SlideShare a Scribd company logo
1 of 18
SYST 542 Paper Presentation:
John Erickson
Fall 2016
From Magnum Opus to Mea Culpa:
A Cautionary Tale of Lessons Learned from a
Failed Decision Support System
Outline
• Introduction
– Critical Success Factors: class reading, David Arnott
– Literature on DSS
• Synopsis of Case Study / Paper
– DSS Justification
– DSS Goals
– DSS Development
– DSS Design
– Model Results
– Upgrade Results
• What Went Wrong
– The Catalyst of the Cataclysm
– The “Postmortem”
– The Mistakes/ Lessons Learned
• Conclusion
Introduction
Why study failures and lessons learned?
1. High Failure Rate: “It is well known in the DSS literature
that all types of DSS projects are high-risk and prone to
failure…Some studies have even reported failure rates as
high as 80%.” (Arnott class paper)
2. Lack of Literature: Gorman claims there are very few
articles on DSS failure: “If one were to review the OR/MS
literature on DSS implementations, that reviewer would
be hard pressed to find published examples of failure.”
Handbook of decision support systems, vol.
1: Basic Themes: Chapter 34, David Arnott
What makes a DSS successful?
*Handbook of decision support systems, vol. 1:
Basic Themes: Chapter 34, David Arnott
Critical Success Factor Model (CSF) Method10 Critical Success Factors
1. Committed and informed executive
sponsor
2. Widespread management support
3. Appropriate team skills
4. Appropriate technology
5. Adequate resources
6. Effective data management
7. Clear link with business
8. Well-defined information and
systems requirements
9. Evolutionary development
10. Management of project scope
“The implication of the table is that if a
reasonable number of CSF’s are not attained
or achieved, a project is likely to fail.”*
Synopsis of Paper
• Michael Gorman, PhD is a management- sciences
university professor and consultant in the freight /
railway industry
• He was hired to build a DSS for “ACME”, a growing
freight transportation company that had volume
growth in the early 2000s (ACME is an alias for
confidentiality).
• It was initially successful, then the DSS failed due to
the Wall Street crisis in 2009
• His paper outlines the conditions for the need of the
DSS, it’s early success, downfall, and then lessons
learned.
ACME’s System Justification
• Why did ACME Build the DSS?
– Previous methods for load acceptance and dispatching
failed to achieve desired economic results
– Frontline decision makers lacked the data, analytical ability,
and global perspective to manage a network that was
growing in complexity.
– Thus, ACME had a strong incentive to change its approach.
“Because of its growing volume and asset base, in 2005, a senior
executive at the company realized that ACME had the scale to
justify a DSS that would support decision makers in the face of
changing fundamental economics resulting from increased asset
intensity. ..”
DSS Goals
• DSS Goals:
1. Support real-time decision making at a
company with annual revenues of $3 billion by
cutting across both marketing and operations
2. Replace suboptimal rules for decision making
(“outdated and myopic” rules)
System Development
• Senior management:
– provided input on system philosophy , design, and marketing
• Mid Level Management:
– operations managers and frontline staff documented the
business decision making process side
• IT professionals:
– established system integration and other technical
requirements.
In short, the DSS had to be integrated in such a way as to minimize
legacy system development costs and disruption to ongoing
operations.
System Design
Orders (These historical
orders were used for
demand forecasting)
Economic Data (Historical
cost, price, and margin
data were used for
predicting profitability)
Service-Time Data
(service times of
transportation providers
affect future supply
Input Output
Supply-and-demand
forecast (supply forecast
predicted equipment
capacity ; demand
forecast predicted daily
customer load tenders)
Load-accept optimization
Dispatch optimization
(load to- equipment
assignment)
-The DSS produced then a ranking of best ways to load/dispatch, based on the lowest overall
cost, including both current conditions at the origin and future conditions at the destination
and the network effects of other assignments to be made.
-It ran without interruption each day for more than two years for forecasting and load
acceptance, and every 15 minutes for load dispatching.
Model Results & Upgrade
Early success
• Model recommendations were generally well received by frontline decision
makers.
• The model recommendations were presented in a similar way as before, as ranked
recommendations, but with more sophisticated model-based rankings.
Motivated by early success, the model was improved:
• Support modules
• More detailed forecasting and dispatching
• These improvements came at a cost—increased model complexity
-The shadow price of violating model-based recommendations was given, showing the
opportunity cost of not following recommendations.
-Global optimization results were presented to dispatchers making sequential decisions
-Any change in supply, demand, or non-model recommended assignments would be
reflected in subsequent model runs
- Current best was always available to dispatchers, given prior events.
What Model Presented in Real-Time:
Upgrade Results (post-upgrade)
• Preliminary results indicated higher revenues, higher service
levels to the best-paying customers, lower costs, and better
asset utilization (i.e., lower excess inventory).
• First year results: formal internal audit after the model’s first
year in production, the results indicated a number of
performance improvements:
– higher asset velocity (more loads per time period),
– a decrease in low-margin loads
– an increase in average margins
– reduction in cost per load.
In the model’s first year, the audit estimated savings of more
than 20 times the system development cost.
Then Things Went South…
• The Perfect Storm:
– Financial credit crisis on Wall Street in 2009-2010
– ACME’s cutting discretionary spending on consulting in 2010
• DSS was shut down and revert back to the simpler but safer
rules of load acceptance and dispatch.
• Management became skeptical of DSS
• Gorman felt the need to catalog the lessons learned and why
his “magnum opus” ultimately failed
Bottom Line:
The beginning of the end was this: the Wall Street credit crisis
caused a perturbation that the DSS couldn’t handle
The Central Failure in the DSS
• The biggest mistake was an implicit assumption surrounding the structural stability of the
demand forecast.
• Although recent low-demand history helped to pull forecasts downward, previous years’
demand levels served to inflate forecasts.
The underlying probability distribution used for the
optimization model turned out to not be an accurate
distribution when the credit crisis hit
“Postmortem”
1. Implicit Assumption:
Demand Forecast Structural
Stability
2. Implicit Assumption: Cost
Structures
3. Too many decision makers
4. Too many parameters
5. Too much data
6. Too Little Knowledge:
Insufficient Skills Transfer
7. Too Little Time Allocated:
Failed Vigilance in Ongoing
DSS Operation
1. More careful and proactive tracking: analysis of
recent forecast errors would have reduced
prolonged use of a consistently high forecast
after the downturn.
2. Be leery of using historical data in the model
3. Limit scope creep: Trying to be “all things to all
people”
4. Limit model parameters: by striving for
perfection, model usefulness and robustness can
often be diminished
5. Robust support staff: Have sufficient support
staff to the data infrastructure model.
6. Ongoing training, problem solving (live case
studies),and troubleshooting between developer
and client
7. Project Manager: a DSS owner, who is
responsible for DSS performance, responding to
customer problems, and uncovering previously
undiscovered anomalies, is necessary.
Mistakes Lessons Learned
Current Situation – No DSS
“The fear and mistrust of optimization
techniques subsequent to this project, coupled
with lack of tolerance for an association with
the model failure, leave little appetite for
advanced analytics to support decision making
at ACME.”
-Gorman
Author’s Final Conclusions
• Beware of implicit assumptions: Models inherently assume
historical data patterns as a guide to future decisions.
• Minimize complexity wherever possible.
• Socialize, train, and experiment with DSS models.
• Maintain constant vigilance of model performance.
“Modeling is never done. When a client asks when the
project will be done, the best reply is “never”; follow this
with an explanation that tells the client that the first version
will begin delivering recommendations by a specific date, at
which time the team will want to continue to assess and
improve on the recommendations through more accurate
data and modeling.”
Takeaways from Paper
Takeway 1: Think about your DSS: A lot of thought should be put into
the scope of the DSS project based on the Critical Success Factors. A
bigger project may not be worth it unless the willpower, resources,
and executives are all on-board.
Takeaway 2: Even if the DSS meets all/most of the CSF, it takes
constant vigilance to keep/upgrade/test/validate the system. This is
a sobering thought for tech/internet startups!!
Critique:
I would have preferred that the author (Gorman) used these 10 CSF (or
a similar metric) throughout his paper so that the reader can get a
measurable sense of what went wrong according to the CSF. Also, use
the CSF’s and compare over time (6-month increments) where the DSS
stands against the CSF.
Questions ?

More Related Content

Similar to Syst 542 paper presentation erickson final

Crisp dm
Crisp dmCrisp dm
Crisp dmakbkck
 
Introduction to Agile Project Management
Introduction to Agile Project ManagementIntroduction to Agile Project Management
Introduction to Agile Project ManagementSemen Arslan
 
Meeting the OTT challenge
Meeting the OTT challengeMeeting the OTT challenge
Meeting the OTT challengeMartin Geddes
 
Using periodic audits to prevent catastrophic project failure
Using periodic audits to prevent catastrophic project failureUsing periodic audits to prevent catastrophic project failure
Using periodic audits to prevent catastrophic project failureicgfmconference
 
Final spiralmodel97
Final spiralmodel97Final spiralmodel97
Final spiralmodel97akshay8835
 
Data-Ed Webinar: Best Practices with the DMM
Data-Ed Webinar: Best Practices with the DMMData-Ed Webinar: Best Practices with the DMM
Data-Ed Webinar: Best Practices with the DMMDATAVERSITY
 
Data Ed: Best Practices with the DMM
Data Ed: Best Practices with the DMMData Ed: Best Practices with the DMM
Data Ed: Best Practices with the DMMData Blueprint
 
Alternative Methodologies for Systems Development
Alternative Methodologies for Systems Development Alternative Methodologies for Systems Development
Alternative Methodologies for Systems Development Sunderland City Council
 
If You Are Not Embedding Analytics Into Your Day To Day Processes, You Are Do...
If You Are Not Embedding Analytics Into Your Day To Day Processes, You Are Do...If You Are Not Embedding Analytics Into Your Day To Day Processes, You Are Do...
If You Are Not Embedding Analytics Into Your Day To Day Processes, You Are Do...Dell World
 
Doing Analytics Right - Building the Analytics Environment
Doing Analytics Right - Building the Analytics EnvironmentDoing Analytics Right - Building the Analytics Environment
Doing Analytics Right - Building the Analytics EnvironmentTasktop
 
Hanno Jarvet - VSM, Planning and Problem Solving - ConFu
Hanno Jarvet - VSM, Planning and Problem Solving - ConFuHanno Jarvet - VSM, Planning and Problem Solving - ConFu
Hanno Jarvet - VSM, Planning and Problem Solving - ConFuDevConFu
 
Management information system prepared by samena
Management information system prepared by samenaManagement information system prepared by samena
Management information system prepared by samenasamena shawon
 

Similar to Syst 542 paper presentation erickson final (20)

Crisp dm
Crisp dmCrisp dm
Crisp dm
 
Introduction to Agile Project Management
Introduction to Agile Project ManagementIntroduction to Agile Project Management
Introduction to Agile Project Management
 
Meeting the OTT challenge
Meeting the OTT challengeMeeting the OTT challenge
Meeting the OTT challenge
 
ml-02x01.pdf
ml-02x01.pdfml-02x01.pdf
ml-02x01.pdf
 
Quality, requirements and success webinar, 6 March 2019
Quality, requirements and success webinar, 6 March 2019Quality, requirements and success webinar, 6 March 2019
Quality, requirements and success webinar, 6 March 2019
 
Using periodic audits to prevent catastrophic project failure
Using periodic audits to prevent catastrophic project failureUsing periodic audits to prevent catastrophic project failure
Using periodic audits to prevent catastrophic project failure
 
Final spiralmodel97
Final spiralmodel97Final spiralmodel97
Final spiralmodel97
 
Data-Ed Webinar: Best Practices with the DMM
Data-Ed Webinar: Best Practices with the DMMData-Ed Webinar: Best Practices with the DMM
Data-Ed Webinar: Best Practices with the DMM
 
Data Ed: Best Practices with the DMM
Data Ed: Best Practices with the DMMData Ed: Best Practices with the DMM
Data Ed: Best Practices with the DMM
 
Alternative Methodologies for Systems Development
Alternative Methodologies for Systems Development Alternative Methodologies for Systems Development
Alternative Methodologies for Systems Development
 
MIS Wk-10.ppt
MIS Wk-10.pptMIS Wk-10.ppt
MIS Wk-10.ppt
 
If You Are Not Embedding Analytics Into Your Day To Day Processes, You Are Do...
If You Are Not Embedding Analytics Into Your Day To Day Processes, You Are Do...If You Are Not Embedding Analytics Into Your Day To Day Processes, You Are Do...
If You Are Not Embedding Analytics Into Your Day To Day Processes, You Are Do...
 
Doing Analytics Right - Building the Analytics Environment
Doing Analytics Right - Building the Analytics EnvironmentDoing Analytics Right - Building the Analytics Environment
Doing Analytics Right - Building the Analytics Environment
 
3.9 báo cáo khuyến nghị
3.9 báo cáo khuyến nghị3.9 báo cáo khuyến nghị
3.9 báo cáo khuyến nghị
 
Scm future2
Scm future2Scm future2
Scm future2
 
Modeling and analysis
Modeling and analysisModeling and analysis
Modeling and analysis
 
Hanno Jarvet - VSM, Planning and Problem Solving - ConFu
Hanno Jarvet - VSM, Planning and Problem Solving - ConFuHanno Jarvet - VSM, Planning and Problem Solving - ConFu
Hanno Jarvet - VSM, Planning and Problem Solving - ConFu
 
Management information system prepared by samena
Management information system prepared by samenaManagement information system prepared by samena
Management information system prepared by samena
 
Lec 24
Lec 24Lec 24
Lec 24
 
Comp587_SEI_CMM.ppt
Comp587_SEI_CMM.pptComp587_SEI_CMM.ppt
Comp587_SEI_CMM.ppt
 

Recently uploaded

High Profile Call Girls Nagpur Isha Call 7001035870 Meet With Nagpur Escorts
High Profile Call Girls Nagpur Isha Call 7001035870 Meet With Nagpur EscortsHigh Profile Call Girls Nagpur Isha Call 7001035870 Meet With Nagpur Escorts
High Profile Call Girls Nagpur Isha Call 7001035870 Meet With Nagpur Escortsranjana rawat
 
Sheet Pile Wall Design and Construction: A Practical Guide for Civil Engineer...
Sheet Pile Wall Design and Construction: A Practical Guide for Civil Engineer...Sheet Pile Wall Design and Construction: A Practical Guide for Civil Engineer...
Sheet Pile Wall Design and Construction: A Practical Guide for Civil Engineer...Dr.Costas Sachpazis
 
Gfe Mayur Vihar Call Girls Service WhatsApp -> 9999965857 Available 24x7 ^ De...
Gfe Mayur Vihar Call Girls Service WhatsApp -> 9999965857 Available 24x7 ^ De...Gfe Mayur Vihar Call Girls Service WhatsApp -> 9999965857 Available 24x7 ^ De...
Gfe Mayur Vihar Call Girls Service WhatsApp -> 9999965857 Available 24x7 ^ De...srsj9000
 
Porous Ceramics seminar and technical writing
Porous Ceramics seminar and technical writingPorous Ceramics seminar and technical writing
Porous Ceramics seminar and technical writingrakeshbaidya232001
 
Coefficient of Thermal Expansion and their Importance.pptx
Coefficient of Thermal Expansion and their Importance.pptxCoefficient of Thermal Expansion and their Importance.pptx
Coefficient of Thermal Expansion and their Importance.pptxAsutosh Ranjan
 
Call for Papers - African Journal of Biological Sciences, E-ISSN: 2663-2187, ...
Call for Papers - African Journal of Biological Sciences, E-ISSN: 2663-2187, ...Call for Papers - African Journal of Biological Sciences, E-ISSN: 2663-2187, ...
Call for Papers - African Journal of Biological Sciences, E-ISSN: 2663-2187, ...Christo Ananth
 
GDSC ASEB Gen AI study jams presentation
GDSC ASEB Gen AI study jams presentationGDSC ASEB Gen AI study jams presentation
GDSC ASEB Gen AI study jams presentationGDSCAESB
 
Introduction to IEEE STANDARDS and its different types.pptx
Introduction to IEEE STANDARDS and its different types.pptxIntroduction to IEEE STANDARDS and its different types.pptx
Introduction to IEEE STANDARDS and its different types.pptxupamatechverse
 
MANUFACTURING PROCESS-II UNIT-5 NC MACHINE TOOLS
MANUFACTURING PROCESS-II UNIT-5 NC MACHINE TOOLSMANUFACTURING PROCESS-II UNIT-5 NC MACHINE TOOLS
MANUFACTURING PROCESS-II UNIT-5 NC MACHINE TOOLSSIVASHANKAR N
 
Extrusion Processes and Their Limitations
Extrusion Processes and Their LimitationsExtrusion Processes and Their Limitations
Extrusion Processes and Their Limitations120cr0395
 
the ladakh protest in leh ladakh 2024 sonam wangchuk.pptx
the ladakh protest in leh ladakh 2024 sonam wangchuk.pptxthe ladakh protest in leh ladakh 2024 sonam wangchuk.pptx
the ladakh protest in leh ladakh 2024 sonam wangchuk.pptxhumanexperienceaaa
 
What are the advantages and disadvantages of membrane structures.pptx
What are the advantages and disadvantages of membrane structures.pptxWhat are the advantages and disadvantages of membrane structures.pptx
What are the advantages and disadvantages of membrane structures.pptxwendy cai
 
Introduction to Multiple Access Protocol.pptx
Introduction to Multiple Access Protocol.pptxIntroduction to Multiple Access Protocol.pptx
Introduction to Multiple Access Protocol.pptxupamatechverse
 
chaitra-1.pptx fake news detection using machine learning
chaitra-1.pptx  fake news detection using machine learningchaitra-1.pptx  fake news detection using machine learning
chaitra-1.pptx fake news detection using machine learningmisbanausheenparvam
 
Software Development Life Cycle By Team Orange (Dept. of Pharmacy)
Software Development Life Cycle By  Team Orange (Dept. of Pharmacy)Software Development Life Cycle By  Team Orange (Dept. of Pharmacy)
Software Development Life Cycle By Team Orange (Dept. of Pharmacy)Suman Mia
 
(ANVI) Koregaon Park Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...
(ANVI) Koregaon Park Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...(ANVI) Koregaon Park Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...
(ANVI) Koregaon Park Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...ranjana rawat
 
High Profile Call Girls Nagpur Meera Call 7001035870 Meet With Nagpur Escorts
High Profile Call Girls Nagpur Meera Call 7001035870 Meet With Nagpur EscortsHigh Profile Call Girls Nagpur Meera Call 7001035870 Meet With Nagpur Escorts
High Profile Call Girls Nagpur Meera Call 7001035870 Meet With Nagpur EscortsCall Girls in Nagpur High Profile
 
VIP Call Girls Service Hitech City Hyderabad Call +91-8250192130
VIP Call Girls Service Hitech City Hyderabad Call +91-8250192130VIP Call Girls Service Hitech City Hyderabad Call +91-8250192130
VIP Call Girls Service Hitech City Hyderabad Call +91-8250192130Suhani Kapoor
 

Recently uploaded (20)

High Profile Call Girls Nagpur Isha Call 7001035870 Meet With Nagpur Escorts
High Profile Call Girls Nagpur Isha Call 7001035870 Meet With Nagpur EscortsHigh Profile Call Girls Nagpur Isha Call 7001035870 Meet With Nagpur Escorts
High Profile Call Girls Nagpur Isha Call 7001035870 Meet With Nagpur Escorts
 
Sheet Pile Wall Design and Construction: A Practical Guide for Civil Engineer...
Sheet Pile Wall Design and Construction: A Practical Guide for Civil Engineer...Sheet Pile Wall Design and Construction: A Practical Guide for Civil Engineer...
Sheet Pile Wall Design and Construction: A Practical Guide for Civil Engineer...
 
Gfe Mayur Vihar Call Girls Service WhatsApp -> 9999965857 Available 24x7 ^ De...
Gfe Mayur Vihar Call Girls Service WhatsApp -> 9999965857 Available 24x7 ^ De...Gfe Mayur Vihar Call Girls Service WhatsApp -> 9999965857 Available 24x7 ^ De...
Gfe Mayur Vihar Call Girls Service WhatsApp -> 9999965857 Available 24x7 ^ De...
 
Porous Ceramics seminar and technical writing
Porous Ceramics seminar and technical writingPorous Ceramics seminar and technical writing
Porous Ceramics seminar and technical writing
 
Exploring_Network_Security_with_JA3_by_Rakesh Seal.pptx
Exploring_Network_Security_with_JA3_by_Rakesh Seal.pptxExploring_Network_Security_with_JA3_by_Rakesh Seal.pptx
Exploring_Network_Security_with_JA3_by_Rakesh Seal.pptx
 
Coefficient of Thermal Expansion and their Importance.pptx
Coefficient of Thermal Expansion and their Importance.pptxCoefficient of Thermal Expansion and their Importance.pptx
Coefficient of Thermal Expansion and their Importance.pptx
 
Call for Papers - African Journal of Biological Sciences, E-ISSN: 2663-2187, ...
Call for Papers - African Journal of Biological Sciences, E-ISSN: 2663-2187, ...Call for Papers - African Journal of Biological Sciences, E-ISSN: 2663-2187, ...
Call for Papers - African Journal of Biological Sciences, E-ISSN: 2663-2187, ...
 
GDSC ASEB Gen AI study jams presentation
GDSC ASEB Gen AI study jams presentationGDSC ASEB Gen AI study jams presentation
GDSC ASEB Gen AI study jams presentation
 
Introduction to IEEE STANDARDS and its different types.pptx
Introduction to IEEE STANDARDS and its different types.pptxIntroduction to IEEE STANDARDS and its different types.pptx
Introduction to IEEE STANDARDS and its different types.pptx
 
MANUFACTURING PROCESS-II UNIT-5 NC MACHINE TOOLS
MANUFACTURING PROCESS-II UNIT-5 NC MACHINE TOOLSMANUFACTURING PROCESS-II UNIT-5 NC MACHINE TOOLS
MANUFACTURING PROCESS-II UNIT-5 NC MACHINE TOOLS
 
Extrusion Processes and Their Limitations
Extrusion Processes and Their LimitationsExtrusion Processes and Their Limitations
Extrusion Processes and Their Limitations
 
Call Us -/9953056974- Call Girls In Vikaspuri-/- Delhi NCR
Call Us -/9953056974- Call Girls In Vikaspuri-/- Delhi NCRCall Us -/9953056974- Call Girls In Vikaspuri-/- Delhi NCR
Call Us -/9953056974- Call Girls In Vikaspuri-/- Delhi NCR
 
the ladakh protest in leh ladakh 2024 sonam wangchuk.pptx
the ladakh protest in leh ladakh 2024 sonam wangchuk.pptxthe ladakh protest in leh ladakh 2024 sonam wangchuk.pptx
the ladakh protest in leh ladakh 2024 sonam wangchuk.pptx
 
What are the advantages and disadvantages of membrane structures.pptx
What are the advantages and disadvantages of membrane structures.pptxWhat are the advantages and disadvantages of membrane structures.pptx
What are the advantages and disadvantages of membrane structures.pptx
 
Introduction to Multiple Access Protocol.pptx
Introduction to Multiple Access Protocol.pptxIntroduction to Multiple Access Protocol.pptx
Introduction to Multiple Access Protocol.pptx
 
chaitra-1.pptx fake news detection using machine learning
chaitra-1.pptx  fake news detection using machine learningchaitra-1.pptx  fake news detection using machine learning
chaitra-1.pptx fake news detection using machine learning
 
Software Development Life Cycle By Team Orange (Dept. of Pharmacy)
Software Development Life Cycle By  Team Orange (Dept. of Pharmacy)Software Development Life Cycle By  Team Orange (Dept. of Pharmacy)
Software Development Life Cycle By Team Orange (Dept. of Pharmacy)
 
(ANVI) Koregaon Park Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...
(ANVI) Koregaon Park Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...(ANVI) Koregaon Park Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...
(ANVI) Koregaon Park Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...
 
High Profile Call Girls Nagpur Meera Call 7001035870 Meet With Nagpur Escorts
High Profile Call Girls Nagpur Meera Call 7001035870 Meet With Nagpur EscortsHigh Profile Call Girls Nagpur Meera Call 7001035870 Meet With Nagpur Escorts
High Profile Call Girls Nagpur Meera Call 7001035870 Meet With Nagpur Escorts
 
VIP Call Girls Service Hitech City Hyderabad Call +91-8250192130
VIP Call Girls Service Hitech City Hyderabad Call +91-8250192130VIP Call Girls Service Hitech City Hyderabad Call +91-8250192130
VIP Call Girls Service Hitech City Hyderabad Call +91-8250192130
 

Syst 542 paper presentation erickson final

  • 1. SYST 542 Paper Presentation: John Erickson Fall 2016 From Magnum Opus to Mea Culpa: A Cautionary Tale of Lessons Learned from a Failed Decision Support System
  • 2. Outline • Introduction – Critical Success Factors: class reading, David Arnott – Literature on DSS • Synopsis of Case Study / Paper – DSS Justification – DSS Goals – DSS Development – DSS Design – Model Results – Upgrade Results • What Went Wrong – The Catalyst of the Cataclysm – The “Postmortem” – The Mistakes/ Lessons Learned • Conclusion
  • 3. Introduction Why study failures and lessons learned? 1. High Failure Rate: “It is well known in the DSS literature that all types of DSS projects are high-risk and prone to failure…Some studies have even reported failure rates as high as 80%.” (Arnott class paper) 2. Lack of Literature: Gorman claims there are very few articles on DSS failure: “If one were to review the OR/MS literature on DSS implementations, that reviewer would be hard pressed to find published examples of failure.” Handbook of decision support systems, vol. 1: Basic Themes: Chapter 34, David Arnott
  • 4. What makes a DSS successful? *Handbook of decision support systems, vol. 1: Basic Themes: Chapter 34, David Arnott Critical Success Factor Model (CSF) Method10 Critical Success Factors 1. Committed and informed executive sponsor 2. Widespread management support 3. Appropriate team skills 4. Appropriate technology 5. Adequate resources 6. Effective data management 7. Clear link with business 8. Well-defined information and systems requirements 9. Evolutionary development 10. Management of project scope “The implication of the table is that if a reasonable number of CSF’s are not attained or achieved, a project is likely to fail.”*
  • 5. Synopsis of Paper • Michael Gorman, PhD is a management- sciences university professor and consultant in the freight / railway industry • He was hired to build a DSS for “ACME”, a growing freight transportation company that had volume growth in the early 2000s (ACME is an alias for confidentiality). • It was initially successful, then the DSS failed due to the Wall Street crisis in 2009 • His paper outlines the conditions for the need of the DSS, it’s early success, downfall, and then lessons learned.
  • 6. ACME’s System Justification • Why did ACME Build the DSS? – Previous methods for load acceptance and dispatching failed to achieve desired economic results – Frontline decision makers lacked the data, analytical ability, and global perspective to manage a network that was growing in complexity. – Thus, ACME had a strong incentive to change its approach. “Because of its growing volume and asset base, in 2005, a senior executive at the company realized that ACME had the scale to justify a DSS that would support decision makers in the face of changing fundamental economics resulting from increased asset intensity. ..”
  • 7. DSS Goals • DSS Goals: 1. Support real-time decision making at a company with annual revenues of $3 billion by cutting across both marketing and operations 2. Replace suboptimal rules for decision making (“outdated and myopic” rules)
  • 8. System Development • Senior management: – provided input on system philosophy , design, and marketing • Mid Level Management: – operations managers and frontline staff documented the business decision making process side • IT professionals: – established system integration and other technical requirements. In short, the DSS had to be integrated in such a way as to minimize legacy system development costs and disruption to ongoing operations.
  • 9. System Design Orders (These historical orders were used for demand forecasting) Economic Data (Historical cost, price, and margin data were used for predicting profitability) Service-Time Data (service times of transportation providers affect future supply Input Output Supply-and-demand forecast (supply forecast predicted equipment capacity ; demand forecast predicted daily customer load tenders) Load-accept optimization Dispatch optimization (load to- equipment assignment) -The DSS produced then a ranking of best ways to load/dispatch, based on the lowest overall cost, including both current conditions at the origin and future conditions at the destination and the network effects of other assignments to be made. -It ran without interruption each day for more than two years for forecasting and load acceptance, and every 15 minutes for load dispatching.
  • 10. Model Results & Upgrade Early success • Model recommendations were generally well received by frontline decision makers. • The model recommendations were presented in a similar way as before, as ranked recommendations, but with more sophisticated model-based rankings. Motivated by early success, the model was improved: • Support modules • More detailed forecasting and dispatching • These improvements came at a cost—increased model complexity -The shadow price of violating model-based recommendations was given, showing the opportunity cost of not following recommendations. -Global optimization results were presented to dispatchers making sequential decisions -Any change in supply, demand, or non-model recommended assignments would be reflected in subsequent model runs - Current best was always available to dispatchers, given prior events. What Model Presented in Real-Time:
  • 11. Upgrade Results (post-upgrade) • Preliminary results indicated higher revenues, higher service levels to the best-paying customers, lower costs, and better asset utilization (i.e., lower excess inventory). • First year results: formal internal audit after the model’s first year in production, the results indicated a number of performance improvements: – higher asset velocity (more loads per time period), – a decrease in low-margin loads – an increase in average margins – reduction in cost per load. In the model’s first year, the audit estimated savings of more than 20 times the system development cost.
  • 12. Then Things Went South… • The Perfect Storm: – Financial credit crisis on Wall Street in 2009-2010 – ACME’s cutting discretionary spending on consulting in 2010 • DSS was shut down and revert back to the simpler but safer rules of load acceptance and dispatch. • Management became skeptical of DSS • Gorman felt the need to catalog the lessons learned and why his “magnum opus” ultimately failed Bottom Line: The beginning of the end was this: the Wall Street credit crisis caused a perturbation that the DSS couldn’t handle
  • 13. The Central Failure in the DSS • The biggest mistake was an implicit assumption surrounding the structural stability of the demand forecast. • Although recent low-demand history helped to pull forecasts downward, previous years’ demand levels served to inflate forecasts. The underlying probability distribution used for the optimization model turned out to not be an accurate distribution when the credit crisis hit
  • 14. “Postmortem” 1. Implicit Assumption: Demand Forecast Structural Stability 2. Implicit Assumption: Cost Structures 3. Too many decision makers 4. Too many parameters 5. Too much data 6. Too Little Knowledge: Insufficient Skills Transfer 7. Too Little Time Allocated: Failed Vigilance in Ongoing DSS Operation 1. More careful and proactive tracking: analysis of recent forecast errors would have reduced prolonged use of a consistently high forecast after the downturn. 2. Be leery of using historical data in the model 3. Limit scope creep: Trying to be “all things to all people” 4. Limit model parameters: by striving for perfection, model usefulness and robustness can often be diminished 5. Robust support staff: Have sufficient support staff to the data infrastructure model. 6. Ongoing training, problem solving (live case studies),and troubleshooting between developer and client 7. Project Manager: a DSS owner, who is responsible for DSS performance, responding to customer problems, and uncovering previously undiscovered anomalies, is necessary. Mistakes Lessons Learned
  • 15. Current Situation – No DSS “The fear and mistrust of optimization techniques subsequent to this project, coupled with lack of tolerance for an association with the model failure, leave little appetite for advanced analytics to support decision making at ACME.” -Gorman
  • 16. Author’s Final Conclusions • Beware of implicit assumptions: Models inherently assume historical data patterns as a guide to future decisions. • Minimize complexity wherever possible. • Socialize, train, and experiment with DSS models. • Maintain constant vigilance of model performance. “Modeling is never done. When a client asks when the project will be done, the best reply is “never”; follow this with an explanation that tells the client that the first version will begin delivering recommendations by a specific date, at which time the team will want to continue to assess and improve on the recommendations through more accurate data and modeling.”
  • 17. Takeaways from Paper Takeway 1: Think about your DSS: A lot of thought should be put into the scope of the DSS project based on the Critical Success Factors. A bigger project may not be worth it unless the willpower, resources, and executives are all on-board. Takeaway 2: Even if the DSS meets all/most of the CSF, it takes constant vigilance to keep/upgrade/test/validate the system. This is a sobering thought for tech/internet startups!! Critique: I would have preferred that the author (Gorman) used these 10 CSF (or a similar metric) throughout his paper so that the reader can get a measurable sense of what went wrong according to the CSF. Also, use the CSF’s and compare over time (6-month increments) where the DSS stands against the CSF.

Editor's Notes

  1. Greeting, Name, Title
  2. Outline of My Presentation: introduction from arnott, breakdown of dss, what went wrong , and finally conclusion.
  3. Why would choose a paper on what went wrong? Both Arnott and Gorman have something to say here.
  4. This is from Arnott’s case study (required reading): The first 2 are about management support; without support from the top, the DSS can’t survive very long. The next group is skills/resources to enable/support the DSS; next is business requirements; last 2 are project management and scope. Of note—management involvement is something that seems to crop up again and again. The picture on the right is Arnott’ way Of showing that if a DSS has most/all the 10 CSF’s, and if the user is satisifed, and if the system makes a positive impact, then the DSS is successful.
  5. ACME is a growing freight transportation company in an evolving industry that had volume growth in the early 2000s As a result of this growth and expanded investment in centrally controlled assets, previous methods for load acceptance and dispatching failed to achieve the desired economic results. Myopic rules of thumb for load acceptance and dispatching failed to achieve economically advantageous results for both asset management and customer service. The new focus on asset management, positioning, and yield rendered previous decision-making methods ineffective.
  6. -better reflect strategic managerial perspectives in frontline decision making -Initial projections of improved performance were supported by positive proof-of-concept results -Because of its growing volume and asset base, in 2005, a senior executive at the company realized that ACME had the scale to justify a DSS that would support decision makers in the face of changing fundamental economics resulting from increased asset intensity The potential gains more than justified the projected system costs (the total internal and external project cost was in the hundreds of thousands of dollars), indicating a large potential return on investment for a completely developed DSS. -A formal organizational audit of the production system one year after its implementation found the anticipated savings to be largely realized, and the DSS was considered a success!! --Shortly thereafter, however, its recommendations began to be questioned, confidence in the model’s recommendations waned, and the system was quickly abandoned. --After the system’s demise, multiple meetings with the client were held to diagnose the causes for the failure and potentially reinstitute the DSS. The system was not recovered, but the information gleaned during this time was valuable nonetheless.
  7. Real-time decision making to increase revenues; replace old/out-dated rules
  8. A short-term demand forecast predicted daily customer load tenders by geographic origin, destination, date, and other attributes (e.g., specific equipment needed) over a two-week period. A supply forecast predicted equipment capacity based on expected termination of previously accepted business To maximize the potential revenue of short-term capacity, a load-accept optimization module determined which tendered orders to accept and which to reject based on anticipated capacity shortfalls, if any Dispatch optimization: The output of the dispatching system was a load to- equipment assignment. It was presented to dispatchers as a list of assignments ranked from best to worst—as they had seen prior to the model’s implementation.
  9. -There were many weaknesses in the system that were not exposed until the financial credit crisis on Wall Street in 2009-2010 -The downturn in 2009 led to revised 2010 corporate forecasts, budgets, and strategies across the U.S. economy. Many shipper contracts were canceled and investments deferred. This affected a corresponding freight demand for ACME and the fundamentals on which the DSS rested. -The dramatic shift in business exposed the instabilities of a well-performing and stable DSS that had been in production for more than two years. The vulnerability of the DSS was compounded by the size, speed, and disruptiveness of the shift in the market, leaving very little time to diagnose new and impactful problems in the model. -These problems were compounded by ACME’s rapidly falling discretionary funds, which caused ACME to cut discretionary spending on consulting in 2010. -I worked with ACME during the last days of the model’s operation to diagnose and rectify problems. With better timing, effort, and luck, the DSS may have been able to overcome the business environment shift in 2009–2010. In an effort to simplify operations and reduce confusion, management decided to shut down the DSS and revert to the simpler but safer rules of load acceptance and dispatch.
  10. Mistakes are primarily related to 1) assumptions; 2) too many or too little of various things… 3)then the time to maintain the DSS on a very vigilant scale. Regarding assumptions, this is really related to the historical data used for the underlying probability distribution. The decision makers aspect was that it tried to be all-satisfying in it’s scope—both frontline users and senior management. This is evident from numbers 3-5. Number 6 and 7 are speaking to the training and maintenance, and constant upgrade and constant vigilance of the DSS. Fine-tuning it continously.
  11. Having been burned by a rapidly changing market and a DSS that did not adjust to such changes, ACME has returned to a much simpler direct-cost, margin-based system. Although the approach is less likely to make a clearly poor recommendation, it regularly omits key considerations that the DSS was designed to include. By throwing away DSS-based opportunities for improvement on every transaction, ACME avoids the risk of overly ambitious recommendations,which can occur when a model ceases to reflect reality.
  12. In the end, Gorman’s summaryizing final conclusions are: beware of implicit assumptions, minimize complexity, socialize train and experiment; and maintain constant vigilance. His quote I think accurately summarizes his point, where he says “ modeling is never done…”