Gaining Competitive Advantage through the Application of Advanced Decision Support and Information Systems Gang Yu, Ph.D. Chairman and Co-founder New Height Corporation 1 Outline Major driving factors Tips for tackling the frequently encountered challenges Case studies Amazon inventory management Dell strategic buys CALEB disruption management and manpower management systems 2 Major Driving Factors Customer Needs Enabling Technology Methodology Advancement Customer Needs More fierce competition Dynamic market and environment More complex business with larger scope and scale Globalization Shortened product life cycle, blossomed product variety, increased customization 4 Enabling Technology Moore’s law Data acquisition and transmission tools Internet, WIFI, UMPC GPS, GIS, RFID, bar coding, 3G/4G mobile devices Advanced software tools for better GUI development 5 Methodology Advancement More efficient algorithms for solving large scale core optimization problems Better modeling techniques Better business understanding results in more fitting models 6 Frequently Encountered Challenges How to define and propose the right DSS/IS project How to get management buy-in How to define, measure, and demonstrate value How to manage the development process and client’s expectations How to get users’ acceptance 7 How to Propose the Right Project Critical to the core business Impacting major metrics Scope can be clearly defined and controlled Proposed by proper people Fitting the current company’s financial situation Right timing, right people, right project. 8 Examples ManpowerSolver project at Continental Airlines undertaken by CALEB Optimal In-stock target at Amazon.com Strategic buy project at Dell Inc. 9 How to Get Management Buy-in Top management sponsorship is critical Present in business terms, avoid technicality Users’ support is important Risk assessment and mitigation 10 Examples CALEB’s CrewSolver ManpowerSolver at Continental Airlines Deployment at Southwest Airlines, Continental Airlines, Northwest Airlines, Delta, JetBlue, etc. all received top management sponsorship Secured support from SVPs through multiple presentations Amazon In-stock target Need to get alignment and support from leaders of operations and all the retail teams managing 23 product lines 11 How to Define, Measure, and Demonstrate Value Show value via prototype systems Simulate decision scenarios Compare with past and current decisions Define proper metrics Set benchmark 12 Examples Metrics used for airlines for measuring the impact of disruption management systems % delays over 15 minutes Average delay time % cancellations Average recovery time Penalty cost for recovering from disruptions Metrics for measuring inventory health Service level % deviation from optimal in-stock % of unhealthy inventory Inventory turn over ratio Average speed of disposition 13 How to Manage the Development Process and Client’s Expectations Simplify, simplify, and simplify Let the users understand decision logic, never build a black box Flexible and modular design to accommodate future changes R&D is required in most cases Work together with client as one team Manage scope tightly and rigorously 14 Examples CALEB Separate OpSolver, CrewSolver, and PassengerSolver Simplified modeling and algorithms Layered design for reusability Amazon Clustering products by demand rate, lead time, and inventory cost for calculating optimal in-stock 15 How to Get Users’ Acceptance Users must be involved in the entire development process GUI design should consider users’ habits Solution presentation should be simple and intuitive Allow users to customize Allow users to interact with the system 16 Examples CALEB Engage crew coordinators and operations managers in the OpSolver/CrewSolver development Frequent team meetings and periodic project reviews Amazon Multiple meetings with various retail organizations to go over our assumptions and secure their input Reach agreement before publishing metrics in the weekly business review 17 Case Study Inventory Management System built by Amazon’s Supply Chain management organization decides the optimal in-stock targets, optimal inventory level and inventory health. Team: Nader Kabbani, Russell Allgor, and many others 18 Amazon Inventory Advantage Physical Store Amazon.com Distribution Centers Stores Inventory 19 Amazon Inventory Classification Unhealthy inventory Healthy overstock generated by dealbuy Optimal inventory target Replenishment point 20 Optimizing Inventory Determine the optimal inventory targets based on inventory costs, lost sale costs, and lead time If inventory < replenish point => automatically trigger replenishment If inventory > unhealthy level => use various disposition channels (vendor returns, liquidation, discount, etc.) to drain inventory For every deal-buy opportunity, decide optimal quantity to buy to ensure inventory health 21 Results Reduced inventory Improved margin + Improved In-stock Improved Customer Experience 22 Case Study CALEB Technologies Corp. CrewSolver – a real-time crew disruption management (recovery) system. Team: Mark Song, Mike Arguello, Ben Thengvall, Sandy McCowan, and others Winner of the Franz Edelman first place prize in 2002. 23 CrewSolver Applications at Continental Airlines Complex, Dynamic Environment Sophisticated Plans Scarce and Coupled Resources FAA Restrictions Contractual Obligations Large Operational Scale 24 The Crew Recovery Project Definition When faced with disruptions including: ¾ ¾ ¾ ¾ ¾ ¾ ¾ Flight delays and cancellations Diversions Adding new flights Reassigning equipment types Sick crew Crew misconnects Crew legality issues How do airlines get crew back to planned pairings in a timely manner? ¾ ¾ Minimize recovery costs while covering all flights Ensure crew qualifications and legalities All scenarios can be treated as resource shortage 25 Continue to observe Operations GUI (Situation monitor) Existing problem Below min threshold Framework for Decision Process or what-if scenario Check threshold s Complete re-planning Above max threshold Frame problem (Isolate impacted areas) Statistical analysis for robust planning Update system logbook Yes No Apply appropriate algorithms from repository Try another iteration ? Generate partial solution No Solution feasible ? Change some parameters or modify the problem Yes List solutions Select appropriat e solution Implement solution No No Look for alternative solutions? Yes Yes 26 System Architecture Crew Client GUI System Environment Variables Configuration Parameters Socket CrewSolver Optimization Server Static City File Crew File Shared Memory Middleware FTP Message Server The Mainframe SOC Database Publish/ Subscribe 27 Prototype for Assessing System Value Value Analysis for Daily Operations ¾ 10 ~ 30 disruptions/day ¾ Takes seconds to generate recovery solutions by applying CrewSolver systems ¾ Saves $4,000 to $39,000 per case ¾ Savings over $15 million/year for daily disruptions System Development and Deployment 29 Testimonial: Impact for Major Disruptions Recovery for the New Year’s Eve 2000 Nor’easter Nor’easter hit New York metropolitan area, worst since 1996 12/29/2000 Newark had 35% reduction resulted in 112 cancellations 12/30/2000 Complete shut down of Newark hub resulted in 113 cancellations 354 Crew members affected Other Airlines Up to 3 days to recover ¾ Unhappy customers ¾ Unhappy crews ¾ Continental with CrewSolver 9 Less than 5 minutes to solve 9 Recovered within the same day 9 No crew complaints 9 Fewer reserves used 9 Over $5 million savings Continental was back to normal operations in record time 30 Testimonial: Impact for Major Disruptions 9/11 Terrorist Attack 9/11/2001 Terrorist Attack to World Trade Center and Pentagon US nationwide airport shutdown for three days Most airports reopened after three days with a reduced flow Largest disruption in air transportation history Other Airlines ¾ ¾ ¾ Took several more days before recovery Used at least 250 man-hours to generate recovery plan Tremendous losses Continental with CrewSolver 9 9 9 9 20 man-hours to generate recovery plan The only carrier to operate at least 55% of its flights by the weekend $30 million savings Largest ever savings from a single IT application for Continental 31 9/ 9/ 9/ 9/ 9/ 9/ 9/ 9/ 9/ 9/ 20 19 18 17 16 15 14 13 12 11 /2 /2 /2 /2 /2 /2 /2 /2 /2 /2 00 00 00 00 00 00 00 00 00 00 1 1 1 1 1 1 1 1 1 1 Major Airlines: % Delays 50 40 CO 30 20 10 AMR DL NWA UAL USAIR WN 0 32 More Testimonials 2002 Franz Edelman Management Science Achievement Award Continental Airlines CEO Gordon Bethune Speech Continental Airlines President Larry Kellner Speech Information Week (9/24/01): “Continental Uses IT To Chart A New Course” Information Week (11/19/01): “How IT Helped An Airline Recover” 33 Conclusion DSS/IS can bring companies remarkable competitive advantage Deployment of DSS/IS requires top management buy-in, users’ support, and proper simplification and adoption of OR modeling and IT architecture The success critically depends on the marriage of OR + IT + Business management 34 Thank You! 35