The Evolution of e-Commerce in the Airline Industry

1
The Evolution of e-Commerce in the Airline
Industry AGIFORS Reservations and Yield
Management Study GroupNew York - March 24, 2000
by Richard Ratliff
2
Overview

• Introduction
• The History of Distribution
• The Internet a New Distribution Channel
• Impact of the Internet on Distribution and
  Planning Systems
• Future Outlook

3
Introduction

• The travel and transportation industry has a long
  history of electronic commerce and communications
• Developed internal communications infrastructures
  to coordinate the activities of staff, aircraft
  and passengers
• In the 1950s, business-to-business systems (ARINC
  and SITA) created to facilitate passenger service
  across airlines
• In the 1960s and 1970s, systems such as Galileo
  and Sabre developed to consolidate airline
  product information (schedules, fares and
  availability) for travel agencies, creating a
  global electronic marketplace for the airline
  industry
• Airlines have taken advantage of the information
  and control available in this environment to
  increase revenues and reduce costs (including
  development of OR applications)

4
Introduction (contd)

• Airline industry has a technical and cultural
  predisposition to e-commerce
• Explosive growth of internet and World Wide Web
  has changed the volume and nature of electronic
  transactions
• Legacy systems have required retooling, new
  business models have been created
• These factors have expanded the actual and
  potential use of Operations Research within the
  travel and transportation industry
• Review the evolution of e-commerce in the travel
  and transportation industry
• Challenges associated with the current
  environment
• Adapting existing models and new OR opportunities

5
The History of Distribution
6
Relevance

• Growth of CRSs and the related use of Operations
  Research in the airline industry provide a strong
  foundation to build upon in the newly evolving
  and expanding world of Internet-based e-commerce
• However, the infrastructure that exists today was
  built up over a 70 year period

7
Early e-Commerce in Air Travel

• The pioneering efforts for airline reservations
  began with the request and reply system used in
  the 1930s
• Through the mid-1940s reservations were recorded
  manually with a pencil on different colored index
  cards, nicknamed Tiffany cards after the lamps
  with the colored glass shades
• Overbooking used to account for misplaced or
  incorrectly filed reservations (no recs)
• After World War II airlines began investing in
  technology

8
How CRSs Originated

• In the late 1950s, air travel was on the brink of
  two key transformations (jet aircraft and IT)
• SITA and ARINC were one of the worlds first
  business-to-business (B-to-B) systems in the
  1950s
• In 1959, AA and IBM jointly announced plans to
  develop a Semi-Automated Business Research
  Environment better know as the Sabre
• CRSs were the first business application of
  real-time computer technology
• Moved from hand-written to electronic passenger
  information records via automated systems
  accessible to any agent

9
YM and the Increasing Importance of Airline OR

• New,start-up carriers in the 1970s (e.g.
  Peoples Express and Texas International)
• Introduction of supersaver fares
• YM fare control began as a defensive measure by
  majors
• Major carriers could utilize the wealth of data
  available from their reservations systems
• Following deregulation, major US carriers were
  uncompetitive on cost
• Saddled with legacy pilot and flight attendant
  union contractual agreements
• Without revenue-enhancing CRS and IT/OR
  technology, majors would have been unable to
  respond to competition

10
Connecting to Travel Agencies Distribution

• As passenger volumes increased, travel agents
  became increasingly concerned about their
  business
• Processes remained paper-intensive and
  time-consuming, offering slower service than the
  airlines could
• Automation was needed to print itineraries,
  invoices, tickets and accounting functions
• JICRS (Joint CRS initiative)
• 1974 - Create one CRS for all airlines
  (participants included American, Eastern, Trans
  World, United, Western)
• 1975 - Failure to reach agreement United
  withdrew
• 1976 - Apollo and Sabre installed in travel
  agencies
• 1978 - The US airline industry is deregulated
• Actions spawned todays multi-CRS and GDS
  environment

11
CRSs are Regulated

• Nov. 1984 - several key CRS functions were
  regulated by the U.S. Civil Aeronautics Board
  (now known as the US DOT)
• Display bias was their primary concern
• Timing of fare releases and ATPCO
• Competitive advance booking data (e.g. MIDT) made
  available
• No differentiation allowed in booking fees by
  agent

12
New Capabilities in the 1980s and 1990s

• Additional functions become available
• CRS hosting
• Frequent Flyer programs
• Hotel, car rental and cruise line availability
• Bargain finder (search multiple fares and advise
  which class is least expensive for flights
  booked)
• Automated yield management
• Direct connect availability
• E-ticketing
• Internet travel sites
• Best fare finders (go directly from low fare to
  flight)

13
GDS - Relationship Changes
1976-1993
1994 - 1999

• 1. Basic Distribution 1976 - 1985 (10
  years)
• 2. Advanced Distribution 1986 - 1999 (14
  years)

• Supplier
• GDS
• Agency
• Traveler

• Supplier
• GDS
• Agency
• Traveler

In 1994, Easy Sabre on Prodigy and AOL In 1997,
the Internet arrived.
14
The Internet a New Distribution Channel
15
Introduction

• Today CRSs and GDSs are the main ticket outlet
  for most airlines
• The internet allows airlines and ticket brokers
  to bypass the travel agent
• Customer needs drive the e-design
• Legacy systems limit the e-design
• Different outlets specialize on different
  customer groups
• Reverse Auctions
• Virtual Travel Agents
• Airlines Sites
• Global Distribution Systems

16
Reverse Auctions

• Customer View
• Web sites like PriceLine.com allow the customer
  to name a price for a travel product
• Customer has to accept any product that matches
  the price
• Infrastructure
• The broker contacts airlines directly and shops
  for the best available fare
• OR Models
• Reverse auction models are useful to determine
  inventory controls in this business model
• Help give information on underlying consumer
  demand

17
The Virtual Travel Agent

• Customer View
• Sites like Sabres Travelocity.com and Preview
  Travel or Microsofts Expedia allow customers to
  pick and choose among different offers online
• Infrastructure
• The sites work on top of existing CRSs and
  emulate the work of travel agents
• Data Needs vs. Data Sources
• Fares include published, off-tariff and
  dynamically created
• OR methods can be used to build an efficient link
  between the GDS and customer sites

18
Finding the Best Fares using OR Techniques

• OR Problem
• Optimize among a broad number of flight and fare
  alternatives and also rank secondary choices
• Problem Characteristics
• Problem space is very large and computational
  time limited
• Side constraints are on the leg and on the path
  level
• Special Considerations
• Algorithm performance depends on efficient fare
  enumeration and rule checking
• Different types of data have different access
  times
• Useful By-Products
• Intermediate search results provide the customer
  with additional information

19
Airline Sites

• Many airlines sell tickets directly through their
  own web sites
• Customer Pros and Cons
• Customers are rewarded by special discounts and
  offers
• But they don't have the opportunity to shop for
  other airlines
• Use of OR Methods
• Airlines use statistical methods to set up
  promotional schemes that target special consumer
  groups
• Provide availability processing and best fare
  search capabilities such as those available in
  the GDSs

20
Global Distribution Systems

• Internet
• GDSs use the internet to extend their reach
• What's new?
• Travel agents and GDSs provide value added
  services to compete with new distribution
  channels (e.g. Virtually There)
• Bundling of services and cross-selling
• OR Applications
• Statistical models are used to find cross-selling
  opportunities
• New YM opportunities for more detailed
  availability control based on customer-specific
  behavior (creates both real-time and profiling
  challenges)

21
Impact of the Internet on Distribution and
Planning Systems
22
Introduction

• Airlines use market analysis and OR based systems
  to maximize expected revenue
• Much of the data that feeds the OR systems are
  collected by CRSs and GDSs
• The advent of a new distribution channel has a
  major impact on the validity and availability of
  the data
• In some cases the OR models themselves have to be
  re-engineered to fit the new business problem
• Example OR applications follow in the next few
  slides

23
CRS Simulation

• CRSs use a set of rules to determine which
  flights are presented upon a given request
• Screen presence has an extraordinary impact on
  customer preferences
• Simulation models can be used to determine the
  effects of different strategies on screen
  presence and market share
• Recent innovations such as web outlets and
  dynamic display rules also need to be considered
• Useful for developing e-mail promotions or those
  via an airlines web site

24
Passenger Preference Modeling

• Passenger preference models became prevalent
  after industry de-regulation
• Schedule design became a very complex problem due
  to a growing number of airports and increasing
  demand
• Models developed to support schedule design by
  evaluating schedule profitability
• These model take account of market size
  forecasts, passenger preference parameters,
  flight schedules, fares and business rules

25
Passenger Preference Modeling (contd)

• The internet results in a large number of
  distribution channels with low volume
• Preference models have to capture passenger
  behavior with respect to all types of
  distribution channels
• Smaller booking volumes per outlet increase data
  variability used to calibrate the customer
  preference model
• Many internet travel sites store customer
  profiles
• May also be used to calibrate passenger
  preference models
• Potential use of clickstream data
• Captures transactions made by customers on web
  sites
• Similar attempts were made in Sabre by recording
  agent key strokes during randomly selected sales
  sessions

26
Passenger Yield Management

• Demand and passenger behavior data is necessary
  to set controls, and CRSs serve as data sources
• Advancements in the OR and processing are moving
  us from separate time-series forecasting and
  leg-based optimization to econometric models and
  ODYM
• Still mostly batch processes today
• Real-time re-forecasting and re-optimization in
  next five years
• Incorporation of still more detailed controls
  with e-channels (customer-specific availability
  via on-line access to historical information and
  rapid profiling of characteristics)
• Competitive closures will be less obvious due to
  reduced use of traditional distribution channels

27
Passenger YM (other impacts)

• Internet sales change size and characteristics of
  demand
• Changes in passenger behavior due to internet
  specific restrictions
• May necessitate re-calibration of overbooking and
  demand forecasts
• Hidden shifts in competitive bookings market
  share due to direct airline web sites
• Internet forces a change in pricing strategy
  (from oligopoly to retail)

28
Cargo YM (B-to-B types)

• Medium-term yield management
• Various forwarders (bulk customers) submit bids
  for shipping capacity on airlines flight network
• The airline optimizes the allocation of available
  capacity to various bids by maximizing the
  expected revenue over a planning period such as
  quarter
• Cargo routing is useful in determining feasible
  and profitable routes for satisfying a shipment
  request
• Being extended to the Internet to efficiently
  integrate the business processes involved with
  the shipper-forwarder interaction
• Can provide dynamic pricing and capacity
  allocation

29
Cargo YM (B-to-C types)

• Short-term yield management to satisfy the ad-hoc
  shipment demand
• Bid prices
• Determined by considering the ad-hoc demand,
  medium-term demand, and available capacity
• Used to accept/reject shipment requests over the
  booking horizon
• Improved consumer cargo search engines via the
  Internet may stimulate additional demand for
  last-minute shipments and drive large changes
  from historical booking behavior

30
Future Outlook
31
Regulation of e-Travel Sites?

• Will Internet travel websites be regulated?
• Neutral, semi-neutral and aligned sites exist
• Up-front disclosure of alignment is important
  in semi-neutral sites (e.g. T2 consortium or
  sites with airline equity investment)
• Customers could be misled into thinking that a
  complete and unbiased range of alternatives will
  be presented
• But even neutral infomediaries may be biased
• Any system will require an algorithm that
  determines what to display and the ordering
  (airlines, mortgages, insurance)
• Volume-based commissions create incentives for
  bias
• Suppliers are paying for essentially two things
  1) to be listed on the website and 2) better
  presence

32
Regulation of e-Travel Sites? (contd)

• Bias in e-commerce travel sites is similar to
  what exists through brick and mortar
  establishments
• Booking direct with airlines is biased
• Everything equal, agents favor airlines with best
  commissions
• But governments have avoided e-Commerce
  regulation
• Secondary market-driven forces may come to the
  rescue
• Studies of best fare comparisons by consumer
  advocacy groups (e.g. Consumer Reports)
• Authentic neutrality may even become a strong
  selling point among the informediaries

33
Search Robots

• Currently, e-commerce on the web is free to the
  user
• Search robots can abuse other sites to shop for
  free information and re-sell it to the customer
• Impacts both the virtual travel agent and airline
  sites
• Increasing sophistication makes robots harder to
  detect
• How can the industry protect itself against this
  abuse?
• Design websites to make it difficult for
  meta-search engines
• Drilling down for information several screens
  deep
• More frequent use of member i.d. logins to
  distinguish genuine users from robots
• Usage-based fees?

34
Airline B-to-B Will Grow

• Successful alliance implementation requires
  seamless integration of various business
  processes and systems
• Internet and related technologies provide the
  communications infrastructure required for the
  business to business integration
• Alliances have a profound impact on the airline
  OR systems
• Need to expanded current models to reflect the
  collaborative planning, marketing, and operating
  efforts among the constituent airlines of the
  alliances
• B-to-B vendors will provide central repositories
  for the data required for alliance related OR
  systems
• Will provide better tools to allow carriers to
  implementing the policies obtained from the
  alliance-based OR models
• e.g. Sabre / Ariba deal to create Sabre
  e-Marketplace

35
Impact on the Airline OR Profession

• Effective implementation of new e-Commerce
  business practices requires investigation using
  OR
• Rapid proliferation of e-Commerce practices is
  putting a strain on the airline OR profession
• The OR model life cycles are decreasing
• The rewards associated with rapid OR modeling are
  becoming high but create greater risk of negative
  impact
• The data are more noisy and the business
  environment is more unstructured than ever before
• Ethical and legal ramifications such as what
  level of detail data can be used from
  click-stream data?
• Confidentiality and privacy issues

36
Thanks

• Other colleagues at Sabre who assisted in
  material presented here
• Dan Delph
• Dirk Guenther
• Beju Rao
• Barry Smith
• Pat Trapp

37
Selected References

• Gilbert Burck, On Line in Real Time,
  FORTUNE magazine, April 1964.
• Copeland, Mason, and McKenney, Sabre The
  Development of Information-Based Competence and
  Execution of Information-Based Competition IEEE
  Annals of the History of Computing, Vol. 17, No.
  3, 1995, pg. 30
• Lee Davis - UNISYS, Real Time- The Ultimate
  OD, AGIFORS RYM, Melbourne, May 1998
• Geraghty, Govil, Guarnieri, Lancaster - Delta
  Technology, Securities Trading Paradigm for
  Revenue Management, AGIFORS RYM, Melbourne, May
  1998
• Guenther, Rao, Ratliff, and Smith - Sabre, A
  Review of the Evolution of e-Commerce and
  Operations Research in Travel and
  Transportation, working paper, March 2000
• Max D. Hopper, Rattling SABRE New Ways to
  Compete on Information, HARVARD BUSINESS REVIEW,
  No. 90307 May-June 1990.
• Startup Muse, FORBES magazine website
  (www.forbes.com), Digital Tool feature, August
  18, 1999 issue
• Thats the Ticket, WALL STREET JOURNAL, Monday,
  July 12, 1999, e-Commerce Section, pg. R45

38
Questions?