Service Engineering:Data-Based Research and Teachingin support of Service Management

Avi(shai) Mandelbaum

Technion, Haifa, Israel

http://ie.technion.ac.il/serveng

ENBIS - Athens, September 2008

1

Background Material (Downloadable)

Technion’s “Service-Engineering" Course ( ≥ 1993):http://ie.technion.ac.il/ServEng

Technion’s SEE Center / Laboratory (≥ 2007):http://ie.technion.ac.il/Labs/ServEng

I Gans (U.S.A.), Koole (Europe), and M. (Israel):“Telephone Call Centers: Tutorial, Review and ResearchProspects." MSOM, 2003.

I Brown, Gans, M., Sakov, Shen, Zeltyn, Zhao:“Statistical Analysis of a Telephone Call Center: AQueueing-Science Perspective." JASA, 2005.

2

Background Material (Downloadable)

Technion’s “Service-Engineering" Course ( ≥ 1993):http://ie.technion.ac.il/ServEng

Technion’s SEE Center / Laboratory (≥ 2007):http://ie.technion.ac.il/Labs/ServEng

I Gans (U.S.A.), Koole (Europe), and M. (Israel):“Telephone Call Centers: Tutorial, Review and ResearchProspects." MSOM, 2003.

I Brown, Gans, M., Sakov, Shen, Zeltyn, Zhao:“Statistical Analysis of a Telephone Call Center: AQueueing-Science Perspective." JASA, 2005.

2

Background Material (Downloadable)

Technion’s “Service-Engineering" Course ( ≥ 1993):http://ie.technion.ac.il/ServEng

Technion’s SEE Center / Laboratory (≥ 2007):http://ie.technion.ac.il/Labs/ServEng

I Gans (U.S.A.), Koole (Europe), and M. (Israel):“Telephone Call Centers: Tutorial, Review and ResearchProspects." MSOM, 2003.

I Brown, Gans, M., Sakov, Shen, Zeltyn, Zhao:“Statistical Analysis of a Telephone Call Center: AQueueing-Science Perspective." JASA, 2005.

2

Evolution

I Math., Computer Science, Operations Research, Statistics

I Queueing TheoryI Application: Service OperationsI Specialization: Telephone Call Centers

I Graduate Research Seminar: Service NetworksI Elective Theoretical course: joint graduate-undergraduateI Elective Theoretical + Data-Based: Service EngineeringI Core Course

I Technion’s SEE Center/Lab (Service Enterprise Engineering)I Further Applications: Healthcare, Internet,. . .

3

Evolution

I Math., Computer Science, Operations Research, Statistics

I Queueing TheoryI Application: Service OperationsI Specialization: Telephone Call Centers

I Graduate Research Seminar: Service NetworksI Elective Theoretical course: joint graduate-undergraduateI Elective Theoretical + Data-Based: Service EngineeringI Core Course

I Technion’s SEE Center/Lab (Service Enterprise Engineering)I Further Applications: Healthcare, Internet,. . .

3

Evolution

I Math., Computer Science, Operations Research, Statistics

I Queueing TheoryI Application: Service OperationsI Specialization: Telephone Call Centers

I Graduate Research Seminar: Service NetworksI Elective Theoretical course: joint graduate-undergraduateI Elective Theoretical + Data-Based: Service EngineeringI Core Course

I Technion’s SEE Center/Lab (Service Enterprise Engineering)I Further Applications: Healthcare, Internet,. . .

3

Evolution

I Math., Computer Science, Operations Research, Statistics

I Queueing TheoryI Application: Service OperationsI Specialization: Telephone Call Centers

I Graduate Research Seminar: Service NetworksI Elective Theoretical course: joint graduate-undergraduateI Elective Theoretical + Data-Based: Service EngineeringI Core Course

I Technion’s SEE Center/Lab (Service Enterprise Engineering)I Further Applications: Healthcare, Internet,. . .

3

Technion (Economy) Evolution

4

Main Messages

1. Simple Useful Models at the Service of Complex Realities.

Note: Useful must be Simple; Simple could require Deep analysis

2. Data-Based Analysis & Teaching is a Must & Fun.Supported by DataMOCCA = Data MOdels for Call Center Analysis;Technion + Wharton research project, available for (academic)adoption. (eg. 2.5 years, 220M/40M telephone calls, 800 agents)

3. Back to the Basic-Research Paradigm (Physics, Biology, . . .):Measure, Model, Experiment, Validate, Refine, etc.

4. Yields scientifically-based design principles and tools(software), that support the balance of service quality, processefficiency and business profitability, from the often-conflicting viewsof customers, service-providers, managers and society:Service Engineering .

5

Main Messages

1. Simple Useful Models at the Service of Complex Realities.Note: Useful must be Simple; Simple could require Deep analysis

2. Data-Based Analysis & Teaching is a Must & Fun.Supported by DataMOCCA = Data MOdels for Call Center Analysis;Technion + Wharton research project, available for (academic)adoption. (eg. 2.5 years, 220M/40M telephone calls, 800 agents)

3. Back to the Basic-Research Paradigm (Physics, Biology, . . .):Measure, Model, Experiment, Validate, Refine, etc.

4. Yields scientifically-based design principles and tools(software), that support the balance of service quality, processefficiency and business profitability, from the often-conflicting viewsof customers, service-providers, managers and society:Service Engineering .

5

Main Messages

1. Simple Useful Models at the Service of Complex Realities.Note: Useful must be Simple; Simple could require Deep analysis

2. Data-Based Analysis & Teaching is a Must & Fun.Supported by DataMOCCA = Data MOdels for Call Center Analysis;Technion + Wharton research project, available for (academic)adoption. (eg. 2.5 years, 220M/40M telephone calls, 800 agents)

3. Back to the Basic-Research Paradigm (Physics, Biology, . . .):Measure, Model, Experiment, Validate, Refine, etc.

4. Yields scientifically-based design principles and tools(software), that support the balance of service quality, processefficiency and business profitability, from the often-conflicting viewsof customers, service-providers, managers and society:Service Engineering .

5

Main Messages

1. Simple Useful Models at the Service of Complex Realities.Note: Useful must be Simple; Simple could require Deep analysis

2. Data-Based Analysis & Teaching is a Must & Fun.Supported by DataMOCCA = Data MOdels for Call Center Analysis;Technion + Wharton research project, available for (academic)adoption. (eg. 2.5 years, 220M/40M telephone calls, 800 agents)

3. Back to the Basic-Research Paradigm (Physics, Biology, . . .):Measure, Model, Experiment, Validate, Refine, etc.

4. Yields scientifically-based design principles and tools(software), that support the balance of service quality, processefficiency and business profitability, from the often-conflicting viewsof customers, service-providers, managers and society:Service Engineering .

5

Main Messages

1. Simple Useful Models at the Service of Complex Realities.Note: Useful must be Simple; Simple could require Deep analysis

2. Data-Based Analysis & Teaching is a Must & Fun.Supported by DataMOCCA = Data MOdels for Call Center Analysis;Technion + Wharton research project, available for (academic)adoption. (eg. 2.5 years, 220M/40M telephone calls, 800 agents)

3. Back to the Basic-Research Paradigm (Physics, Biology, . . .):Measure, Model, Experiment, Validate, Refine, etc.

4. Yields scientifically-based design principles and tools(software), that support the balance of service quality, processefficiency and business profitability, from the often-conflicting viewsof customers, service-providers, managers and society:Service Engineering .

5

Staffing: How Many? When? What? Who?

Fundamental challenge in Services: Healthcare, . . . , Call Centers

I Reality Complex and becoming more soI Staffing is based on The Erlang-C (M/M/n) model (1913!)

=⇒ Solutions urgently needed.

Consider, for example, Palm/Erlang-A: a simple (but not too simple)Mathematical Model of the complex reality of call centers.

6

Staffing: How Many? When? What? Who?

Fundamental challenge in Services: Healthcare, . . . , Call Centers

I Reality Complex and becoming more soI Staffing is based on The Erlang-C (M/M/n) model (1913!)

=⇒ Solutions urgently needed.

Consider, for example, Palm/Erlang-A: a simple (but not too simple)Mathematical Model of the complex reality of call centers.

6

Staffing: How Many? When? What? Who?

Fundamental challenge in Services: Healthcare, . . . , Call Centers

I Reality Complex and becoming more soI Staffing is based on The Erlang-C (M/M/n) model (1913!)

=⇒ Solutions urgently needed.

Consider, for example, Palm/Erlang-A: a simple (but not too simple)Mathematical Model of the complex reality of call centers.

6

Complex Reality: Call-Center Network

Agents(CSRs)

Back-Office

Experts)(Consultants

VIP)Training(

Arrivals(Business Frontier

of the21th Century)

Redial(Retrial)

Busy)Rare(

Goodor

Bad

Positive: Repeat BusinessNegative: New Complaint

Lost Calls

Abandonment

Agents

ServiceCompletion

Service Engineering: Multi-Disciplinary Process View

ForecastingStatistics

New Services Design (R&D)Operations,Marketing

Organization Design:Parallel (Flat)Sequential (Hierarchical)Sociology/Psychology,Operations Research

Human Resource Management

Service Process Design

To Avoid Delay

To Avoid Starvation Skill Based Routing

(SBR) DesignMarketing,Human Resources,Operations Research,MIS

Customers Interface Design

Computer-Telephony Integration - CTIMIS/CS

Marketing

Operations/BusinessProcessArchiveDatabaseDesignData Mining:MIS, Statistics, Operations Research, Marketing

InternetChatEmailFax

Lost Calls

Service Completion)75% in Banks(

( Waiting TimeReturn Time)

Logistics

Customers Segmentation -CRM

Psychology, Operations Research,Marketing

Expect 3 minWilling 8 minPerceive 15 min

PsychologicalProcessArchive

Psychology,Statistics

Training, IncentivesJob Enrichment

Marketing,Operations Research

Human Factors Engineering

VRU/IVR

Queue)Invisible(

VIP Queue

(If Required 15 min,then Waited 8 min)(If Required 6 min, then Waited 6 min)

Information DesignFunctionScientific DisciplineMulti-Disciplinary

IndexCall Center Design

(Turnover up to 200% per Year)(Sweat Shops

of the21th Century)

Tele-StressPsychology

7

Simple Model: Palm/Erlang-Aagents

arrivals

abandonment

λ

µ

1

2

n

…

queue

θ

Erlang-A Parameters (Math. Assumptions):

I λ – Arrival rate (Poisson)I µ – Service rate (Exponential)I θ – Impatience rate (Exponential)I n – Number of Service-Agents.

8

Arrivals to Service: Poisson-RelativesArrival Rates on Tuesdays in a September – U.S. Bank

0

500

1000

1500

2000

2500

3000

3500

0:00 2:00 4:00 6:00 8:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00

Time (Resolution 30 min.)

Arrival rate

04.09.2001 11.09.2001 18.09.2001 25.09.2001

I Tuesday, September 4th: Heavy, following Labor DayI Tuesdays, September 18 & 25: Normal

I Tuesday, September 11th, 2001.

9

Arrivals to Service: Poisson-RelativesArrival Rates on Tuesdays in a September – U.S. Bank

0

500

1000

1500

2000

2500

3000

3500

0:00 2:00 4:00 6:00 8:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00

Time (Resolution 30 min.)

Arrival rate

04.09.2001 11.09.2001 18.09.2001 25.09.2001

I Tuesday, September 4th: Heavy, following Labor DayI Tuesdays, September 18 & 25: NormalI Tuesday, September 11th, 2001.

9

Service Durations: The LogNormal Law

Service Durations in a Typical (?) Call Center

January-October November-December

Beyond Data Averages Short Service Times

AVG: 200 STD: 249

AVG: 185 STD: 238

7.2 % ? Jan – Oct:

Log-Normal AVG: 200 STD: 249

Nov – Dec:

27

Beyond Data Averages Short Service Times

AVG: 200 STD: 249

AVG: 185 STD: 238

7.2 % ? Jan – Oct:

Log-Normal AVG: 200 STD: 249

Nov – Dec:

27

I Lognormal service times prevalent in call centers

I 7.2% Short-Services: Agents’ “Abandon" (improve bonus, rest)I Distributions, not only Averages, must be measured.

10

Service Durations: The LogNormal Law

Service Durations in a Typical (?) Call Center

January-October November-December

Beyond Data Averages Short Service Times

AVG: 200 STD: 249

AVG: 185 STD: 238

7.2 % ? Jan – Oct:

Log-Normal AVG: 200 STD: 249

Nov – Dec:

27

Beyond Data Averages Short Service Times

AVG: 200 STD: 249

AVG: 185 STD: 238

7.2 % ? Jan – Oct:

Log-Normal AVG: 200 STD: 249

Nov – Dec:

27

I Lognormal service times prevalent in call centersI 7.2% Short-Services: Agents’ “Abandon" (improve bonus, rest)I Distributions, not only Averages, must be measured.

10

(Im)Patience While Waiting: Palm’s Law of Irritation

Hazard Rates of (Im)Patience – Israeli Bank:Regular over VIP Customers

14

16

I Peaks of abandonment at times of AnnouncementsI VIP are more Patient (Needy) than the OthersI Call-by-Call Data (DataMOCCA) required (Un-Censoring).

11

(Im)Patience While Waiting: Palm’s Law of Irritation

Hazard Rates of (Im)Patience – Israeli Bank:Regular over VIP Customers

14

16

I Peaks of abandonment at times of AnnouncementsI VIP are more Patient (Needy) than the OthersI Call-by-Call Data (DataMOCCA) required (Un-Censoring).

11

Erlang-A: Simple, Useful, Robust, Insightful, Optimal

I Simple: 4CallCenters calculator (download in our Website)

All Palm/Erlang-A assumptions are violated.Yet the model often fits very well, so much so that the model is

I Useful: Replaces Erlang-C as the WFM standard

I Robust: QED asymptotics (moderate-to-large systems)

I Insightful: Square-Root Staffing rules; E.O.S.

I Optimal: Could save significant $’s

I and Generalizable: Time-Varying, CRM/SBR, . . .,

Still has its Limitations, theoretical & practical, all of which simulates

=⇒ Current Research

12

Erlang-A: Simple, Useful, Robust, Insightful, Optimal

I Simple: 4CallCenters calculator (download in our Website)

All Palm/Erlang-A assumptions are violated.Yet the model often fits very well, so much so that the model is

I Useful: Replaces Erlang-C as the WFM standard

I Robust: QED asymptotics (moderate-to-large systems)

I Insightful: Square-Root Staffing rules; E.O.S.

I Optimal: Could save significant $’s

I and Generalizable: Time-Varying, CRM/SBR, . . .,

Still has its Limitations, theoretical & practical, all of which simulates

=⇒ Current Research

12

Erlang-A: Simple, Useful, Robust, Insightful, Optimal

I Simple: 4CallCenters calculator (download in our Website)

All Palm/Erlang-A assumptions are violated.Yet the model often fits very well, so much so that the model is

I Useful: Replaces Erlang-C as the WFM standard

I Robust: QED asymptotics (moderate-to-large systems)

I Insightful: Square-Root Staffing rules; E.O.S.

I Optimal: Could save significant $’s

I and Generalizable: Time-Varying, CRM/SBR, . . .,

Still has its Limitations, theoretical & practical, all of which simulates

=⇒ Current Research

12

Erlang-A: Simple, Useful, Robust, Insightful, Optimal

I Simple: 4CallCenters calculator (download in our Website)

All Palm/Erlang-A assumptions are violated.Yet the model often fits very well, so much so that the model is

I Useful: Replaces Erlang-C as the WFM standard

I Robust: QED asymptotics (moderate-to-large systems)

I Insightful: Square-Root Staffing rules; E.O.S.

I Optimal: Could save significant $’s

I and Generalizable: Time-Varying, CRM/SBR, . . .,

Still has its Limitations, theoretical & practical, all of which simulates

=⇒ Current Research

12

Back to Main Messages: Summary of Erlang-A

1. Simple useful model, requiring and stimulating deep analysis.

2. Supported by Data-Based research & teaching.(DataMOCCA, available for (academic) adoption.)

3. Takes one back to the basic-research paradigm:Measure, Model, Experiment, Validate, Refine, etc.

4. Generates scientifically-based design principles, tools(software) and teaching material, downloadable at the

Service-Engineering Course websitehttp://ie.technnion.ac.il/ServEng

SEE Center/Laboratory websitehttp://ie.technion.ac.il/Labs/ServEng

13