Help Desk Customer Service and Support With thanks

Help Desk, Customer Service and Support With thanks to Rabih Neouchi William Luthi University of Alberta ICT-management Fall 2004 Wim De Bruyn, Bert Van Vreckem

Outline n n n n Introduction to Help Desks Help Desk Case Logging Distributed Help Desks Examples of Help Desk Technologies Security Issues Future Trends (AI Approaches) Summary References

Introduction to Help Desks n n n Motivation of Help Desks Architecture of Classic Models Problems in Classic Models Politics What Users Expect

Motivation of Help Desks n n Cope with evolving technology As you increase complexity n n n Probability of failure grows exponentially The number of available technicians becomes limited Help desk operator’s expertise tend to diverge (do not overlap) Problem solving experience becomes distributed and changes over time Expertise needed exceeds resources available to end users Assist the user’s needs

Architecture of Classic Models Example of Support Layers [1]

Architecture of Classic Models Call Flow in a Help Desk Structure [2]

Problems n n Handling support requests efficiently Efficient use of employee knowledge Reduction of maintenance costs Politics Potential Pitfalls of a Dispatcher [2]

Politics [5]

What Users Expect n Help Desk Operators to provide n n n Guidelines on how to use products Maintenance tasks Solutions to problems: n n n In a short time With short notice Expertise in all product areas

Help Desk Case Logging n n n n What is Case Logging? General Case Structure Why Log Cases? Help Desk Knowledge Repository Case-Based Reasoning (CBR) Case Reuse Knowledge Engineering of CBHD System

What is Case Logging? n Store request and solutions for each help desk request case [3] n n n n Case Number Date User Help Desk Technician Assigned Problem Description Reason Solution Other Notes

General Case Structure Help Desk Case [1]

Why Log Cases? n n Create Knowledge Base Case-Based Reasoning (CBR) The Case-Based Reasoning Process Model [3]

Help Desk Knowledge Repository Knowledge acquisition and Transfer [1]

Case-Based Reasoning (CBR) n Idea: n n n Uses previous support cases and adapts them to solve new problems If no solution is found, a new case is entered into the KB Stores, shares and re-uses experiences

Case Reuse Process of a case-based help desk system [3]

How to Knowledge Engineer a Case-Based Help Desk System Development Procedure [7]

Homer Case Example [3, 11]

How to Knowledge Engineer a Case-Based Help Desk System WP What is Category? Symptoms Tree [7] Print How To Error What is the error message? Integer divide by zero Missing DLL file GPF Insufficient Memory Other What is your print problem? Can’t print Can’t find printer output Prints Garbage

Distributed Help Desk n n n Online Support Model Advantages and Disadvantages of Online Support Problem Prevention and Troubleshooting of Online Support Systems Centralized vs. Decentralized vs. Hybrid Call centers Customer Support Agents

Online Support Models n Motivation n Long distance phone calls are [10]: n n n Inefficient Ineffective Costly Time consuming Poor in quality

Online Support Model Generations [4]

First Generation n Informational First Generation Support Site [4]

Second Generation n Conversational Second Generation Support Site [4]

Third Generation n Decisional Third Generation Support Site [4]

Advantages of Online Support [4] n n n Self-service Remote access Single source access Information sharing Assisted service Integration

Potential Disadvantages of Online Support [4] n n n n Not knowing your customers Useless functions and information Outdated information Staff may not have skills for updating the site Poor Marketing Site does not change with organization Poor site design Negative impact on bottom line

Problem Prevention and Troubleshooting [4] n Problem symptoms: n n n At the earliest stages During the scope definition During the staff selection and design Involving the functions and tools Involving the processes and implementation During site operation

Earliest Problem Symptoms [4] n “For heaven’s sake, everyone knows we need a support site. Let’s just do it. ” Symptom Want to build site with no planning Eventual Results Negative support site Possible Causes No formal justification Prognosis and Suggestions Do a cost justification for Recovery Prevention Measure actual costs and return on investment

Problem symptoms during scope definition [4] n “We want the site to eliminate all other support, triple sales, bring world peace, and so on. ” Symptom Scope is unrealistic Eventual Results The site is late, over budget, and poorly implemented Possible Causes Stakeholders have unrealistic expectations Prognosis and Suggestions Negotiate with clients and present a cost for Recovery benefit analysis Prevention Make cost justifications mandatory

Problem Symptoms during Staff Selection and Design [4] n “Jim can set up the knowledge base. He’s a whiz at HTML. ” Symptom Person does not know about knowledge bases Eventual Results Knowledge base’s information is incorrect or unclear Possible Causes Perception technical wizards can do anything Prognosis and Suggestions Train the person or find a more suitable for Recovery candidate Prevention Have appropriate people trained in knowledge bases

Problem symptoms involving functions and tools [4] n “Let’s build our own tools. It can’t be that hard. ” Symptom Want to create tools rather than buy them Eventual Results Project is late and over budget Possible Causes A lack of understanding of tool functionality and complexity Prognosis and Suggestions Scrap it and buy something that works or do for Recovery proper research and analysis for construction Prevention Do thorough analysis of tool requirements

Problem symptoms involving processes & implementation [4] n “We can’t really test this stuff until our site is up. ” Symptom Site goes into production without being tested Eventual Results Customers experience problems Possible Causes You don’t know how to test the site Prognosis and Suggestions Poor. Outsource extra help while you test the for Recovery site to fix problems Prevention Make testing mandatory for each site iteration

Problem symptoms during site operation [4] n “Jenna can manage the site along with the four other departments she manages. ” Symptom Site manager does not have time for the site Eventual Results Diminished return and usefulness Possible Causes Lack of understanding or staff shortages Prognosis and Suggestions Create a cost justification for management to for Recovery hire more people for the site Prevention Define site management requirements and create a job description

Centralized vs. Decentralized vs. Hybrid n n n Centralized Decentralized Hybrid

Centralized Help Desk n Advantage n n Collaboration Disadvantage n May not be responsive to local needs

Decentralized Help Desk n Advantage n n Each unit can serve local needs Disadvantage n May produced inefficiencies and redundancies

Hybrid Approach n Tuft’s IT Support Services Model n n Takes best from both worlds “A seven person support team is funded by the central IT organization but ‘lives’ in the University’s local units’’ [8] Collaborative support team Whole university’s IT needs and local interests are served

Call Centers n n n Telephony call centers Web-based using telephone and Web-based multimedia call centers

Web-based Multimedia Call Centers n Use Vo. IP (Voice over IP) Internet Call Me/Callback Request [10]

Customer Support Agents n Callback services n n n Audiovisual interaction with help desks Co-browsing and multimedia presentations Three way teleconferences (customer, agent, another customer) Customer support agent for an e-commerce site [10]

Examples of Help Desk Technologies n n Avrasoft Size. Master Help. Desk People. Soft Help. Desk Blue Ocean Track-It! HOMER

Size. Master Help. Desk [14] n http: //www. avrasoft. com/sizemaster/

People. Soft Help. Desk [15] n http: //www. peoplesoft. com/corp/en/products /line/crm/help_desk/features. asp

Blue Ocean Track-It [16] n http: //www. blueocean. com/smallbusiness/tw ebinfo. htm

The HOMER System n HOMER: HOtline Mit ERfahrung Hotline with Experience CAD/CAM Help Desk Daimler-Benz

The HOMER Architecture [1, 11] Domain Modelling [1]

Modes of Execution in HOMER n User Driven Mode n n Operator is experienced System Driven Mode n Operator lacks experience User Driven Mode [1]

HOMER Performance n n Period: 2 Months 2002 Evaluation: n n n 102 calls 45 unsuitable 57 suitable: n n n 18 (32%) - Solved by HOMER 34 (59%) – New case created 5 (9%) - Case not found Calls Suitable for HOMER [11]

HOMER Performance Call Resolution Time [11]

HOMER Along Product Lines Methodology Benefits [11]

HOMER Along Product Lines Methodology Benefits [11]

Security Issues n n Help Desk assisting in identification of incidents Vendors n n n Confidentiality Privacy User Threats n n Malicious users Denial of Service Attacks

Security Issues n Identification of Incidents [13] n n Potential incidents are identified by user community First call goes to Help Desk A properly trained staff can identify incidents from calls or requests (cases) Help Desk staff must capture information so incidents can be dealt with swiftly

Future Trends (AI Approaches) n Automatic Assignment of Technicians n n NL Agents for Help Desk Requests n n n Information Retrieval Natural language driven automated help desks ANN (Artificial Neural Networks) CBR-ANN (Web. Service System) n Hybrid case-based reasoning combined with ANN approach for machine fault diagnosis

Automatic Assignment of Technicians Process Flow [9]

Automatic Assignment of Technicians n IR techniques n n n Parsing Stop word removal Keyword matching Keyword weighting Model Stages n n n User report analysis Selection of Technician Refine selection with assignment rules

Automatic Assignment Results n Study conducted in 2002: n n 48% of the system assigned cases were better than human experts’ 92% of system cases were good or better [9]

NL Agent Help Desks NL Help Desk Architecture [6] n Main components n n n Textual input analysis Request additional information from user, if needed Propose solution and verify

NL Agent Results n n NL interfaces are user friendly Generic system that can be adapted to different media: n n Email driven systems Web based chat Telephony services Expect significant reduction in maintenance costs [6]

Web. Service System [12]

CBR-ANN Help Desks Fault Diagnosis Engine [12]

CBR-ANN Results n Trained with inputs and solutions from a knowledge base [12]

Summary n Help Desk Importance n n As complexity increases, systems are more vulnerable to failure User support is necessary to increase productivity Poor support strategies place burden on service agents Support should not be neglected n Problems should be dealt with fast and resource-efficient strategies

Summary (continued) n Whenever dealing with a CBR (OO) system: n n n Case-base enables n n Should log all your cases Continuous knowledge acquisition and maintenance are necessary Access Reuse Knowledge extension Gains n n n Availability of solutions to previous problems Knowledge is cumulated and preserved Problem solving time is optimized

Summary (continued) n Distributed Help Desks n n Online support models Call centers

References n n n n [1] Mehmet Göker et. al. The Development of HOMER: A Case-Based CAD/CAM Help-Desk Support Tool. LNCS, Vol. 1488, pp 346 -357, 1998. [2] Barbara Czegel. Running an Effective Help Desk. John Wiley & Sons. 1998. [3] Mehmet Göker and Thomas Roth-Berghofer. Development and Utilization of a Case-Based Help-Desk Support System in a Corporate Environment. LNAI, Vol. 1650, pp 132 -146, 1999. [4] Barbara Czegel. Technical Support on the Web: Designing and Managing an Effective E-Support. John Wiley & Sons. 2001. [5] Keith Patching and Robina Chatham. Corporate Politics for IT Managers: How to Get Streetwise. Butterworth-Heinemann. 2000 [6] Melanie Knapp and Jens Woch. Towards a Natural Language Driven Automated Help Desk. LNCS, Vol. 2276, pp 96 -105, 2002. [7] Christine W. Chan et. al. Knowledge engineering for an intelligent case-based system for help desk operations. Expert Systems with Applications, Vol. 18, No. 2, pp 125 -132, 2000.

References (continued) n n n [8] Kathleen Cummings. Reinventing Support Services: Transcending the Centralized-Decentralized Support Model Debate. In Proc. SIGUCCS’ 02, pp 232 -233, 2002. [9] Avinoam Lazarov and Peter Shoval. A rule-based system for automatic assignment of technicians to service faults. Decision Support Systems, Vol. 32. pp 343 -360, 2002. [10] Doo-Hyun Kim et. al. Collaborative Multimedia Architecture and Advanced Internet Call Center. In Proc. ICOIN’ 01, pp 246 -250, 2001. [11] Ralph Bergmann. Experience Management for Self-Service and Help-Desk Support. LNAI, Vol. 2432, pp 315 -346, 2002. [12] S. C. Hui et. al. A web-based intelligent fault diagnosis system for customer service support, Engineering Applications of Artificial Intelligence, Vol. 14, No. 4, pp 537 -548, 2001. [13] Eric Maiwald and William Sieglein. Security Planning & Disaster Recovery. Osborne/Mc. Graw-Hill. 2002.

References (continued) n n [14] Size. Master Help Desk. Avrasoft Corporation, 2003. http: //www. avrasoft. com/sizemaster/ [15] People. Soft Inc. Help Desk. People. Soft, 2003. http: //www. peoplesoft. com/corp/en/products/line/crm/help_desk/featu res. asp [16] Track-It! Demo. Blue Ocean Software Inc. , 2002. http: //www. blueocean. com/smallbusiness/twebinfo. htm Rabih Neouchi William Luthi CMPUT 660 Winter 2003, University of Alberta

Questions & Answers