Tom Gregorson
Vice President
Products & Solutions
Tom GregorsonVice President,
Products & SolutionsATPCO
2015: WE STARTED WITH A WHITEPAPER
Dynamic Offering of Fare Levels
Create Services and Brands
Personalization Dynamic Fare Adjustment Generations
DRIVING THE INDUSTRY FORWARD
INDUSTRY WORKING GROUPS FURTHER STAKEHOLDER INPUT
Research Paper with PODS, LLC
Research and Alternatives to Expand RBDs
Dynamic Adjustment Pilots
Most Recent Dynamic Price WG
73 organizations
257participants
2016 20182017
RESEARCH PAPER AGENDA
• Background on Revenue Management
• Definition Framework
• Pricing & Revenue Management in Other Industries
• Next Generation Pricing Mechanisms
• Implications of Next Generation Pricing Mechanisms
• Summary and Conclusions
PILOT APPROACH 1 – PROS-ATPCO
Pricing
(Published Fares)
Request Availability
Availability Price with “Trigger”
Price AdjustmentAssign Code
“Trigger”
Return Availability
+ Trigger
Data Collection
and Distribution
CONTENT PROS
Request Response
Price Adj +Trigger
Availability Schedules
Fares/Rules Ancillaries
ATPCO
AIRLINE
PRICING
SYSTEM
FBR with Code
“Trigger” at all
relevant price points
“Trigger” reflected on
the ticket. No change to
downline processing
PILOT APPROACH 2 - SABRE-FARELOGIX-ATPCO
Dynamic Price Adjustment
Shopping
Booking
Pricing
Reporting
Inventory
ET Server
Dynamic
Pricing
Engine
BSPRevenue
Accounting
ATPCO Message Hub and DPE
Private Fare Filling
External Data Inputs
(e.g. Lowfare Shopping Cache)
CRS/GDS AIRLINE SYSTEM(S)
shoppingID (for reference only)
pricingID
RET HOT LIFT
Ticketing
Manish NagpalVice President,
Program DevelopmentFarelogix
John McBrideHead of Global Travel Product Management
PROS
Richard RatcliffSenior Research
ScientistSabre
Peter BelobabaPrincipal Research Scientist
International Center for Air TransportationMassachusetts Institute of Technology
Peter BelobabaPrincipal Research Scientist
International Center for Air TransportationMassachusetts Institute of Technology
Advances in Airline Pricing, Revenue Management, & Distribution
Implications for the Airline Industry
Peter BelobabaBill Brunger
Michael D. Wittman
Prepared for ATPCO by PODS Research LLC
October 2017
Purpose of Research Study• ATPCO commissioned PODS Research LLC to review current and next-
generation pricing practices in the airline industry.
Review the historical development of pricing and RM in
the airline industry
Examine pricing and RM practices in other industries and
describe similarities/differences with airlines
Describe next-generation pricing practices that are
currently under development in the airline industry
Assess potential implications of next-generation pricing
on revenues, competition, customers,
and internal processes
Pricing and Revenue Management Are Small Parts of a Complex Airfare Shopping Process
PRICING
Q
Y
M
B
X
REVENUEMANAGEMENT
DISTRIBUTION
File a set of fare products
Y $400 No AP N/A
B $300 3D AP NONREF
M $200 7D AP NONREF
Q $100 14D AP NONCHG
Send farestructure
to RM
Sendbooking limits/bid prices
Poll INV for availability
Referencefare filingsto determinefeasible fareproducts
INVENTORY
Calculate fare quote
FARE QUOTE
The New Distribution Capability Could Start to Change Some of these Legacy Processes
• IATA’s New Distribution Capability (NDC) is an XML-based standard for distribution communication.
• NDC allows for more information to be exchanged in the ticket shopping process besides itineraries, availabilities, and fares.
• Including information regarding ancillary services, rich media, and optional personal information about the customer making the request
• Prices and product offerings could be customized to each request and generated in real time.
• NDC has started a discussion about “next-generation” approaches to airline pricing and revenue management.
Source: IATA
A Definitional Framework of Pricing Mechanisms
• Framework to describe selection of price for a given product.
• Traditional airline pricing and revenue management can be best classified as assortment optimization.
AssortmentOptimization
Dynamic PriceAdjustment
Start with assortment optimization, then adjust price points up or down in certain situations.
Continuous Pricing
Select prices freely from among a continuous range of possible values.
Create a finite set of possible price points, and then select prices from this set to offer to customers.Q
Y
M
B
X$249
$229
$499
$199
Price Selection Could Be Made Infrequently, or on a Transaction-by-Transaction Basis
• Pricing mechanisms vary on the number of allowable price points and the frequency with which prices are selected:
• With transactional continuous pricing, prices are computed individually for each transaction.
Frequency of Price Selection
Less Frequent(e.g., Weekly)
More Frequent(e.g., Transactional)
Number of Possible Price Points
Fewer Price Points(Assortment Opt.)
More Price Points(Continuous Pricing)
$$
Pricing Mechanisms in a Variety of Industries Were Reviewed
Industry Pricing MechanismPublicly-Available
Price Structure?Tra
vel-
Rela
ted
w/ A
sst. O
pt.
Airlines Assortment optimization Yes
Hotels Assortment optimization No
Passenger rail Assortment optimization Yes
Rental cars Assortment optimization No
Ind
ustr
ies
with
Advanced P
ricin
g
Online Retail Dynamic price adjustment No
Airbnb Continuous pricing No
On-demand transportDynamic price adjustment &
transactional continuous pricingYes/No
Airlines remain sophisticated compared to other travel-related industries, but other industries may not have the same technological constraints.
Six Next-Generation Pricing Mechanisms Are Currently Under Development in the Airline Industry
1. More Frequent Updating of Fare Structures
2. Dynamic Availability of Fare Products
4. Dynamic Pricing Engines
3. Advanced RBD Capabilities
6. Dynamic Offer Generation
5. Continuous Pricing
Most Complex
Least Complex
Mechanism 1: More Frequent Updating of Fare Structures• The frequency of fare filings has increased in recent years:
• More frequent filing of fares allows airlines to update the menu of possible price points more often.
• But the airline still practices assortment optimization by selecting prices among a pre-set menu of possible price points.
• At the limit, a unique fare structure could be filed for each departure day in each market.
2007 2017
Distribution FrequencyInternationalUS/CA Domestic
8 times/day3 times/day
Every hour4 times/day
Average Fare Changes per day 669k 3.9M
Source: ATPCO
Mechanism 2: Dynamic Availability of Fare Products
• Dynamic availability changes the RM availability of fare products based on characteristics of booking requests.
• Dynamic availability is an assortment optimization mechanism.
Y
B
M
QX
Class
If load factor < 50%and Corporate Code
= ABCDEFG…
Y
B
M
Q
Class
RM
Syste
m A
vailab
ilit
y Dyn
am
ic A
vaila
bility
XX
…open Class M for this transaction
Mechanism 3: Additional RBD Capabilities
• ATPCO’s Dynamic Pricing Working Group has investigated increasing the capabilities of existing RBDs.
• The current limit of 26 price points for each market may not be sufficient as airlines develop more complex products.
• With additional RBD capabilities, a single RBD could be divided into multiple price points that could be controlled independently.
• These changes would likely require more changes to existing technology than more frequent refiling of fares or dynamic availability.
Mechanism 4: Dynamic Pricing Engines• Dynamic Pricing Engines (DPEs) are a dynamic price adjustment mechanism
that adjusts (up or down) the prices of pre-filed fare products in certain situations.
• The DPE concept also emerged from the ATPCO Dynamic Pricing Working Group.
DynamicDiscounting
Dynamic Incrementing
In certain situations, apply a discount to the price of the lowest-available fare product for lower-WTP requests.
In certain situations, apply an increment to the price of the lowest-available fare product for higher-WTP requests.
Examples of Dynamic Pricing Mechanisms
Mechanism 5: Continuous Pricing
• Unlike Dynamic Pricing Engines, continuous pricing does not rely on a set of pre-determined price points.
• Mechanically, two possible options for implementing continuous pricing:• File a separate fare basis for each possible price point, then use continuous pricing to select which price to
display.
• Use the New Distribution Capability to distribute continuously chosen prices without reference to pre-determined price points.
• Either approach would require new WTP-based RM forecasting and optimization processes.• MIT PODS Consortium currently evaluating methods and outcomes of new approaches.
Mechanism 6: Dynamic Offer Generation• Dynamic offer generation combines the product creation and price
selection processes into a single mechanism:
Ancillaries Itineraries
• Preferences • WTP
Select and price a set of offer(s) that maximizes expected revenue from each booking request
Offer Set
Next-Gen. Pricing Mechanisms Could Have Wide-Reaching Effects on the Airline Industry• Next-generation mechanisms have the potential to impact:
Airline Competition
Legacy Processes
Consumers
Regulators
Airline Revenues
Simulations Have Suggested that Dynamic Pricing Engines Could Lead to Revenue Gains
0.4%
0.7%
1.3%
0.0%
0.5%
1.0%
1.5%
2.0%
2.5%
3.0%
3.5%
4.0%
Low Demand MediumDemand
High Demand% C
hange in A
L1 R
evenue fro
m R
ele
vant
Base
% Change in Revenue from Base when AL1 Uses
Dynamic Incrementing
3.7%
3.0%
2.1%
Low Demand MediumDemand
High Demand
% Change in Revenue from Base when AL1 Uses
Dynamic Discounting
Implications for Airline Competition
• Simulation studies do not capture the psychology of airline competition.
• Next-generation pricing technologies could lead to aggressive discounting to fill empty seats.
• There may be an incentive for each firm to undercut the other until one firm reaches its marginal cost.
• Could mitigated by the availability of improved pricing data.
AL1
$350$349$349$347
$99
…
AL2
$350$350$348$348
$100 (MC)…
Next-Generation Mechanisms Could Require Changes to Legacy Systems and Processes
Mechanism Most Likely to Affect…
More frequent updates to fare structuresPricingRevenue Accounting
Dynamic availability
PricingRevenue ManagementSales & Marketing
Additional RBD capabilities
Sales & MarketingDistributionPricing
Dynamic Pricing Engines
PricingRevenue ManagementRevenue IntegrityRevenue Accounting
Continuous PricingPricingRevenue Management
Dynamic Offer Generation
Merchandising and SalesDistributionRevenue ManagementPricing
Consumers May (or May Not) Notice Changes to Next-Generation Pricing Mechanisms
• Airline customers may be affected by changes to airline pricing mechanisms.
• The extent to which customers will notice and react to these changes will likely depend on the mechanism.
• Advanced RBD capabilities will remain out of sight for most customers, whereas dynamic offer generation and dynamic pricing engines could be more visible.
• Some customers believe that airlines are already using “dynamic” or even “customized” pricing—e.g., changing prices in response to a customer’s search or travel history.
• Customer perceptions of price “fairness” will likely depend on how next-generation mechanisms are framed.
Regulators Have Been Permissive of Next-Generation Pricing So Far: Will This Change?
• In the U.S., the DOT considered preliminary arguments about next-generation pricing in 2014.
• Price discrimination is not allowed on the basis of protected classes (i.e. race, gender, age), and anonymous shopping options will be required in the U.S.
• Transactional pricing methods may be more likely to draw regulatory review than other pricing mechanisms.
“We are tentatively not prepared to prohibit future [pricing] innovations that may better
match capacity with demand.”
Source: U.S. DOT Order to Show Cause for IATA Resolution 787, May 2014
Viewpoints on Next-Generation Pricing
• There is a debate in the industry regarding the feasibility and prudence of next-generation pricing mechanisms.
• Our report outlines two opposing viewpoints:
• These viewpoints do not necessarily reflect the views of the study’s authors, ATPCO or any particular airline.
Viewpoint 1: The Airline Industry Should Not Miss this Opportunity to Modernize its Pricing and RM Practices
Viewpoint 2: The Inherent Risks of Next-Generation Pricing Are Too High to Justify Further Development
Viewpoint 1: The Airline Industry Should Take Advantage of Next-Generation Pricing Mechanisms
The revenue potential of next-generation pricing and revenue management appears strong.
Next-generation pricing will finally allow airlines to move beyond constraints imposed by legacy systems.
New markets for information will emerge that are compatible with next-generation pricing mechanisms.
Some airlines will experiment with next-generation pricing, and a first-mover advantage appears to exist.
The greatest risk from next-generation pricing is likely from regulation and consumer reactions.
Viewpoint 2: Next-Generation Pricing is Too Risky to Justify its Implementation
Airlines are currently profitable and have well-understood pricing and revenue management processes.
Airlines are subject to inertia and resistance to change: why change what is not broken?
Next-generation pricing mechanisms risk destabilization of the underlying fare structure.
Next-generation pricing could reduce the ability of airlines to evaluate and respond to competitor price changes.
Next-generation pricing could have a disruptive influence on distribution relationships.
Conclusions – Preparing for a World with Next-Generation Airline Pricing
• Most stakeholders at airlines will have their own views that fall somewhere between these two extremes.
• The airline industry should proceed under the assumption that next-generation pricing will develop in some form.
• To prepare for next-generation pricing, airlines will need to develop new processes, techniques, and core competencies:
Automated processesfor next-gen pricing
New RM/forecasting for Conditional WTP
Interoperability with existing systems/practices
Corporate strategies fornew pricing mechanisms
HOW CAN YOU GET INVOLVED?
Use Research Attend Next Working GroupParticipate in Pilot*
Gianni [email protected]
Fred [email protected]
Melanie [email protected]
* C O N T A C T S
Manish NagpalVice President,
Program DevelopmentFarelogix
Manish Nagpal
A Complex Problem To Solve
FAR
ORD
DEL
FAR - DEL
LHR
FAR
ORD
LHR
DEL
600,000Fares
FAR - DELFAR – ORD, ORD – LHR, LHR- DEL
DEL – LHR, LHR- ORD, ORD- FAR
PRICING
FAR
ORD
LON
DEL
1.5Billion Fares
FAR - DELFAR – ORD, ORD – LHR, LHR- DEL
DEL – LHR, LHR- ORD, ORD- FAR
SHOPPING 50 inbound flights, 50 outbound flights andfind top 100 cheapest shopping solutions.
2500 flight combination * 600,000=
Oh, and it changes every hour
“Cat 25 fare by”
multiplier effect
Hundreds of rules per fare
20M filed fares
1 Airline
The Airline’s Challenge
What’s Driving Demand for Next Gen Shopping/Pricing?
Rise of Meta Search10,000 look to book is the norm
Dynamic Pricing Optimize offers (adjust ATP) using rules (today) or science (tomorrow)
NDC – Delivery vehicleAirline’s own shopping engine; airline as single source of truth
Pace of InnovationIntroduce products/differentiate on the airline’s timeline
NDC
Manish Nagpal
“ATPCO is essential.
FULL ATPCO PRICING CAPABILITY
• Proven system used by the world’s airlines
• Essential common model of distributed pricing
• Community model “by design”
New Generation Shopping and Pricing
– what’s it take?
New Gen Offer Optimization
OFFER
ATPCO
fare
New Gen Offer Optimization
OFFER
EXTERNAL DATA:
Social, Browser, Events, Trending, Buyer
Propensity, Weather, Trip purpose, Culture, Brand
Preferences
New Gen Offer Optimization
OFFER
AIRLINE DATA:
Purchase History, Load Factor, CRM, Search Data, RM, Loyalty, Preferences,
Corporate ID
EXTERNAL DATA:
Social, Browser, Events, Trending, Buyer
Propensity, Weather, Trip purpose, Culture, Brand
Preferences
New Gen Offer Optimization
OFFER
EXTERNAL DATA:
Social, Browser, Events, Trending, Buyer
Propensity, Weather, Trip purpose, Culture, Brand
Preferences
AIRLINE DATA:
Purchase History, Load Factor, CRM, Search Data, RM, Loyalty, Preferences,
Corporate ID
DATA SCIENCE
Machine Learning, AI, Predictive Analytics
New Gen Offer Optimization
BEST
OFFER
EXTERNAL DATA:
Social, Browser, Events, Trending, Buyer
Propensity, Weather, Trip purpose, Culture, Brand
Preferences
INFLUENCING RULES:
Channel, Corp. ID, Market, Season,
Preferred Connection, and more
DATA SCIENCE
Machine Learning, AI, Predictive Analytics
AIRLINE DATA:
Purchase History, Load Factor, CRM, Search Data, RM, Loyalty, Preferences,
Corporate ID
“Superior performance
PERFORMANT AND BUILT FOR PURPOSE
• Support Massive Volumes
• No look to book limitations/fees
• Subsecond response time
• No cache-based responses
• Adjust offers in real time
• Rules-based and easily updated by humans (today) or science (tomorrow)
New Generation Shopping and Pricing
– what’s it take?
“Maximum control for lowest cost of ownership
DESIGNED FOR AIRLINE CONTROLAND LOW COST OF
OWNERSHIP
• Linearly scalable on commodity hardware
• Limit PSS dependency
• Off-host availability and connection building
• Portable solution with airline hosting option
New Generation Shopping and Pricing
– what’s it take?
FLX Platform acting as DPE where offer is consumed by Sabre GDS
“New Gen Shopping and Pricing in Action
ATPCO Innovation Forum
Thursday, 11 October
FLXShop &Price
Thank you
John McBrideHead of Global Travel
Product ManagementPROS
John J. McBride, PhD
PROS
30 + years in revenue management
30300
Number of scientistsNumber of engineers 25% revenue into
R&D
About Vayant Travel Technologies
PROS + Vayant aligns revenue management with shopping and
merchandising, accelerating the path to full offer optimization
Enabling Airline Retailing &
Optimizing Distribution
Created in 2007
Global B2B IT service provider to airlines, agencies, and
PSSs
Supports Distribution and E-Commerce
GDS and PSS Agnostic
IATA NDC certified
Core values:
o Agility
o Control to Customer: Satisfaction & Value
o Advanced Technology
o Future-proof Solutions
What is Dynamic Pricing?
Decouples price from availability• Price is the determining factor – not class availability
• Fare families provide segmentation, but no classes within a family technically required
Generates an optimal price for an offer in real time• Price is based on expected demand and revenue to come
• Price depends on dimensions other than passenger type, itinerary, capacity, etc.
Dynamic Pricing #’s
Production Since 2013
10+ Airlines Using DP
5+ Airlines Migrating 2018
Shopping
Selling
IBE/GDSPROS
OneSearch
IBE/GDSPROS
OneSearch
Shopping Request
Offer Response
Request Info
Price and
Availability
Selling RequestCheck Availability
(Price)
Confirm Availability
(Price)
Issue Ticket
ATPCO
Check prices to issue ticket
Persisted ID needed
Airline pre-files set of discount codes
Increase
Revenue
+ (3 – 10)%
Increase
Conversion
+50%
Increase
Brand Loyalty
Why Dynamic Pricing?
Richard RatcliffSenior Research
ScientistSabre
Industry-standard Specifications for Air Dynamic Pricing Engines:
Progress Update
Richard Ratliff, Sr. Research Scientist, Sabre
ATPCO Global Conference
Washington, DC
Oct. 10, 2017
• Airline dynamic pricing has been discussed in various forums for more than a decade, but little progress was made due to both the disruptive nature of the technology and a lack of standardization
• Carriers want to be right-priced in all channels• Implication is that most carriers cannot unilaterally implement dynamic pricing in their CRS or with a
single GDS without causing issues with other GDS’s or agencies
• Beneficial for both airlines and GDS’s to adopt a common specification for dynamic pricing engines
• Less overall effort• Faster time to market
• Sabre and other parties recognized the need for an industry-standard specification in order to simplify and promote adoption
• Conduct underlying research work to demonstrate the benefits
• Good progress on DP research over the past few years• At AGIFORS RM in 2015, Amadeus reported simulation analysis on DPE
• Demonstrated net revenue improvements ranging from 5% to 7%
• Also supported by internal Sabre studies (unpublished) showing positive benefits
• MIT research showing benefits from DPE also been reported at AGIFORS and PODS meetings
• In 2017, Prof. Guillermo Gallego reported on benefits of dynamic pricing for both suppliers and consumers
Background
Basic Concepts of Dynamic Pricing
• For several years, Sabre has been working on the design and development of new shopping data sources, customer choice models and fare optimization algorithms to help airlines continually adapt their selling price positioning to maximize profits
• These market-adaptive dynamic pricing models are emerging as the next wave of advanced pricing and revenue management (RM) science
• Dynamic pricing team has been working on approaches on how these new tools could be employed by airlines for improved pricing and RM performance in their own and global distribution system channels
• Sabre’s vision of a comprehensive, real-time dynamic fares engine is based on three main components (documented in a white paper written in Feb. 2016)
• Customer Retailing• Accurate inventory controls• Market Adaptive Pricing (customer choice and
optimization models considering relative price and schedule quality)
Background
Elements of Market Adaptive Pricing
Optimize host itinerary prices trade off
selection prob vs. yield
Lowfare Search results by mkt, POS,
date
Assess attractiveness
Estimate itinerary probability of selection
Choice models provide ability to estimate selection probabilities considering competition
Price optimization model
Gather current competitive conditions
Customer-choice models
• What re-pricing opportunities do you observe in this JFK-ATH example?
• SU?• FI?• QR?
Market Adaptive Pricing Example: Discussion
Progress on the Pilot
InventoryInventoryDynamic Pricing Engine
API Validation Pilot: Progress-to-date (ATPCO, Farelogix, Sabre)
Shopping
ET server*
BSP*Revenue
Accounting*
Dynamic Pricing Engines
(Farelogixand
Sabre)
Booking*
Pricing
Ticketing*
Reporting
Sabre Vendor Systems
RET HOT
In-scope items shown in GREEN* Items not included in pilot testing
New items in RED
Inventory
LIFT
shoppingID (for reference only)
pricingID*
ATPCO Message Hub and DPE Private Fare Filing
External Data Inputs(e.g. Lowfare Shopping Cache)
new