Maximizing Revenue with a Leg Revenue Management System

1
Maximizing Revenuewith aLeg Revenue Management
System

• E. Andrew Boyd
• Vice President, Research and Design
• PROS Revenue Management

2
Revenue Management Goal
Generate increased revenues by seeking to accept
the highest paying mix of passengers

• To achieve the goal of increased revenues, the
  value of each ticket sold must be known with some
  certainty

3
A Perfect World

• One flight leg
• One fare per booking class
• Y 800
• M 600
• B 500
• Q 350

4
Perfect Fares
Fare
1000
800
800
600
600
500
400
350
200
Class
Y
M
B
Q
5
Fares in Reality
Fare
1000
800
600
FareRange
400
200
Class
Y
M
B
Q
6
Where Does Fare Variation Come From?

• Source 1 Distribution Channel

7
Where Does Fare Variation Come From?

• Source 1 Distribution Channel
• Y class tickets on flight 111 may sell for
  different fares because of distribution channel

Alliances
TravelAgencies
Code ShareArrangements
Inventory
Internet Sales
TourOperators
CorporateAccounts
InternetAuction
FrequentFlier Programs
8
Distribution ChannelsFront End or Back End Fare
Control?

• Front End Fare Control
• Make sure fares align across distribution
  channels when they are established
• Back End Fare Control
• Assume fares will not be aligned, and account for
  differences when a purchase request arrives

Front End
Back End
9
Front End Fare Control

• Good front end control is extremely difficult to
  achieve, but is the only alternative for carriers
  in a non-seamless / leg environment
• How to assure that all Y class tickets on flight
  111 sell for approximately the same price?

10
Back End Fare Control

• Seamless availability / OD provides the
  opportunity to address many distribution channel
  issues from the back end
• Dynamic fare calculation at time of
  requestassigning the right fare to each
  transaction

11
The Value of Back End Fare Control
Quality of Front End Fare Alignment

• The worse fair alignment is on the front end, the
  greater the impact of back end fare control
• Back end fair control can have a major revenue
  impact

Below Upper Bound
12
Simulation

• Purpose
• To simulate the effects of poor fare class
  rationalization
• Simulation details
• Eight itinerary/fare classes
• Four fare class scenarios established by setting
  upper and lower limits on fare class fares
• Top fare for each itinerary established from
  network data
• Within each fare class, each fare value between
  the upper and lower limit is equally likely

13
Where Does Fare Variation Come From?

• Source 2 Leg Control

14
Where Does Fare Variation Come From?

• Source 2 Leg Control Fares are provided at the
  OD level, but availability is controlled at the
  leg fare class level
• Flight networks greatly complicate the process of
  defining leg fare class fares

15
Leg Control Fare VariationThe Problem

• Example Y class fares on a simple network

800
STL
ELP
IAH
400
MSY
700
16
Leg Control Fare VariationThe Problem

• Y fare tickets on the leg ELP-IAH take on the
  values 400, 700, and 800 a wide fare variation

Should these fares all be assigned to Y class?
Should these fares be reassigned to
differentclasses that reduce the fare variation?
17
Leg Control Fare VariationThe Problem

• Fare class realignment may be required

Reassign to M Class
800
STL
ELP
IAH
400
MSY
700
Reassign to B Class
Leave in Y Class
18
Fare Class Rationalization

• Very Important Fact
• Fares throughout the network must be aligned
  within fare classes if leg control revenue
  management systems are to function properly

• How can this be achieved?

19
Network Optimizationfor Fare Class
Rationalization

• Assign OD fares to fare classes so that a leg
  control revenue management system generates
  maximum revenue
• Determine leg fare class fares for leg
  optimization algorithms
• Adhere to all business constraints

20
Observations

• The myriad of business constraints leads to an
  extremely challenging problem that is not easily
  handled by an optimization algorithm alone
• If fare class rationalization is not addressed,
  even a good leg revenue management system can be
  severely compromised
• Fare class rationalization is an important aspect
  of aligning fare classes within an alliance

21
Fare Class RationalizationThe Wrong Way

• Ordering OD values from highest to lowest fare
  value and clustering to form fare classes is a
  poor idea

22
Fare Class RationalizationThe Right Way

• OD valuation should incorporate some measure of
  displacement cost
• Good leg fare class fares must be determined

23
A Process For Fare Class Rationalization Getting
to Leg Class Fares

• In the case of the ELP-IAH-STL itinerary, it is
  incorrect to attribute a revenue of 800 to the
  ELP-IAH leg since the itinerary uses two legs
• The situation is the same for theELP-IAH-MSY leg

24
A Process For Fare Class Rationalization Getting
to Leg Class Fares

• A way of addressing multi-leg fares is to prorate
  the OD fare to the legs

25
A Process For Fare Class Rationalization Getting
to Leg Class Fares

• Important Fact
• There are mathematically based ways to prorate
  itinerary fares into leg fares itinerary fares
  can then be treated as consisting of leg fares
  for purposes of fare class rationalization

800
200
600
A True Proration
800
300
700
A Pseudo-Proration
26
A Process For Fare Class Rationalization Getting
to Leg Class Fares

• With fares prorated to legs, it is possible to
  determine good leg fare class fares by taking a
  weighted average of the prorated fares within the
  fare class
• Prorated fares still have the likely problem of
  being poorly rationalized at the leg level

27
A Process For Fare Class Rationalization Getting
to Leg Class Fares

• Proration alone does not solve the problem of
  fare class rationalization

Itinerary Y Fare Prorated Fare for ELP-IAH Leg
ELP-IAH 400 400
ELP-IAH-STL 800 100
ELP-IAH-MSY 700 425

• Itineraries ELP-IAH and ELP-IAH-MSY align well
  onleg ELP-IAH-MSY, while itinerary
  ELP-IAH-STLdoes not

28
A Process For Fare Class Rationalization Getting
to Leg Class Fares

• Proration does suggest how fare class
  rationalization might be achieved

Itinerary Y Fare Prorated Fare for ELP-IAH Leg
ELP-IAH 400 400
ELP-IAH-STL 800 100
ELP-IAH-MSY 700 425

• Y Class ELP-IAH and ELP-IAH-MSY
• M Class ELP-IAH-STL

29
A Process For Fare Class Rationalization Getting
to Leg Class Fares

• The problem with rationalizing fare classes on
  one leg is that the rationalization may not make
  sense for other legs

30
A Process For Fare Class RationalizationPseudofa
res

• Pseudofares are one method of prorating itinerary
  fares to leg fares

Pseudofarek(l1,l2,,lk,,lm) Fare(l1,l2,,lk,,l
m) Displacement Cost (l1,l2,,lk,,lm)
Displacement Cost (lk)
31
A Process For Fare Class RationalizationPseudofa
re Example
800 600 100
800 600 500
800
700
300
100
500
Displacement Costs
32
A Process For Fare Class Rationalization

• Pseudofares provide a good alternative for
  performing fare class rationalization at the leg
  level

Itinerary Y Fare DisplacementCost Adjustment Pseudofare for ELP-IAH Leg
ELP-IAH 400 0 400
ELP-IAH-STL 800 700 100
ELP-IAH-MSY 700 275 425
33
A Process For Fare Class Rationalization

• Important FactPseudofares have the special
  property that they preserve order under fare
  proration to different legs

• This special property of pseudofares makes them
  ideal for the purpose of fare class
  rationalization

34
Ultimate Fare Class Rationalization

• Fare class rationalization is a front end fare
  control solution
• Seeks to align fares so that the allocation of
  inventory through leg fare class limits yields
  revenue maximizing decisions
• Can leg fare classes ever be fully rationalized?
• Is it possible to achieve a perfect world in
  which every ticket sold in a leg class has the
  same fare?

35
Ultimate Fare Class Rationalization

• The Bad News
• Restricting inventory control to availability in
  leg fare classes, the answer is noa perfect
  world cannot be achieved
• Fare class rationalization is necessary to get
  the most from a leg control system, but there are
  inherent limitations

36
Ultimate Fare Class Rationalization

• The Good News
• Stepping outside of leg fare class control to OD
  control, a perfect world can be achieved
• Leg level Displacement Adjusted Virtual Nesting,
  for example, is the obvious and inevitable
  resolution of the problem of trying to
  rationalize fare classes
• Fare class rationalization is simply thrown out
  the window and replaced by leg value buckets
• Bid price methods achieve the same goal in a more
  sophisticated fashion

37
Conclusions Distribution Channel

• Fare variation as a result of different fares in
  different distribution channels is a huge problem
  that must be addressed
• The move to seamless availability / OD control
  opens up substantial opportunities for revenue
  improvement through back end inventory control
• Without seamless availability / OD control,
  airlines must be diligent in aligning fares
  onthe front end

38
Conclusions Leg Control

• In a leg control environment, it is vital to have
  fare classes rationalized at the network level in
  order to achieve maximum revenue gains from a leg
  control revenue management system
• Good fare class rationalization is difficult and
  has inherent limitations
• OD control overcomes the limitations inherent in
  a leg control revenue management system

39
Maximizing Revenuewith aLeg Revenue Management
System

• E. Andrew Boyd
• Vice President, Research and Design
• PROS Revenue Management