1 / 23

DataMOCCA

DataMOCCA. D ata MO dels for C all C enter A nalysis. Project Partners: Technion: Paul Feigin, Avi Mandelbaum, Sergey Zeltyn (IBM Research) SEElab : Valery Trofimov, Ella Nadjharov, Igor Gavako Students, RA’s

millicent
Download Presentation

DataMOCCA

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. DataMOCCA Data MOdels for Call Center Analysis Project Partners: Technion: Paul Feigin, Avi Mandelbaum, Sergey Zeltyn (IBM Research) SEElab:Valery Trofimov, Ella Nadjharov, Igor Gavako Students, RA’s Wharton: Larry Brown, Noah Gans, Haipeng Shen (N. Carolina), Linda Zhao Students, Wharton Financial Institutions Center Companies: US Bank, IL Telecom (both Aspect-Based), IL Bank (Genesys), …

  2. Project DataMOCCA • Goal: Designing and Implementing a • (universal) data-base/data-repository and interface • for storing, retrieving, analyzing and displaying • Call-by-Call-based data/information Enables the Study of: - Customers - Service Providers / Agents - Managers / System • Wait Time, Abandonment, Retrials • Service Duration, Activity Profile • Loads,Queue Lengths, Trends

  3. DataMOCCA History: The Data Challenge • Queueing Research lead to Service Operations (early 90’s) • Services started with Call Centers which, in turn, created data-needs • Queueing Theory had to expand to Queueing Science: Fascinating • WFM was Erlang-C based, but customers abandon! (Im)Patience? • (Im)Patience censored hence Call-by-Call data required: 4-5 years saga • Finally Data: a small call center in a small IL bank (15 agents, 4 services) • Technion Stat Lab, guided by Queueing Science: Descriptive Analysis • Building blocks (Arrivals, Services, (Im)Patiece): even more Fascinating

  4. DataMOCCA History: Research & Teaching • Large well-run call centers beyond conventional Queueing Theory: • Both Quality and Efficiency Driven (vs. tradeoff) • Multi-Disciplinary view: OR/OM, HRM (Psychology), Marketing, MIS • Research: Asymptotic analysis of the Palm/Erlang-A model, in the Halfin-Whitt regime = QEDRegime • Teaching: Service Management + Industrial Engineering = Service Engineering / Science • Wharton: Mini-course (OPIM, Morris Cohen) + Seminar (Statistics, Larry Brown + Call Center Forum (Noah Gans) = cooperation with a large(r) banking call center (1000 agents), …

  5. DataMOCCA: Present • System Components: • Clean databases: operational histories of • individual calls, agents and sometimes customers (ID’s). • 2. SEEStat (friendly powerful interface): online access to (mostly) operationa and (some) administrative data; flexible customization, statistics • Currently Three Databases: • US Bank (220/40M calls, 1000 agents, 2.5 years;7-48GB; 3GB DVD). • Israeli Telecom (800 agents, 3.5 years; 55GB; ongoing). • IsraeliBank (500 agents, 1.5 years; ongoing).

  6. Service Engineering: Multi-Disciplinary View Index Function Scientific Discipline Multi-Disciplinary Call Center Design Service Completion (75% in Banks) Organization Design: Parallel (Flat) Sequential (Hierarchical) Information Design Marketing, Operations Research Operations/ Business Process Archive ( Waiting Time Return Time) Lost Calls Sociology, Psychology, Operations Research Experts Agents Queue (Invisible) Redial (Retrial) Database Design Busy (Rare) Computer-Telephony Integration - CTI Job Enrichment Tele-Stress Data Mining: MIS, Statistics,Operations Research,Marketing Training (Turnover up to 200% per Year) (Sweat Shops of the 21th Century) Psychology MIS/CS HRM Arrivals (Business Frontier of the 21th Century) Good or Bad Incentives Game Theory, Economics Service Completion Agents (CSRs) VRU/ IVR Psychological Process Archive Forecasting Internet Chat Email Fax Efficiency Statistics, Human Resource Management (HRM) Skill Based Routing (SBR) Design Expect 3 min Willing 8 min Perceive 15 min Customers Segmentation - CRM Marketing, HRM, Operations Research, MIS Customers Interface Design Quality Back-Office Marketing (If Required 15 min, then Waited 8 min) (If Required 6 min, then Waited 8 min) Human Factors Engineering VIP (Training) VIP Queue Service Process Design Abandonment Psychology,Operations Research, Marketing New Services Design (R&D) Operations Research, Economics, HRM Logistics Psychology, Statistics Lost Calls Operations, Marketing, MIS Positive: Repeat Business Negative: New Complaint

  7. DataMOCCA: Future • 1. Operational (ACD) data with Business (CRM) data, plus Contents • 2.Contact Centers: IVR, Internet, Chats, Emails, … • 3. Daily update (as opposed to montly DVD’s) • 4. Web-access (Research, Applications) • 5. Nurture Research, for example • - Skills-Based Routing: Control, staffing, design, online; HRM • - The Human Factor: Service-anatomy, agents learning, incentives, • 6. Hospitals (already started, with IBM, Haifa hospital): Operational, Human-Factors, Medical & Financial data; RFID’s for flow-tracking

  8. DataMOCCA Interface: SEEStat • - Daily / monthly / yearly reports & flow-charts for a complete operational view. • - Graphs and tables, in customized resolutions (month, days, hours, minutes, seconds) for a variety of (pre-designed) operational measures (arrival rates, abandonment counts, service- and wait-time distribution, utilization profiles,…). • Graphs and tables for new user-defined measures. • Direct access to the raw (cleaned) data: export, import.

  9. VRU 3760 Entries 3716 Exits Abnormal Termination 44 28968 Announce 15 48240 Offered Volume Handled Continued 9130 28953 42103 84645 Message 10394 37723 4247 1387 603 295 42542 Direct 10394 Abandon Short Abandon Cancel Disconnect DataMOCCA – SEEStat Daily Report of April 13, 2004 – Regular Day

  10. VRU 779 Entries 765 Exits Abnormal Termination 14 22650 Announce 3 22157 Offered Volume Handled Continued 22647 2867 19989 46849 Message 4941 18581 2215 709 435 107 26860 Direct 4941 Abandon Short Abandon Cancel Discon nect DataMOCCA – SEEStat Daily Report of April 27, 2004 - Holiday

  11. VRU 5609 Entries 5567 Exits Abnormal Termination 42 Announce 26614 28 62866 26586 Offered Volume Handled Continued 11138 63993 157820 Message 15964 49748 15791 1980 1024 432 93827 Direct 15964 Abandon Short Abandon Cancel Discon nect Interesting Scenarios: Peak Days Daily Report of April 20, 2004 – Heavily Loaded Day

  12. DataMOCCA - SEEStat • Pre-designed Operational Measures: %Abandonment, TSF Time Series

  13. Interesting Scenarios: Platinum Customers (VIP’s ?) • Pre-designed Operational Measures: %Abandonment Time Series

  14. DataMOCCA - SEEStat • Pre-designed Operational Measures: Arrival Rates, 2/2005 Daily Reports

  15. DataMOCCA - SEEStat • Pre-designed Operational Measures: Service Times Histograms Fitting Distribution

  16. DataMOCCA – SEEStat Fitting Mixture of 5 Lognormal Components 1 3 4 2 5

  17. Interesting Scenarios: VRU/IVR Fitting Mixture of 5 Gamma Components

  18. Interesting Scenarios: Service Science • US Bank: Service Time Histograms for Telesales, 2001-3

  19. Interesting Scenarios: Psychological vs. System • Israeli Bank: Histogram of Waiting Time

  20. Interesting Scenarios: Skills-Based-Routing Performance level: Regular versus VIP More Regular get service immediately But VIP customers wait less

  21. Interesting Scenarios: Scenario Analysis • Israeli Call Center: Technical Service • Unhandled Calls on May 24th, 2005 • Agents Status on May 24th, 2005

  22. Data-Based Research: Must (?) & Fun • - Contrast with “EmpOM”: Industry / Company / Survey Data (Social Sciences) • - Converge to: Measure, Model, Validate, Experiment, Refine (Physics, Biology,…) • - Prerequisites: OR/OM, (Marketing) – for Design; Computer Science,Information Systems, Statistics – for Implementation. • - Outcomes: Relevance, Credibility, Interest; Pilot (eg. Healthcare). Moreover, • Teaching: Class, Homework (Experimental Data Analysis); Cases. • Research: Test (Queueing) Theory / Laws, Stimulate New Models / Theory. • Practice: OM Tools (Scenario Analysis), Mktg (Trends, Benchmarking).

More Related