultimately, this reduced both the number of help desk calls and our process quality: the percentage of abandoned calls. • Training (to allow technicians to gain knowledge and experience) Other actions we took based on DOE results include: • Remote control (to solve issues remotely) • Software capability • Adjustment to the new process model represented in Figure 5 • Team management • Implementation of help desk remote control tools • Work schedule changes based on call volume patterns Improve • Improvements in ticket classification and escalation (once a ticket wasn’t resolved by the first and second support level, it had to advance to third-level support within the required time To identify process improvements, we used design of experiments limit of 24 hours) (DOE), a method for conducting a series of tests to determine the • Operating system upgrades for help desk customers relative importance of the key variables and to assist in selecting optimum operating values. We limited our initial number of Control variables so that the time, effort, and cost of testing were not excessive. After the array of tests we could illustrate the impact those variables had on the process. In the final stage of the project, we tracked the changes we made and established feedback into the process. This was essential for Tests conducted: guaranteeing process continuity. • Implementing a patch installation (to reduce the number of Our process control tasks included: problem issues, the IT department would develop a software patch for clients) • Keeping control of all open tickets within 24 hours • Improving software categorization (technicians would have to • Updating personal computers categorize each service ticket correctly every time a new user • Weekly revision of key performance indicators called for service) • Monthly reports of top application errors • Improving training versus increasing staff • Monthly training • Upgrading operating systems • Ongoing analysis of the hidden factory • Implementing a new process model As shown in Figure 6, documented benefits to the company as a The most significant experiment was a test of the effect of result of our project include the following: training versus incrementing help desk technicians. Results Figure 4 Two-sample T-test demonstrated that both training and increasing the number of technicians would reduce our average solution time to 16 Ho = No significance (With SLB employees working in Call Center A OR without SLB Figure 3 Fishbone Cause and Effect Diagram for employees working in Call Center B) Help Desk Process Ha = Yes significance (With SLB employees working in Call Center A OR without SLB employees working in Call Center B) Measurements Materials Personnel Two-Sample T-Test and CI: Call Center A, Help Desk B Lack of Remote Training Two-Sample T for Call Center A vs Help Desk B Control Control Tools Motivation N Mean StDev SE Mean Audits RBS A 10 84.24 2.90 0.92 Issues Procedures Experience report RBS B 10 85.54 3.65 1.2 Help Desk Process Failure Difference = µ Call Center A – Help Desk B Environmental Servers Software conditions Estimate for difference: -1.30 Application Management QHSE Laptops 95% CI for difference: (-4.41, 1.81) (Quality, Health, Safety, & T-Test of difference = 0 (vs not =): T-Value = -0.88 P-Value = 0.390 DF = 17 Process Desktops Environment) Result: P-Values > 0.05, Accept Ho. Environment Methods Machines The American Society for Quality ■ www.asq.org Page 3 of 4
  • 4. • Decreased the number of help desk issues by 32% the Information Technology Infrastructure Library (ITIL), and Six • Raised our capability to meet our target of 15,000 total issues Sigma, which brought results for us. These approaches will only from 1.5 sigma to 4 sigma (see Figure 7) work, however, if they focus on business reasons and seek to add • Reduced average desktop solution time from 9.75 hours to value to the business and for the customer. only 1 hour, an improvement of 89.5% For More Information • Reduced our call abandonment rate from 44% to 26% • Reduced help desk operating costs by 69% • To learn more about this help desk project, contact Francisco Endara at 593 2 2806283 or francisco_endara@hotmail.com. Focus on Business Reasons for Improvement • To learn more about Schlumberger, visit http://www.slb.com/. • Learn more about the Six Sigma methodology and the quality Today the help desk is one of the principal engines of the infor- tools used in this improvement project. Visit ASQ’s Learn mation technology area in any company, which increases the About Quality Web pages: http://www.asq.org/learn-about- urgency of offering efficient and successful service. Help desk quality/index.html. analysts think of themselves as process-driven problem-solvers for technical issues, but they too often do not focus this energy About the Author on the business of operating a help desk. Francisco Endara M. is a service desk manager for Many of the problems help desks face can be traced to a struc- Schlumberger. He earned his MBA in quality and productivity ture that is undefined or not focused on business. Help desk from Catholic University of Ecuador in 2005, and he is an ASQ analysts need to concentrate on applying their problem-solving member. He has six years of help desk experience, working as an abilities toward eliminating problems and reasons for calls and engineer, team leader, and service manager. minimizing downtime for customers. The big question is, How? The answer for our project, and likely for many help desks fac- ing similar crises, was a process approach to improvement: Figure 6 Help Desk Project Results 1. Implement a help desk process model—identify the elements Beginning Project goal Project results of the process and standardize the process measure 2. Implement policies and procedures for providing service Number of issues Reduce to approach 20,400 30,000 3. Identify and use tools to automate solutions and prevent (32% improvement) reported per month ideal target of 15,000 problems Improve to approach 1.5σ 4σ Process capability ideal of 6σ 4. Design effective work groups and teams 5. Collect feedback from customer surveys to measure perform- Average solution time 1 hour Reduce by 50% 9.75 hours (89.5% improvement) for reported issues ance; document internal process measurements monthly Reduce as much 6. Improve the help desk based on feedback Call abandonment 44% 26% as possible 7. Control process improvements to sustain the gains Help desk operating Confidential Reduce by 60% 69% reduction costs Help desks have many options that provide methodologies for pur- suing improvement, including total quality management (TQM), Figure 7 Process Capability After Six Sigma Implementation for Total Number of Issues Help Desk Process Performance Figure 5 New Help Desk Process Model (Total Number of Issues) Process Data USL 25555.0 Provider Feed Forward Target 15000.0 Client Feed Forward User profile LSL 5000.0 LSL Target USL Ticket closed Type of required Mean 15775.4 equipment Sample N 5 StDev 2252.22 Provider Problems Client Potential (Within) Capability Call Center Call Center Requirements Cp 1.52 CAVS IT Help Desk CPU 1.45 CAVS Call Attention Services Process IT/Technical CPL 1.59 Questions Others Cpk 1.45 Others Externals Orders Externals 4σ Client Feed Back Feed Back to Provider Ticket solution Complaints Standards Service recalls 5000 10000 15000 20000 25000 Policies Satisfaction surveys Number of issues The American Society for Quality ■ www.asq.org Page 4 of 4