SlideShare a Scribd company logo
1 of 11
Download to read offline
PRACTICE ARTICLE
An approach to offer management: maximizing sales with fare
products and ancillaries
Ben Vinod1 • Richard Ratliff1 • Vikram Jayaram1
Revised: 1 March 2017
Ó Macmillan Publishers Ltd 2018
Abstract With the growth in ancillary sales, an area of
increasing importance for airlines is the concept of offer
management, which entails the creation of dynamic, cus-
tom, personalized offers consisting of a flight itinerary and
ancillary products offered by an airline. This practice-ori-
ented, overview paper provides an end-to-end, future-ori-
ented framework for determining the composition of
optimal base fare and ancillary bundles by customer trip-
purpose segment followed by 1:1 personalization to max-
imize total sales. Our focus in this paper is primarily on the
proposed offer management framework and its sub-
components.
Keywords Revenue management Á Ancillary products Á
Willingness-to-pay Á Bundles Á Customer segmentation Á
Personalization Á Controlled experiments
Overview
Airline offer management is the practice of creating
dynamic, custom, personalized offers consisting of a flight
itinerary and ancillary products sold by an airline. An
important precept of offer management which enables
maximizing ancillary revenue is to maintain the identity of
each discrete ancillary in the bundle. The objective of offer
management is to offer the right bundles to the right cus-
tomer at the right price at the right time.
Offer management is an extension of airline revenue
management (RM). Given the growing importance of air-
line ancillary sales over the past decade and because the
type and volume of ancillary sales vary by fare product
type considered, many airlines have started incorporating
ancillary revenue streams into their RM systems. This
process of including the sale of ancillaries to the base fare
in the traditional revenue management context has been
referred to as ‘total revenue management’ (Rickey 2014).
Surveys (Alexander 2006) have shown that travelers
would pay for extra perks, such as more frequent flyer
miles, more overhead bin space, and the choice to sit in a
child-free section of the aircraft. Ancillary revenues and
new product offerings in the form of airline branded fares
have grown rapidly over the past decade, and new decision
support tools (including both offer management and total
RM) are emerging in this area.
During the early days of promoting ancillaries, ultra-low
cost carriers (LCC’s) such as Spirit Airlines, Ryanair and
easyJet gained notoriety due to their business models that
were based on maximizing total revenue. This total
includes revenues generated from fares and add-ons such as
fees for carry-on bags, checked luggage and seat assign-
ments. For these LCC’s, ancillary revenues can comprise
[11% of their total revenue, with some airlines reaching as
high as 25% (IdeaWorks 2016). The full service carriers on
the other hand were slow to adopt add-ons, but that’s
changing. In 2015, single-year growth of ancillary revenues
for full service carriers was greater than 13% (IdeaWorks
2016) comprised primarily of baggage fees, food and
beverages, and premium seat assignments. The sale of
miles or points to banks for co-branded credit card loyalty
programs was another large component.
The sale of ancillary services started in the airline direct
channel and has become important for the indirect channel
& Ben Vinod
Ben.Vinod@sabre.com
1
Sabre Research, 3150 Sabre Drive, Southlake, TX 76092,
USA
J Revenue Pricing Manag
https://doi.org/10.1057/s41272-017-0121-1
[i.e. Global Distribution Systems (GDS) and Online Travel
Agencies (OTA)] as well. Annual growth of ancillary sales
through Sabre channels worldwide was greater than 120%
during June 2015 compared to June 2016, a representative
period of peak summer travel, as shown in Fig. 1. Several
airlines worldwide have signed up to distribute ancillary
services through the Sabre channel.
Traditional ancillary sales pioneered by Spirit Airlines
in the 1990s focused ancillary sales efforts after the
booking was made (via e-mail in the later years). Besides
Spirit, other airlines that led the pack include Aer Lingus,
AirAsia, easyJet, Ryanair and Vueling.
An alternative to post-booking ancillary sales is the at-
booking approach; i.e. promoting bundles during the
shopping and booking process itself. The at-booking
approach is the main method for travel agents to sell
ancillaries, and it is also widely used in airline-direct
channels. This conclusion was based on an analysis of
EMD (Electronic Miscellaneous Document) sales in Sabre
over a 1-year period. EMD’s are an International Air
Transport Association (IATA) standard for electronically
tracking airline ancillary sales and revenue; they come in
two types:
EMD-A (air): associated with an air e-ticket
EMD-S (standalone): sold on a standalone basis
Table 1 shows a percentage breakdown of the types of
EMD’s issued in a large data sample comprised of both
travel agency and airline sales.
At-booking bundles will always have a lower sales
conversion rate than standalone air sales (due to the
increased purchase complexity and total price), but travel
agency capabilities to sell ancillaries are improving and
include many advancements over the past decade. For
example, specific functionality such as Midwest Airlines
pay-for-seats (‘‘Signature Seats’’) was introduced in Sabre
in 2007. Branded Fares availability, shopping, fare quote
and booking capabilities have been in place since 2010.
Two other important agency ancillary sales enablers are
more recent. In 2011, EMD functionality was added so
travel agents can shop, book and sell airline ancillaries
(either at or post-booking). Although we do not yet have
specific information to compare the effectiveness of at-
booking versus post-booking ancillary sales efforts, it is an
interesting topic for future research.
The promotion of service differentiation with branded
fare products and sale of ancillaries are collectively refer-
red to as ‘‘merchandising’’ or ‘‘airline retailing. Branded
fare families were first introduced by Air New Zealand in
2004 followed by Air Canada (Vinod 2008). Many airlines
followed and initially adopted branded fare products with
pre-defined bundles. Some airlines, led by US majors and
regional carriers, promoted unbundled pricing with a la
carte sale of ancillaries. However, today a large number of
airlines have evolved to a hybrid model. In this hybrid
model, branded fare products are retained and include
support for standalone sales of additional ancillaries that
are not included in the branded fare product. Other varia-
tions also exist; some airlines offer bundles of ancillaries
(e.g. Delta Air Lines’ ‘‘Lift Package’’ provides price sav-
ings for priority boarding when combined with a 1000 mile
frequent flyer booster). Figure 2 shows the evolving trend
in merchandising and retailing.
For example, one of the early adopters, Air Canada1
originally had three branded products in economy (Tango,
Tango Plus, Latitude) and two branded products in the
premium cabin (Executive Class Lowest, Executive Class
Flexible). Each branded fare product is a combination of an
airfare with bundled attributes, such as access to the Maple
Leaf lounge, priority baggage check in, fare refundability,
Fig. 1 Ancillary sales growth in Sabre channels
1
Air Canada currently has the Air Canada Altitude Program https://
altitude.aircanada.com/status/home.
B. Vinod et al.
advance seat selection, frequent flyer miles accrual, and
change fees. In addition, even with the purchase of a lower
valued branded fare product, customers can pay for specific
ancillary services such as advance seat selection. The dis-
tribution infrastructure of suppliers and the GDSs is
maturing to simplify branded fare filing and settlement.
The Airline Tariff Publishing Company (ATPCo) in con-
junction with the International Air Transport Association
(IATA) has provided a service fee solution. The fee types
are OA (booking fees), OB (ticketing fees) and OC (op-
tional service fees). Airlines that fail to adopt an a` la carte
pricing model will likely lose customers and potential
revenue (Nason 2009). In addition, the S8 record from
ATPCo that links the branded fare products to ancillaries is
being adopted by a large number of airlines such as Delta
Air Lines, Virgin Australia, and Kulula.
Significance of offer management
According to IdeaWorks (2016), airline ancillary revenue
statistics compiled from Air Transport World, Airline
Business and airline websites totaled $59.2 billion world-
wide in 2015 (projected to grow to $67.4 billion in 2016).
With this growth in airline ancillary sales, offer manage-
ment is increasingly important; it consists of creation of
dynamic custom personalized offers based on customer
traits, customer value score, flight and schedule attributes,
ancillaries and non-air products through the airline’s pre-
ferred channels of distribution. The initial focus has been to
sell ancillaries and sell-up of branded fares through the
direct and agency channels for online and codeshare part-
ner flights (Smith et al. 2006; Vinod and Moore 2009;
Zouaoui and Rao 2009).
The initial focus of airlines in the area of merchandising
and ancillaries has been on execution; these execution-
oriented systems provide the ability to sell ancillaries and
sell-up of branded fares through the direct channel and
agency channels for online and codeshare partner flights
(Vinod 2015). An offer management platform should
include decision support capabilities as well complement
real-time execution systems for generating targeted offers
to customers that maximize revenues.
The new distribution capability proposed by IATA is a
travel industry-supported program for the development and
adoption of a new data transmission standard for airlines to
communicate with retails sales outlets such as travel
agents, corporations and travelers. Towards this end,
application program interface (API) standards have been
defined for various functions as shopping, booking, ancil-
laries and fulfillment. The offer management solution for
both the direct and indirect channel are NDC compliant in
the send that they would use the same NDC API’s as a
means to communicated between the airline and the retail
outlet. Offer management is important for IATA’s New
Distribution Capability (NDC), and it also extends across
all channels of distribution.
This paper provides a framework for offer management
decision support including the process of determining the
composition of the optimal bundles for customer segments
and their respective price points (Fig. 2). A bundle consists
of both a base fare and targeted ancillaries with a total price
Table 1 Comparison of ancillary sales by EMD type for airline and
agency channels
Type Airlines (%) Travel agencies (%)
EMD-A (air e-ticket) 92.3 99.1
EMD-S (standalone) 7.7 0.9
Fig. 2 Trends in airline price bundling
An approach to offer management: maximizing sales with fare products and ancillaries
that is promoted by an airline to a customer segment or an
individual passenger.
The remainder of this paper involves the non-traditional
revenue management components as shown in Fig. 3 and is
organized as follows (the direction of red arrows in Fig. 3
indicate the step by step flow in the proposed offer man-
agement framework).
Trip-purpose segmentation
The main requirement in making effective bundled offers
to customers is to understand their wants and needs for
their upcoming trip. Determining the trip purpose (e.g.
business or leisure, duration, origin and destination, etc.)
has a strong bearing on the customer’s ancillaries pur-
chased and price sensitivity. So the need for trip-based
segmentation for bundling of offers cannot be understated.
Much of the marketing literature describes customer
profiles based on historical purchases and declared pref-
erences by specific customers (e.g. using their name, credit
card, phone number or frequent flyer ID); although this
information is important, it may not be applicable for
future flights. This is because the typical traveler has
multiple profiles depending on the purpose of the trip (e.g.
business, traveling with family, visiting friends and
relatives, etc.). Also in most travel purchase situations (e.g.
booking via a travel agent or a website without cookies
enabled), shopping for flights is done on an anonymous
basis.
In the authors’ experience, we find trip-based segmen-
tation to be the most practical initial method of classifying
customer types; it can be augmented with customer-
specific information in a later step when the actual cus-
tomer is declared such as when the flight purchase is
completed and full customer details are obtained. Classi-
fying customers can be based on trip characteristics such
as:
• origin and destination,
• how far in advance they book,
• distribution channel (e.g. airline direct or via a travel
agency),
• number in party,
• children or infants,
• length of stay,
• departure and return day-of-week,
• Saturday night stay-away classification,
• season,
• holiday versus non-holiday period,
• and presence or absence of special qualifiers on the
low-fare search, etc.
Fig. 3 Offer management framework
B. Vinod et al.
These are all examples of variables that can be used to help
ascertain the trip purpose and classify customers into
variants of leisure and business types. Table 2 illustrates an
example of trip-purpose segments for air. This particular
rule-based, ten cluster classification scheme was based on
1 year of a multi-agency and multi-airline Sabre ticketing
data sample (including both domestic and international
markets), but it is important to note that best-performing
rules would vary depending on the particular airlines,
agencies and networks considered. As a pragmatic matter,
our definitions of these segments are mutually exclusive
(based on fixed classification rules); although that can
potentially worsen the clustering performance, it does
make it easier to apply the algorithm in practice. Although,
we used data analysis in finding good breakpoints for each
dimension, we rounded them to nearby values which cor-
respond more closely with standard airline pricing restric-
tions to make them more intuitive to airline users.
We were concerned that using fixed, rules-based clas-
sification schemes could lead to unacceptably high
volatility among items mapped within each cluster when
compared to more standard clustering methods. As a quick
validation test, for each data point in our sample, we
computed its Euclidean distance to the centroid of the
cluster it was assigned to. Euclidean distance is the square
root of the sum of the squared differences across the dif-
ferent dimensions considered; in our case, the dimensions
used were all numeric (i.e. advance purchase period, length
of stay and number of passengers). We then summed the
Euclidean distances across all the data in our fixed classi-
fication scheme and compared those totals to two other
clustering approaches: Ward-based hierarchical clustering
with nine clusters (the tenth one was removed because it
only had a few observations mapped to it) and a random
clustering approach (purely to provide another baseline
reference). Ranking the approaches from least to greatest
total Euclidean distance we found:
1. Fixed classification (smallest total difference).
2. Hierarchical clustering.
3. Random assignment (greatest total difference).
Although there were many other validation tests we could
have performed, we concluded from the Euclidean distance
test that the rule-based assignments were not unreasonable
in terms of within-cluster minimization performance when
compared with a commonly-used automated approach.
In actual practice, our plan is to make additional dis-
tinctions. These include short haul (800 miles) versus
long haul (C800 miles) markets since purchase of ancil-
laries is influenced by length of haul. If Saturday Night
Stay is added, it is a mechanism to address day of week by
saying that the segment either includes or excluded
Saturday Night Stay. Selling channel is another obvious
dimension to consider (direct vs. indirect). Hence, if short
vs. long-haul, Saturday Night Stay and channel dimensions
are added, then the number of proposed air trip-purpose
segments would increase.
Market basket analysis
Ancillaries promoted by airlines fall into two broad cate-
gories: dominant ancillaries and non-dominant ancillaries.
Examples of dominant ancillaries are baggage, pre-re-
served seats, premium seats (extra leg room, aisle, window,
etc.). Non-dominant ancillaries are the extra features that
are nice to have but are usually not a necessity (e.g. meals
in coach, lounge access, newspaper, etc.). Due to differ-
ences in elapsed flight time, seating configuration, etc.,
each flight leg can have a unique relationship between the
dominant and non-dominant attributes.
Figure 4 shows an example of ancillary type associa-
tions in two different markets where a pre-reserved seat
was a dominant ancillary and the other non-dominant
ancillaries were co-purchased in each case. The number of
connection lines between the dominant and the non-dom-
inant ancillary in the two types indicate the strength of the
association or higher number of co-purchases. Notice in
this example that, even though the pre-reserved seat is the
dominant ancillary in both markets, the non-dominant
ancillaries vary in each case; this situation is typical in
practice. In Fig. 4, the acronym FFP stands for frequent
flying programs.
Table 2 Air trip purpose segment definitions
Air
segment
Id
Type Business/
leisure
Advance
purchase
Length
of stay
A1 Individual Business 0–6 days 0–1 days
A2 Individual Business/leisure 0–6 days 2? days
A3 Individual Business/leisure 7–13 days 0–3 days
A4 Individual Leisure 7–13 days 4? days
A5 Individual Business/leisure 14–20 days Any
A6 Individual Business 21? days 0–3 days
A7 Individual Leisure 21? days 4? days
A8 Couple (2) Leisure 0–20 days Any
A9 Couple (2) Leisure 21? days Any
A10 Family ([2) Leisure Any Any
An approach to offer management: maximizing sales with fare products and ancillaries
Willingness-to-pay
An active area of research is to quantify a customer’s
willingness to pay for ancillary services to determine their
value (Ratliff and Gallego 2013). Fortunately, during the
past few years, EMD data have become available, thus
allowing a consumer’s willingness to pay to be calibrated
for dynamic pricing of ancillaries.
Established methods include multinomial logit choice
analysis (Ben-Akiva and Lerman 1985; Train 2003; Bal-
combe et al. 2009), the van Westendorp pricing model (van
Westendorp 1976; Hague 2008) and conjoint analysis
(Green et al. 2001; Hair et al. 1984) which considers
tradeoffs between various combinations of price and pro-
duct features. A stated preference model (Martin et al.
2008) can be used to provide empirical evidence of esti-
mated valuations air passengers have on quality of service
attributes such as comfort, food, ticket change fees, fre-
quency and reliability. Survey tools can be used to estimate
a customer’s willingness to pay based on stated prefer-
ences. This approach is closer to a customer’s shopping
experience, and hence it is an improvement over single
dimensional surveys such as the van Westendorp (1976)
pricing model. The outputs of the model are customer
willing to pay for ancillary products and the relative
importance of each. This information is important in cor-
rectly setting prices for ancillaries and branded fare bun-
dles. Experimental design (described in ‘‘Controlled
experimentation’’ of this paper) is another emerging area
for willingness-to-pay estimation which is becoming more
mature (Besbes and Zeevi 2015).
An alternative approach which can be used to indirectly
estimate ancillary willingness to pay is based on the
(seemingly mild) assumption that ancillary sales have a
similar price elasticity to air travel in general. A simple
heuristic approach for estimating air travel demand price
elasticity using commonly available airline revenue man-
agement history data can be found in (Ratliff and Vinod
2016); this approach uses a cumulative sales history across
differently priced fare products together with log–log
regression methods to model the demand curve. For situ-
ations where relatively poor historical data are available for
ancillary willingness-to-pay estimation and/or direct
experimentation of different ancillary prices is impractical,
then assuming that ancillary price elasticity is similar to air
travel in general seems to be a reasonable alternative.
Competitive positioning
Perhaps the most important (but often ignored), component
of an effective retailing system is competitive positioning.
The benefits of effective customer retailing and accurate
availability information are well understood by the airlines,
but ongoing monitoring poses challenges owing to the data
required. In the context of offer management, such com-
petitive information includes both airfares as well as
ancillary prices. In this section, we will begin with airfare
competitiveness (for which some automated data sources
exist) followed by ancillaries (for which commercially
available comparison data are lacking).
Applications which help airlines to automatically
respond to changing competitive circumstances for airfares
are emerging; one recent reported example is the market
adaptive pricing (MAP) module (Ratliff 2016) which is a
component of an airfare dynamic pricing engine. Tradi-
tionally, most airlines have used the Airline Tariff Pub-
lishing Company (ATPCo) to file fares and compare prices.
Fig. 4 An example of dominant ancillary with its non-dominant association
B. Vinod et al.
While very useful for many purposes, unfortunately these
filed fare data do not consider availability. MAP processing
relies on real-time (or recent) low-fare search results
(which do consider availability) by market, point of sale
(POS) and date to understand an airline’s current compet-
itive positioning (considering both price and schedule
quality aspects). Customer choice and optimization models
are used which consider price, schedule quality and cus-
tomer segmentation to estimate probability of selection and
expected revenue of available itineraries. MAP provides a
real-time, ‘sense and respond’ capability for maximizing
expected itinerary profits (i.e. expected revenue net of
estimated flight opportunity costs); see Ratliff and Vinod
(2005) and Cary (2004).
Understanding your airline’s current air availability and
price competitiveness is at the core of market adaptive
pricing. The low-fare search results by market, date and
POS from GDS and third-party robotic vendors are a good
source for this type of information because they provide
results across multiple airlines. Ideally, these shopping data
would be comprised of streaming, real-time information.
However, given the large number of items (i.e. markets,
departure and return date combinations) that most airlines
want to monitor (often [1 million), cached shopping
results are typically used instead (Ratliff 2016). Real-time,
MAP optimization requires accurate estimates of marginal
cost (i.e. flight bid prices by itinerary); see Gallego and Hu
(2014) and Choubert et al. (2015). The potential benefits of
MAP are large; Choubert reported simulation results indi-
cating net revenue improvements in the 4–7% range.
Accurate product availability information also affects the
execution of an airline’s retailing strategy (Ratliff et al.
2003). Incorrect availability can lead to the wrong (or in
some cases no) product might be offered to the customer,
or inconsistencies can result between the products dis-
played during the shopping process versus those which are
actually bookable.
Inasmuch as airfare availability and shopping data pose
data collection challenges, at least there exist GDS (global
distribution system) and web-based tools for such infor-
mation; in fact, there are even some commercial vendors
providing shopping data solutions (e.g. Infare and QL2).
Unfortunately vendor solutions for gathering and compar-
ing competitor ancillary information are still immature, and
most airlines rely on manually conducting such competi-
tive comparisons. At present the best automated source of
ancillary price information are the ATPCo optional ser-
vices records:
• S5—sub code services record (detailed service defini-
tions to allow cross-airline comparisons).
• S6—concurrence record (for flight-related or rule
buster services to ascertain partner airlines’
concurrence).
• S7—provision record (to identify the travel, passenger,
geographic, flight, fare and sales requirements for each
specific sub code service in the S5).
Carriers often ask ‘‘how important is it to be competitive
with respect to specific ancillaries?’’ In our opinion, total
price (i.e. airfare ? ancillary purchases) is the main con-
sideration by customers. Because of the difficulties in
gathering current airfare and ancillary information, direct
comparisons at the detailed level are not easy. Fortunately,
although there are hundreds of different possible types of
ancillaries, the majority of the sales are comprised of only a
few major items. By knowing which ancillary items are the
most important ones by market, a carrier can simplify its
competitive comparisons by focusing its monitoring efforts
on airfares and top-selling ancillaries.
Dynamic pricing of bundled offer
Most airlines today have ancillary prices that are the same
(or nearly so) for all their markets in a given region (e.g.
$25 baggage fees on all domestic flights). In the future we
expect a broader trend of airlines experimenting with
market and date-specific ancillary prices, possibly even
further varying prices in other contexts such as time to
departure or scarcity of the ancillary products (e.g. extra
legroom seats). Ancillary pricing at these more detailed
levels will necessitate expanded data collection efforts and
further complicate an airline’s understanding of its com-
petitive position.
This variable ancillary pricing by market for services
such as pre-reserved seats based on competitive market
conditions and distance is frequently referred to as dynamic
pricing of ancillaries (Ødegaard and Wilson 2016). In
recent years there has been considerable research work in
the area of dynamic pricing for airfares (Choubert et al.
2015), and an ATPCo working group was formed in 2016
to create common input/output specifications for dynamic
pricing engines for air travel. Our expectation is that sim-
ilar capabilities will emerge in regard to ancillary dynamic
pricing in the near future.
Dynamic pricing of the bundle is based on a consumer’s
willingness to pay, which is calibrated by trip-purpose
segment (previously described in ‘‘Trip-purpose segmen-
tation’’). Figure 5 below illustrates how ancillary prefer-
ence rankings vary by trip-purpose segment for a large,
multi-airline data sample we tested.
An approach to offer management: maximizing sales with fare products and ancillaries
Personalization
Today’s customers have high expectations for personalized
services. Successful customer experiences promote rev-
enue growth and shall influence the future product invest-
ments from suppliers, intermediaries and software vendors
to ensure the loyalty and retention of profitable customers.
While trip-segmentation itself can be used to determine the
composition of the initial bundle to be offered to cus-
tomers, the composition of the bundle can be augmented
and refined once the customer’s identity is known. Past
travel history of the customer, preferences, loyalty program
affiliation and a customer’s preferences for the specific trip
all play a role in fine tuning and presenting the final trav-
eler preferred offer. The trade-off analytics algorithm is a
technique for ordering preferences by similarity to the ideal
solution. It ranks the bundles based on a traveler’s relative
trade-offs between the various attributes of the ancillaries
and the budget constraint. The algorithm scores all bundles
based on the trade-off evaluation based on the relative
importance of the ancillaries and displays the bundles in
the order of the scores. The resulting improvements to the
composition of the customized bundled offer and display
order should help improve sales conversion rates.
Branded Fares is also an important consideration in
creating the final offer. When a bundled offer is deter-
mined, it should be mapped to the attributes included in
each branded fare product to ensure the lowest cost option
is offered to the customer. Hence for example it may be
economical to offer a customer a cheaper branded fare
product and an ancillary priced separately than offering the
higher valued branded fare product with the ancillary
included. Hence offer management provides a range of
upsell opportunities tied to the various branded fare prod-
ucts that are available for sale, and the recommendation
can be based on the least total cost to the customer.
Controlled experimentation
Controlled experimentation is often useful either as an
alternative to model-based tools or when used in con-
junction with model-based tools as a validation and fine-
tuning mechanism. In recent years, direct experimentation
applications have become more widespread (e.g. Airbnb,
Amazon, Expedia, Facebook, Google, Microsoft, Staples,
and Walmart).
Business experimentation provides a direct approach to
running controlled tests to learn customer behavior. Unlike
model-based decision support tools, business experimen-
tation is in many respects simpler to use and implement.
Models range from simple to sophisticated. However, all of
them are only as good as the quality of the underlying input
estimates (e.g. demand forecasts, operational schedule,
weather patterns, guest checkout patterns, etc.). Big data
can only provide clues about the past behavior of cus-
tomers (i.e. out-of-sample predictions are questionable).
Also, there are often situations in which no historical data
exists (e.g. for new markets or products), and business
experimentation may be the only viable approach in the
short term.
The idea of business experimentation is to directly test
new prices, product designs, systems or processes com-
pared to current ones to see if there is performance
Fig. 5 An example of ancillary preferences ranked by revenue in each trip segment
B. Vinod et al.
improvement. Experimentation can help to answer com-
mon business questions such as:
• What is the best price point for maximizing revenue?
• Which particular algorithm or business approach
works best?
• How are my revenues and conversion rates impacted by
changes to my display rules?
• Which bundled offer maximizes revenue?
Thus, even with reasonably advanced models, estimates of
trip purpose, willingness-to-pay, ancillary preference order,
etc. are still imperfect. Direct, controlled experiments are
one way to estimate willingness-to-pay for ancillaries and
bundles. Such controlled experiments involve randomly
allocating different experiments (e.g. prices) across various
test subjects (which can be organized by session, departure
dates, markets, agencies, etc.). Randomization reduces bias
by equalizing other features that have not been explicitly
accounted for in the experimental design.
Simple binary tests (e.g. old vs. new) are commonly
referred to as A/B testing. One of the more advanced (and
widely used) experimental methods today is the multi-
armed bandit (MAB) algorithm, and there are a number of
good reasons it is used in our proposed offer management
framework. MAB implementations are more general, are
adaptive (i.e. learns over time), and can support running
multiple experiments simultaneously (e.g. A/B/C/D, etc.).
We chose the multi-armed bandit methodology because it
is statistically sound and open-source software is available.
The multi-armed bandit method continually learns as
experiments are conducted; the proportion of experiments
being run can be changed dynamically depending on cur-
rent results, so instead of fixed proportions, as test results
are obtained, more weight is given to those experiments
which show better performance (and vice-versa). Advo-
cates of MAB describe it as balancing ‘experimentation’
(e.g. testing different prices) with ‘exploitation’ (by
dynamically increasing the proportion of those experiments
showing better performance). An example of this dynamic
reweighting of the experiments over time is shown in
Fig. 6.
There are various heuristics used to compute the
dynamic reweighting proportions. The authors recommend
Bayesian approaches which use randomized probability
matching (e.g. Thompson Sampling). The underlying
principle behind these methods is to randomly allocate
each experiment according to its probability of being
optimal (Scott 2010).
Fig. 6 Multi-armed bandit dynamic reweighting example
Fig. 7 End-to-end offer management business process
An approach to offer management: maximizing sales with fare products and ancillaries
The end-to-end view: putting it all together
Synthesizing all the various components described in ear-
lier sections of this paper, we provide an illustration in
Fig. 7 which shows the end-to-end, at-booking, offer
management process. Notice that offer management is an
extension of existing systems; a key input in creating the
bundled offer price for seats and ancillaries is the tradi-
tional revenue management control mechanism (e.g. leg/
segment, O&D or dynamic pricing). The recommendation
engine and offer engine rely on historical booking data as
the source of data to recommend relevant content and
create the customer specific bundle respectively. The rec-
ommendation engine determines the relevant content that is
applicable to a segment based on historical behavior. The
offer engine takes the results from the recommendation
engine together with user specific preferences to generate
the 1-to-1 bundle based on a trade-off analytics model
which ranks bundles under consideration based on a trav-
eler’s relative trade-off between the various bundles sub-
ject to a budget constraint. The model scores all bundles
based on the trade-off evaluation using relative weights
which imply importance of the specific ancillary as part of
the bundle and displays the bundles in order of the scores.
As the figure indicates, the workflow begins with a user
executing a shopping request, selects an itinerary which is
followed by a ranked list of default bundles which are by
trip purpose segment. The user proceeds to select a default
bundle and modify the composition of the bundle- ancil-
laries and relative weights based on importance of the
preference before the trade-off analytics model recom-
mends a bundle specific to the traveler.
The offer management process for the post-booking
model is very similar, with the exception that the workflow
begins after the air booking has been made.
Conclusions
Decision support for offer management requires an
understanding of customer preferences and generating
pertinent targeted responses to customer requests during
the sales process life cycle. Generating a multitude of
offers to customers who have to select from a myriad of
options slows down the sales process and leads to aban-
donment. A key to customer centricity is the ability to offer
the right product or bundle to the right customer at the right
price at the right time based on derived customer prefer-
ences. Displaying targeted offers that resonate with cus-
tomers requires an investment in a data infrastructure and
advanced analytics to understand consumer behavior and
preferences. This approach in turn generates incremental
revenues with targeted offers and ensures repeatable,
profitable customers. Development of new business pro-
cesses and management tools to support total revenue
management will continue to be an important growth area.
References
Alexander, K.L. 2006. Paying more for small extras. Washington
Post, January 31. http://www.washingtonpost.com/wp-dyn/con
tent/article/2006/01/30/AR2006013001282.html.
Balcombe, K., I. Fraser, and L. Harris. 2009. Consumer willingness to
pay for in-flight service and comfort levels: A choice experi-
ment. Journal of Air Transport Management 15: 221–226.
Ben-Akiva, M., and S. Lerman. 1985. Discrete choice analysis:
Theory and application to travel demand. Cambridge, MA: MIT
Press.
Besbes, O., and A. Zeevi. 2015. On the (surprising) sufficiency of
linear models for dynamic pricing with demand learning.
Management Science 61 (4): 723–739.
Cary, D. 2004. A view from the inside. Journal of Pricing and
Revenue Management 3 (2): 200–203.
Choubert, L., T. Fiig, and V. Viale. 2015. ‘‘Amadeus Dynamic
Pricing’’, presentation at the AGIFORS Revenue Management
and Distribution Study Group meeting, Shanghai, China.
Gallego, G., and M. Hu. 2014. Dynamic pricing of perishable assets
under competition. Management Science 60 (5): 1241–1259.
Green, P.E., A.M. Krieger, and Y. Wind. 2001. Thirty years of
conjoint analysis: Reflections and prospects. Interfaces 31: S56–
S73.
Hague, N. 2008. The problem with price. B2B International. http://
www.b2binternational.com/library/whitepapers/pdf/the_pro
blem_with_price.pdf.
Hair, J.S., R.E. Anderson, and R.T. Tatham. 1984. Multivariate data
analysis with readings, 2nd ed. New York: Macmillan.
IdeaWorks. 2016. Airline ancillary revenue projected to be $67.4
billion worldwide in 2016. Press release, 29 Nov. 29, 2016. No.
115. http://www.ideaworkscompany.com/wp-content/uploads/
2016/11/Press-Release-115-Global-Estimate.pdf.
Martin, J.C., C. Roman, and R. Espino. 2008. Willingness to pay for
airline service quality. Transportation Reviews 28 (2): 199–217.
Nason, S.D. 2009. The future of a la carte pricing in the airline
industry. Journal of Revenue and Pricing Management 8 (5):
467–468.
Ødegaard, F., and J.G. Wilson. 2016. Dynamic pricing of primary
products and ancillary services. European Journal of Opera-
tional Research 251 (2): 586–599.
Ratliff, R.M., A. Walker, B.C. Smith, and T. Brice. 2003. Availability
based value creation method and systems, U.S. Patent
20030191725, filed March 26, 2003, and issued October 9, 2003.
Ratliff, R.M. 2016. A business overview of market adaptive dynamic
pricing. Sabre white paper.
Ratliff, R., and G. Gallego. 2013. Estimating sales and profitability
impacts of airline branded-fares product design and pricing
decisions using customer choice models. Journal of Revenue and
Pricing Management 12 (6): 509–523.
Ratliff, R.M., and B. Vinod. 2005. Airline pricing and revenue
management: A future outlook. Journal of Revenue and Pricing
Management 4 (3): 302–307.
Ratliff, R.M., and B. Vinod. 2016. An applied process for airline
strategic fare optimization. Journal of Revenue and Pricing
Management 15 (5): 320–333.
Rickey, D. 2014. Total revenue management. Ascend 13 (4): 20–22.
B. Vinod et al.
Scott, S. 2010. A modern Bayesian look at the multi-armed bandit.
Applied Stochastic Models in Business and Industry 26: 639–658.
Smith, B.C., R. Darrow, J. Elieson, D. Guenther, B.V. Rao, and F.
Zouaoui. 2006. Travelocity becomes a travel retailer. Interfaces
37 (1): 68–81.
Train, K.E. 2003. Discrete choice methods with simulation. New
York: Cambridge University Press.
van Westendorp, P.H. 1976. NSS—price sensitivity meter (PSM)—a
new approach to study consumer perception of price. Proceed-
ings of the ESOMAR Congress, Venice.
Vinod, B. 2008. The continuing evolution: Customer centric revenue
management. Journal of Revenue and Pricing Management 7
(1): 27–39.
Vinod, B. 2015. The expanding role of revenue management in the
airline industry. Journal of Revenue and Pricing Management 14
(6): 391–399.
Vinod, B., and K. Moore. 2009. Promoting branded fare families and
ancillary services: Merchandising and its impacts on the travel
value chain. Journal of Revenue and Pricing Management 8
(2–3): 174–186.
Zouaoui, F., and B.V. Rao. 2009. Dynamic pricing of opaque airline
tickets. Journal of Revenue and Pricing Management 8 (2–3):
148–154.
An approach to offer management: maximizing sales with fare products and ancillaries

More Related Content

What's hot

Southwest Airline 2009
Southwest Airline 2009Southwest Airline 2009
Southwest Airline 2009Shakhzod44
 
Easy Jet Case Study - Mis
Easy Jet Case Study   -  MisEasy Jet Case Study   -  Mis
Easy Jet Case Study - MisBERHMANI Samuel
 
Mod. 6: Segmentation
Mod. 6: SegmentationMod. 6: Segmentation
Mod. 6: SegmentationRaul Revuelta
 
Amadeus mobile solutions for airlines
Amadeus mobile solutions for airlinesAmadeus mobile solutions for airlines
Amadeus mobile solutions for airlinesYANNIS A. POLLALIS
 
Spirit Airlines: Strategic Management Case Study
Spirit Airlines: Strategic Management Case StudySpirit Airlines: Strategic Management Case Study
Spirit Airlines: Strategic Management Case StudyMarissa Pié
 
SWA Case Study Presentation
SWA Case Study PresentationSWA Case Study Presentation
SWA Case Study PresentationReisha Bernard
 
Global Strategy Assignment
Global Strategy AssignmentGlobal Strategy Assignment
Global Strategy AssignmentJames Bowyer
 
Airline ancillaries: What is working in today’s marketplace
Airline ancillaries: What is working in today’s marketplaceAirline ancillaries: What is working in today’s marketplace
Airline ancillaries: What is working in today’s marketplaceKevin May
 
Malaysia Airlines, Soaring with the Phoenix
Malaysia Airlines, Soaring with the PhoenixMalaysia Airlines, Soaring with the Phoenix
Malaysia Airlines, Soaring with the PhoenixJun Hao Lim
 
Capstone Research-Southwest Airlines
Capstone Research-Southwest AirlinesCapstone Research-Southwest Airlines
Capstone Research-Southwest Airlinescpedersen
 
Southwest Airlines Way
Southwest Airlines WaySouthwest Airlines Way
Southwest Airlines WayGMR Group
 
Airlines umair shah
Airlines   umair shahAirlines   umair shah
Airlines umair shahSaad Munir
 
Southwest airlines final
Southwest airlines finalSouthwest airlines final
Southwest airlines finalCullen Griffin
 
American Airlines Vs Southwest Airlines
American Airlines Vs Southwest AirlinesAmerican Airlines Vs Southwest Airlines
American Airlines Vs Southwest AirlinesDam Frank
 
Case Study Southwest
Case Study SouthwestCase Study Southwest
Case Study Southwesttltutortutor
 
Strategic Management - Southwest Airlines Review
Strategic Management - Southwest Airlines ReviewStrategic Management - Southwest Airlines Review
Strategic Management - Southwest Airlines ReviewStacey Troup
 
Airlines 2020 substitution and commoditization
Airlines 2020   substitution and commoditizationAirlines 2020   substitution and commoditization
Airlines 2020 substitution and commoditizationMarinet Ltd
 
Southwest Airlines : Case Study 2016 (Group Work)
Southwest Airlines : Case Study 2016 (Group Work)Southwest Airlines : Case Study 2016 (Group Work)
Southwest Airlines : Case Study 2016 (Group Work)Sriwiyata Ismail Zainuddin
 

What's hot (19)

Southwest Airline 2009
Southwest Airline 2009Southwest Airline 2009
Southwest Airline 2009
 
Easy Jet Case Study - Mis
Easy Jet Case Study   -  MisEasy Jet Case Study   -  Mis
Easy Jet Case Study - Mis
 
Mod. 6: Segmentation
Mod. 6: SegmentationMod. 6: Segmentation
Mod. 6: Segmentation
 
Amadeus mobile solutions for airlines
Amadeus mobile solutions for airlinesAmadeus mobile solutions for airlines
Amadeus mobile solutions for airlines
 
Low frill airlines
Low frill airlinesLow frill airlines
Low frill airlines
 
Spirit Airlines: Strategic Management Case Study
Spirit Airlines: Strategic Management Case StudySpirit Airlines: Strategic Management Case Study
Spirit Airlines: Strategic Management Case Study
 
SWA Case Study Presentation
SWA Case Study PresentationSWA Case Study Presentation
SWA Case Study Presentation
 
Global Strategy Assignment
Global Strategy AssignmentGlobal Strategy Assignment
Global Strategy Assignment
 
Airline ancillaries: What is working in today’s marketplace
Airline ancillaries: What is working in today’s marketplaceAirline ancillaries: What is working in today’s marketplace
Airline ancillaries: What is working in today’s marketplace
 
Malaysia Airlines, Soaring with the Phoenix
Malaysia Airlines, Soaring with the PhoenixMalaysia Airlines, Soaring with the Phoenix
Malaysia Airlines, Soaring with the Phoenix
 
Capstone Research-Southwest Airlines
Capstone Research-Southwest AirlinesCapstone Research-Southwest Airlines
Capstone Research-Southwest Airlines
 
Southwest Airlines Way
Southwest Airlines WaySouthwest Airlines Way
Southwest Airlines Way
 
Airlines umair shah
Airlines   umair shahAirlines   umair shah
Airlines umair shah
 
Southwest airlines final
Southwest airlines finalSouthwest airlines final
Southwest airlines final
 
American Airlines Vs Southwest Airlines
American Airlines Vs Southwest AirlinesAmerican Airlines Vs Southwest Airlines
American Airlines Vs Southwest Airlines
 
Case Study Southwest
Case Study SouthwestCase Study Southwest
Case Study Southwest
 
Strategic Management - Southwest Airlines Review
Strategic Management - Southwest Airlines ReviewStrategic Management - Southwest Airlines Review
Strategic Management - Southwest Airlines Review
 
Airlines 2020 substitution and commoditization
Airlines 2020   substitution and commoditizationAirlines 2020   substitution and commoditization
Airlines 2020 substitution and commoditization
 
Southwest Airlines : Case Study 2016 (Group Work)
Southwest Airlines : Case Study 2016 (Group Work)Southwest Airlines : Case Study 2016 (Group Work)
Southwest Airlines : Case Study 2016 (Group Work)
 

Similar to An approach to offer management: maximizing sales with fare products and ancillaries

Ancillary Revenues soaring Opportunities in Airline Operations
Ancillary Revenues soaring Opportunities in Airline OperationsAncillary Revenues soaring Opportunities in Airline Operations
Ancillary Revenues soaring Opportunities in Airline OperationsNIIT Technologies
 
16Virgin Atlantic SWOT AnalysisIntrodu
16Virgin Atlantic SWOT AnalysisIntrodu16Virgin Atlantic SWOT AnalysisIntrodu
16Virgin Atlantic SWOT AnalysisIntroduEttaBenton28
 
The Future of Airline Retail - Fast Future Report 19 07 11
The Future of Airline Retail - Fast Future Report 19 07 11The Future of Airline Retail - Fast Future Report 19 07 11
The Future of Airline Retail - Fast Future Report 19 07 11Rohit Talwar
 
Intern report on marketing and sales tactics of travel Agency
Intern report on marketing and sales tactics of travel AgencyIntern report on marketing and sales tactics of travel Agency
Intern report on marketing and sales tactics of travel AgencyMd. Mamun Hasan Biddut
 
Article review on Strategic Methods for cost cutting and increasing profits i...
Article review on Strategic Methods for cost cutting and increasing profits i...Article review on Strategic Methods for cost cutting and increasing profits i...
Article review on Strategic Methods for cost cutting and increasing profits i...TDakshinamurthyMBAKB
 
Southwest Airlines And United Airlines
Southwest Airlines And United AirlinesSouthwest Airlines And United Airlines
Southwest Airlines And United AirlinesPeggy Johnson
 
Ryanair - Brand Audit
Ryanair - Brand AuditRyanair - Brand Audit
Ryanair - Brand AuditJoshua Peace
 
Ryanair industry analysis – A case study report
Ryanair industry analysis – A case study reportRyanair industry analysis – A case study report
Ryanair industry analysis – A case study reportPaulius Bagdanskas
 

Similar to An approach to offer management: maximizing sales with fare products and ancillaries (18)

Ancillary Revenues soaring Opportunities in Airline Operations
Ancillary Revenues soaring Opportunities in Airline OperationsAncillary Revenues soaring Opportunities in Airline Operations
Ancillary Revenues soaring Opportunities in Airline Operations
 
AIR FRANCE
AIR FRANCEAIR FRANCE
AIR FRANCE
 
AIR FRANCE
AIR FRANCEAIR FRANCE
AIR FRANCE
 
Presentation1
Presentation1Presentation1
Presentation1
 
16Virgin Atlantic SWOT AnalysisIntrodu
16Virgin Atlantic SWOT AnalysisIntrodu16Virgin Atlantic SWOT AnalysisIntrodu
16Virgin Atlantic SWOT AnalysisIntrodu
 
Star alliance
Star allianceStar alliance
Star alliance
 
Presentation by CBS students on Star alliance
Presentation by CBS students on Star alliancePresentation by CBS students on Star alliance
Presentation by CBS students on Star alliance
 
Presentatin by CBS students on Star alliance
Presentatin by CBS students on Star alliancePresentatin by CBS students on Star alliance
Presentatin by CBS students on Star alliance
 
Core product
Core productCore product
Core product
 
The Future of Airline Retail - Fast Future Report 19 07 11
The Future of Airline Retail - Fast Future Report 19 07 11The Future of Airline Retail - Fast Future Report 19 07 11
The Future of Airline Retail - Fast Future Report 19 07 11
 
Achievement
AchievementAchievement
Achievement
 
Intern report on marketing and sales tactics of travel Agency
Intern report on marketing and sales tactics of travel AgencyIntern report on marketing and sales tactics of travel Agency
Intern report on marketing and sales tactics of travel Agency
 
Article review on Strategic Methods for cost cutting and increasing profits i...
Article review on Strategic Methods for cost cutting and increasing profits i...Article review on Strategic Methods for cost cutting and increasing profits i...
Article review on Strategic Methods for cost cutting and increasing profits i...
 
E Com Price Line
E Com Price LineE Com Price Line
E Com Price Line
 
Southwest Airlines And United Airlines
Southwest Airlines And United AirlinesSouthwest Airlines And United Airlines
Southwest Airlines And United Airlines
 
Ryanair - Brand Audit
Ryanair - Brand AuditRyanair - Brand Audit
Ryanair - Brand Audit
 
Ryanair industry analysis – A case study report
Ryanair industry analysis – A case study reportRyanair industry analysis – A case study report
Ryanair industry analysis – A case study report
 
Environment-of-business
Environment-of-businessEnvironment-of-business
Environment-of-business
 

More from Pioneer Natural Resources

Hydraulic Fracturing Stimulation Monitoring with Distributed Fiber Optic Sens...
Hydraulic Fracturing Stimulation Monitoring with Distributed Fiber Optic Sens...Hydraulic Fracturing Stimulation Monitoring with Distributed Fiber Optic Sens...
Hydraulic Fracturing Stimulation Monitoring with Distributed Fiber Optic Sens...Pioneer Natural Resources
 
Interpretation Special-Section: Insights into digital oilfield data using ar...
 Interpretation Special-Section: Insights into digital oilfield data using ar... Interpretation Special-Section: Insights into digital oilfield data using ar...
Interpretation Special-Section: Insights into digital oilfield data using ar...Pioneer Natural Resources
 
A comparison of classification techniques for seismic facies recognition
A comparison of classification techniques for seismic facies recognitionA comparison of classification techniques for seismic facies recognition
A comparison of classification techniques for seismic facies recognitionPioneer Natural Resources
 
3D Gravity Modeling of Osage County Oklahoma for 3D Gravity Interpretation
3D Gravity Modeling of Osage County Oklahoma for 3D Gravity Interpretation3D Gravity Modeling of Osage County Oklahoma for 3D Gravity Interpretation
3D Gravity Modeling of Osage County Oklahoma for 3D Gravity InterpretationPioneer Natural Resources
 
OPTIMIZED RATE ALLOCATION OF HYPERSPECTRAL IMAGES IN COMPRESSED DOMAIN USING ...
OPTIMIZED RATE ALLOCATION OF HYPERSPECTRAL IMAGES IN COMPRESSED DOMAIN USING ...OPTIMIZED RATE ALLOCATION OF HYPERSPECTRAL IMAGES IN COMPRESSED DOMAIN USING ...
OPTIMIZED RATE ALLOCATION OF HYPERSPECTRAL IMAGES IN COMPRESSED DOMAIN USING ...Pioneer Natural Resources
 
Receiver deghosting method to mitigate F-­K transform artifacts: A non-­windo...
Receiver deghosting method to mitigate F-­K transform artifacts: A non-­windo...Receiver deghosting method to mitigate F-­K transform artifacts: A non-­windo...
Receiver deghosting method to mitigate F-­K transform artifacts: A non-­windo...Pioneer Natural Resources
 
Distance Metric Based Multi-Attribute Seismic Facies Classification to Identi...
Distance Metric Based Multi-Attribute Seismic Facies Classification to Identi...Distance Metric Based Multi-Attribute Seismic Facies Classification to Identi...
Distance Metric Based Multi-Attribute Seismic Facies Classification to Identi...Pioneer Natural Resources
 
Active learning algorithms in seismic facies classification
Active learning algorithms in seismic facies classificationActive learning algorithms in seismic facies classification
Active learning algorithms in seismic facies classificationPioneer Natural Resources
 
A Rapid Location Independent Full Tensor Gravity Algorithm
A Rapid Location Independent Full Tensor Gravity AlgorithmA Rapid Location Independent Full Tensor Gravity Algorithm
A Rapid Location Independent Full Tensor Gravity AlgorithmPioneer Natural Resources
 
Directional Analysis and Filtering for Dust Storm detection in NOAA-AVHRR Ima...
Directional Analysis and Filtering for Dust Storm detection in NOAA-AVHRR Ima...Directional Analysis and Filtering for Dust Storm detection in NOAA-AVHRR Ima...
Directional Analysis and Filtering for Dust Storm detection in NOAA-AVHRR Ima...Pioneer Natural Resources
 
Detection and Classification in Hyperspectral Images using Rate Distortion an...
Detection and Classification in Hyperspectral Images using Rate Distortion an...Detection and Classification in Hyperspectral Images using Rate Distortion an...
Detection and Classification in Hyperspectral Images using Rate Distortion an...Pioneer Natural Resources
 

More from Pioneer Natural Resources (17)

Hydraulic Fracturing Stimulation Monitoring with Distributed Fiber Optic Sens...
Hydraulic Fracturing Stimulation Monitoring with Distributed Fiber Optic Sens...Hydraulic Fracturing Stimulation Monitoring with Distributed Fiber Optic Sens...
Hydraulic Fracturing Stimulation Monitoring with Distributed Fiber Optic Sens...
 
Interpretation Special-Section: Insights into digital oilfield data using ar...
 Interpretation Special-Section: Insights into digital oilfield data using ar... Interpretation Special-Section: Insights into digital oilfield data using ar...
Interpretation Special-Section: Insights into digital oilfield data using ar...
 
Machine learning for Seismic Data Analysis
Machine learning for Seismic Data AnalysisMachine learning for Seismic Data Analysis
Machine learning for Seismic Data Analysis
 
A comparison of classification techniques for seismic facies recognition
A comparison of classification techniques for seismic facies recognitionA comparison of classification techniques for seismic facies recognition
A comparison of classification techniques for seismic facies recognition
 
3D Gravity Modeling of Osage County Oklahoma for 3D Gravity Interpretation
3D Gravity Modeling of Osage County Oklahoma for 3D Gravity Interpretation3D Gravity Modeling of Osage County Oklahoma for 3D Gravity Interpretation
3D Gravity Modeling of Osage County Oklahoma for 3D Gravity Interpretation
 
OPTIMIZED RATE ALLOCATION OF HYPERSPECTRAL IMAGES IN COMPRESSED DOMAIN USING ...
OPTIMIZED RATE ALLOCATION OF HYPERSPECTRAL IMAGES IN COMPRESSED DOMAIN USING ...OPTIMIZED RATE ALLOCATION OF HYPERSPECTRAL IMAGES IN COMPRESSED DOMAIN USING ...
OPTIMIZED RATE ALLOCATION OF HYPERSPECTRAL IMAGES IN COMPRESSED DOMAIN USING ...
 
Receiver deghosting method to mitigate F-­K transform artifacts: A non-­windo...
Receiver deghosting method to mitigate F-­K transform artifacts: A non-­windo...Receiver deghosting method to mitigate F-­K transform artifacts: A non-­windo...
Receiver deghosting method to mitigate F-­K transform artifacts: A non-­windo...
 
Distance Metric Based Multi-Attribute Seismic Facies Classification to Identi...
Distance Metric Based Multi-Attribute Seismic Facies Classification to Identi...Distance Metric Based Multi-Attribute Seismic Facies Classification to Identi...
Distance Metric Based Multi-Attribute Seismic Facies Classification to Identi...
 
Active learning algorithms in seismic facies classification
Active learning algorithms in seismic facies classificationActive learning algorithms in seismic facies classification
Active learning algorithms in seismic facies classification
 
A Rapid Location Independent Full Tensor Gravity Algorithm
A Rapid Location Independent Full Tensor Gravity AlgorithmA Rapid Location Independent Full Tensor Gravity Algorithm
A Rapid Location Independent Full Tensor Gravity Algorithm
 
2009 spie hmm
2009 spie hmm2009 spie hmm
2009 spie hmm
 
Directional Analysis and Filtering for Dust Storm detection in NOAA-AVHRR Ima...
Directional Analysis and Filtering for Dust Storm detection in NOAA-AVHRR Ima...Directional Analysis and Filtering for Dust Storm detection in NOAA-AVHRR Ima...
Directional Analysis and Filtering for Dust Storm detection in NOAA-AVHRR Ima...
 
2009 asilomar
2009 asilomar2009 asilomar
2009 asilomar
 
2008 spie gmm
2008 spie gmm2008 spie gmm
2008 spie gmm
 
2007 asprs
2007 asprs2007 asprs
2007 asprs
 
2006 ssiai
2006 ssiai2006 ssiai
2006 ssiai
 
Detection and Classification in Hyperspectral Images using Rate Distortion an...
Detection and Classification in Hyperspectral Images using Rate Distortion an...Detection and Classification in Hyperspectral Images using Rate Distortion an...
Detection and Classification in Hyperspectral Images using Rate Distortion an...
 

Recently uploaded

From idea to production in a day – Leveraging Azure ML and Streamlit to build...
From idea to production in a day – Leveraging Azure ML and Streamlit to build...From idea to production in a day – Leveraging Azure ML and Streamlit to build...
From idea to production in a day – Leveraging Azure ML and Streamlit to build...Florian Roscheck
 
1:1定制(UQ毕业证)昆士兰大学毕业证成绩单修改留信学历认证原版一模一样
1:1定制(UQ毕业证)昆士兰大学毕业证成绩单修改留信学历认证原版一模一样1:1定制(UQ毕业证)昆士兰大学毕业证成绩单修改留信学历认证原版一模一样
1:1定制(UQ毕业证)昆士兰大学毕业证成绩单修改留信学历认证原版一模一样vhwb25kk
 
20240419 - Measurecamp Amsterdam - SAM.pdf
20240419 - Measurecamp Amsterdam - SAM.pdf20240419 - Measurecamp Amsterdam - SAM.pdf
20240419 - Measurecamp Amsterdam - SAM.pdfHuman37
 
VIP Call Girls Service Miyapur Hyderabad Call +91-8250192130
VIP Call Girls Service Miyapur Hyderabad Call +91-8250192130VIP Call Girls Service Miyapur Hyderabad Call +91-8250192130
VIP Call Girls Service Miyapur Hyderabad Call +91-8250192130Suhani Kapoor
 
Kantar AI Summit- Under Embargo till Wednesday, 24th April 2024, 4 PM, IST.pdf
Kantar AI Summit- Under Embargo till Wednesday, 24th April 2024, 4 PM, IST.pdfKantar AI Summit- Under Embargo till Wednesday, 24th April 2024, 4 PM, IST.pdf
Kantar AI Summit- Under Embargo till Wednesday, 24th April 2024, 4 PM, IST.pdfSocial Samosa
 
꧁❤ Greater Noida Call Girls Delhi ❤꧂ 9711199171 ☎️ Hard And Sexy Vip Call
꧁❤ Greater Noida Call Girls Delhi ❤꧂ 9711199171 ☎️ Hard And Sexy Vip Call꧁❤ Greater Noida Call Girls Delhi ❤꧂ 9711199171 ☎️ Hard And Sexy Vip Call
꧁❤ Greater Noida Call Girls Delhi ❤꧂ 9711199171 ☎️ Hard And Sexy Vip Callshivangimorya083
 
PKS-TGC-1084-630 - Stage 1 Proposal.pptx
PKS-TGC-1084-630 - Stage 1 Proposal.pptxPKS-TGC-1084-630 - Stage 1 Proposal.pptx
PKS-TGC-1084-630 - Stage 1 Proposal.pptxPramod Kumar Srivastava
 
Data Science Jobs and Salaries Analysis.pptx
Data Science Jobs and Salaries Analysis.pptxData Science Jobs and Salaries Analysis.pptx
Data Science Jobs and Salaries Analysis.pptxFurkanTasci3
 
INTERNSHIP ON PURBASHA COMPOSITE TEX LTD
INTERNSHIP ON PURBASHA COMPOSITE TEX LTDINTERNSHIP ON PURBASHA COMPOSITE TEX LTD
INTERNSHIP ON PURBASHA COMPOSITE TEX LTDRafezzaman
 
04242024_CCC TUG_Joins and Relationships
04242024_CCC TUG_Joins and Relationships04242024_CCC TUG_Joins and Relationships
04242024_CCC TUG_Joins and Relationshipsccctableauusergroup
 
办理学位证中佛罗里达大学毕业证,UCF成绩单原版一比一
办理学位证中佛罗里达大学毕业证,UCF成绩单原版一比一办理学位证中佛罗里达大学毕业证,UCF成绩单原版一比一
办理学位证中佛罗里达大学毕业证,UCF成绩单原版一比一F sss
 
GA4 Without Cookies [Measure Camp AMS]
GA4 Without Cookies [Measure Camp AMS]GA4 Without Cookies [Measure Camp AMS]
GA4 Without Cookies [Measure Camp AMS]📊 Markus Baersch
 
代办国外大学文凭《原版美国UCLA文凭证书》加州大学洛杉矶分校毕业证制作成绩单修改
代办国外大学文凭《原版美国UCLA文凭证书》加州大学洛杉矶分校毕业证制作成绩单修改代办国外大学文凭《原版美国UCLA文凭证书》加州大学洛杉矶分校毕业证制作成绩单修改
代办国外大学文凭《原版美国UCLA文凭证书》加州大学洛杉矶分校毕业证制作成绩单修改atducpo
 
Customer Service Analytics - Make Sense of All Your Data.pptx
Customer Service Analytics - Make Sense of All Your Data.pptxCustomer Service Analytics - Make Sense of All Your Data.pptx
Customer Service Analytics - Make Sense of All Your Data.pptxEmmanuel Dauda
 
Industrialised data - the key to AI success.pdf
Industrialised data - the key to AI success.pdfIndustrialised data - the key to AI success.pdf
Industrialised data - the key to AI success.pdfLars Albertsson
 
办理(Vancouver毕业证书)加拿大温哥华岛大学毕业证成绩单原版一比一
办理(Vancouver毕业证书)加拿大温哥华岛大学毕业证成绩单原版一比一办理(Vancouver毕业证书)加拿大温哥华岛大学毕业证成绩单原版一比一
办理(Vancouver毕业证书)加拿大温哥华岛大学毕业证成绩单原版一比一F La
 
Building on a FAIRly Strong Foundation to Connect Academic Research to Transl...
Building on a FAIRly Strong Foundation to Connect Academic Research to Transl...Building on a FAIRly Strong Foundation to Connect Academic Research to Transl...
Building on a FAIRly Strong Foundation to Connect Academic Research to Transl...Jack DiGiovanna
 
Amazon TQM (2) Amazon TQM (2)Amazon TQM (2).pptx
Amazon TQM (2) Amazon TQM (2)Amazon TQM (2).pptxAmazon TQM (2) Amazon TQM (2)Amazon TQM (2).pptx
Amazon TQM (2) Amazon TQM (2)Amazon TQM (2).pptxAbdelrhman abooda
 

Recently uploaded (20)

From idea to production in a day – Leveraging Azure ML and Streamlit to build...
From idea to production in a day – Leveraging Azure ML and Streamlit to build...From idea to production in a day – Leveraging Azure ML and Streamlit to build...
From idea to production in a day – Leveraging Azure ML and Streamlit to build...
 
1:1定制(UQ毕业证)昆士兰大学毕业证成绩单修改留信学历认证原版一模一样
1:1定制(UQ毕业证)昆士兰大学毕业证成绩单修改留信学历认证原版一模一样1:1定制(UQ毕业证)昆士兰大学毕业证成绩单修改留信学历认证原版一模一样
1:1定制(UQ毕业证)昆士兰大学毕业证成绩单修改留信学历认证原版一模一样
 
E-Commerce Order PredictionShraddha Kamble.pptx
E-Commerce Order PredictionShraddha Kamble.pptxE-Commerce Order PredictionShraddha Kamble.pptx
E-Commerce Order PredictionShraddha Kamble.pptx
 
20240419 - Measurecamp Amsterdam - SAM.pdf
20240419 - Measurecamp Amsterdam - SAM.pdf20240419 - Measurecamp Amsterdam - SAM.pdf
20240419 - Measurecamp Amsterdam - SAM.pdf
 
VIP Call Girls Service Miyapur Hyderabad Call +91-8250192130
VIP Call Girls Service Miyapur Hyderabad Call +91-8250192130VIP Call Girls Service Miyapur Hyderabad Call +91-8250192130
VIP Call Girls Service Miyapur Hyderabad Call +91-8250192130
 
Kantar AI Summit- Under Embargo till Wednesday, 24th April 2024, 4 PM, IST.pdf
Kantar AI Summit- Under Embargo till Wednesday, 24th April 2024, 4 PM, IST.pdfKantar AI Summit- Under Embargo till Wednesday, 24th April 2024, 4 PM, IST.pdf
Kantar AI Summit- Under Embargo till Wednesday, 24th April 2024, 4 PM, IST.pdf
 
꧁❤ Greater Noida Call Girls Delhi ❤꧂ 9711199171 ☎️ Hard And Sexy Vip Call
꧁❤ Greater Noida Call Girls Delhi ❤꧂ 9711199171 ☎️ Hard And Sexy Vip Call꧁❤ Greater Noida Call Girls Delhi ❤꧂ 9711199171 ☎️ Hard And Sexy Vip Call
꧁❤ Greater Noida Call Girls Delhi ❤꧂ 9711199171 ☎️ Hard And Sexy Vip Call
 
Call Girls in Saket 99530🔝 56974 Escort Service
Call Girls in Saket 99530🔝 56974 Escort ServiceCall Girls in Saket 99530🔝 56974 Escort Service
Call Girls in Saket 99530🔝 56974 Escort Service
 
PKS-TGC-1084-630 - Stage 1 Proposal.pptx
PKS-TGC-1084-630 - Stage 1 Proposal.pptxPKS-TGC-1084-630 - Stage 1 Proposal.pptx
PKS-TGC-1084-630 - Stage 1 Proposal.pptx
 
Data Science Jobs and Salaries Analysis.pptx
Data Science Jobs and Salaries Analysis.pptxData Science Jobs and Salaries Analysis.pptx
Data Science Jobs and Salaries Analysis.pptx
 
INTERNSHIP ON PURBASHA COMPOSITE TEX LTD
INTERNSHIP ON PURBASHA COMPOSITE TEX LTDINTERNSHIP ON PURBASHA COMPOSITE TEX LTD
INTERNSHIP ON PURBASHA COMPOSITE TEX LTD
 
04242024_CCC TUG_Joins and Relationships
04242024_CCC TUG_Joins and Relationships04242024_CCC TUG_Joins and Relationships
04242024_CCC TUG_Joins and Relationships
 
办理学位证中佛罗里达大学毕业证,UCF成绩单原版一比一
办理学位证中佛罗里达大学毕业证,UCF成绩单原版一比一办理学位证中佛罗里达大学毕业证,UCF成绩单原版一比一
办理学位证中佛罗里达大学毕业证,UCF成绩单原版一比一
 
GA4 Without Cookies [Measure Camp AMS]
GA4 Without Cookies [Measure Camp AMS]GA4 Without Cookies [Measure Camp AMS]
GA4 Without Cookies [Measure Camp AMS]
 
代办国外大学文凭《原版美国UCLA文凭证书》加州大学洛杉矶分校毕业证制作成绩单修改
代办国外大学文凭《原版美国UCLA文凭证书》加州大学洛杉矶分校毕业证制作成绩单修改代办国外大学文凭《原版美国UCLA文凭证书》加州大学洛杉矶分校毕业证制作成绩单修改
代办国外大学文凭《原版美国UCLA文凭证书》加州大学洛杉矶分校毕业证制作成绩单修改
 
Customer Service Analytics - Make Sense of All Your Data.pptx
Customer Service Analytics - Make Sense of All Your Data.pptxCustomer Service Analytics - Make Sense of All Your Data.pptx
Customer Service Analytics - Make Sense of All Your Data.pptx
 
Industrialised data - the key to AI success.pdf
Industrialised data - the key to AI success.pdfIndustrialised data - the key to AI success.pdf
Industrialised data - the key to AI success.pdf
 
办理(Vancouver毕业证书)加拿大温哥华岛大学毕业证成绩单原版一比一
办理(Vancouver毕业证书)加拿大温哥华岛大学毕业证成绩单原版一比一办理(Vancouver毕业证书)加拿大温哥华岛大学毕业证成绩单原版一比一
办理(Vancouver毕业证书)加拿大温哥华岛大学毕业证成绩单原版一比一
 
Building on a FAIRly Strong Foundation to Connect Academic Research to Transl...
Building on a FAIRly Strong Foundation to Connect Academic Research to Transl...Building on a FAIRly Strong Foundation to Connect Academic Research to Transl...
Building on a FAIRly Strong Foundation to Connect Academic Research to Transl...
 
Amazon TQM (2) Amazon TQM (2)Amazon TQM (2).pptx
Amazon TQM (2) Amazon TQM (2)Amazon TQM (2).pptxAmazon TQM (2) Amazon TQM (2)Amazon TQM (2).pptx
Amazon TQM (2) Amazon TQM (2)Amazon TQM (2).pptx
 

An approach to offer management: maximizing sales with fare products and ancillaries

  • 1. PRACTICE ARTICLE An approach to offer management: maximizing sales with fare products and ancillaries Ben Vinod1 • Richard Ratliff1 • Vikram Jayaram1 Revised: 1 March 2017 Ó Macmillan Publishers Ltd 2018 Abstract With the growth in ancillary sales, an area of increasing importance for airlines is the concept of offer management, which entails the creation of dynamic, cus- tom, personalized offers consisting of a flight itinerary and ancillary products offered by an airline. This practice-ori- ented, overview paper provides an end-to-end, future-ori- ented framework for determining the composition of optimal base fare and ancillary bundles by customer trip- purpose segment followed by 1:1 personalization to max- imize total sales. Our focus in this paper is primarily on the proposed offer management framework and its sub- components. Keywords Revenue management Á Ancillary products Á Willingness-to-pay Á Bundles Á Customer segmentation Á Personalization Á Controlled experiments Overview Airline offer management is the practice of creating dynamic, custom, personalized offers consisting of a flight itinerary and ancillary products sold by an airline. An important precept of offer management which enables maximizing ancillary revenue is to maintain the identity of each discrete ancillary in the bundle. The objective of offer management is to offer the right bundles to the right cus- tomer at the right price at the right time. Offer management is an extension of airline revenue management (RM). Given the growing importance of air- line ancillary sales over the past decade and because the type and volume of ancillary sales vary by fare product type considered, many airlines have started incorporating ancillary revenue streams into their RM systems. This process of including the sale of ancillaries to the base fare in the traditional revenue management context has been referred to as ‘total revenue management’ (Rickey 2014). Surveys (Alexander 2006) have shown that travelers would pay for extra perks, such as more frequent flyer miles, more overhead bin space, and the choice to sit in a child-free section of the aircraft. Ancillary revenues and new product offerings in the form of airline branded fares have grown rapidly over the past decade, and new decision support tools (including both offer management and total RM) are emerging in this area. During the early days of promoting ancillaries, ultra-low cost carriers (LCC’s) such as Spirit Airlines, Ryanair and easyJet gained notoriety due to their business models that were based on maximizing total revenue. This total includes revenues generated from fares and add-ons such as fees for carry-on bags, checked luggage and seat assign- ments. For these LCC’s, ancillary revenues can comprise [11% of their total revenue, with some airlines reaching as high as 25% (IdeaWorks 2016). The full service carriers on the other hand were slow to adopt add-ons, but that’s changing. In 2015, single-year growth of ancillary revenues for full service carriers was greater than 13% (IdeaWorks 2016) comprised primarily of baggage fees, food and beverages, and premium seat assignments. The sale of miles or points to banks for co-branded credit card loyalty programs was another large component. The sale of ancillary services started in the airline direct channel and has become important for the indirect channel & Ben Vinod Ben.Vinod@sabre.com 1 Sabre Research, 3150 Sabre Drive, Southlake, TX 76092, USA J Revenue Pricing Manag https://doi.org/10.1057/s41272-017-0121-1
  • 2. [i.e. Global Distribution Systems (GDS) and Online Travel Agencies (OTA)] as well. Annual growth of ancillary sales through Sabre channels worldwide was greater than 120% during June 2015 compared to June 2016, a representative period of peak summer travel, as shown in Fig. 1. Several airlines worldwide have signed up to distribute ancillary services through the Sabre channel. Traditional ancillary sales pioneered by Spirit Airlines in the 1990s focused ancillary sales efforts after the booking was made (via e-mail in the later years). Besides Spirit, other airlines that led the pack include Aer Lingus, AirAsia, easyJet, Ryanair and Vueling. An alternative to post-booking ancillary sales is the at- booking approach; i.e. promoting bundles during the shopping and booking process itself. The at-booking approach is the main method for travel agents to sell ancillaries, and it is also widely used in airline-direct channels. This conclusion was based on an analysis of EMD (Electronic Miscellaneous Document) sales in Sabre over a 1-year period. EMD’s are an International Air Transport Association (IATA) standard for electronically tracking airline ancillary sales and revenue; they come in two types: EMD-A (air): associated with an air e-ticket EMD-S (standalone): sold on a standalone basis Table 1 shows a percentage breakdown of the types of EMD’s issued in a large data sample comprised of both travel agency and airline sales. At-booking bundles will always have a lower sales conversion rate than standalone air sales (due to the increased purchase complexity and total price), but travel agency capabilities to sell ancillaries are improving and include many advancements over the past decade. For example, specific functionality such as Midwest Airlines pay-for-seats (‘‘Signature Seats’’) was introduced in Sabre in 2007. Branded Fares availability, shopping, fare quote and booking capabilities have been in place since 2010. Two other important agency ancillary sales enablers are more recent. In 2011, EMD functionality was added so travel agents can shop, book and sell airline ancillaries (either at or post-booking). Although we do not yet have specific information to compare the effectiveness of at- booking versus post-booking ancillary sales efforts, it is an interesting topic for future research. The promotion of service differentiation with branded fare products and sale of ancillaries are collectively refer- red to as ‘‘merchandising’’ or ‘‘airline retailing. Branded fare families were first introduced by Air New Zealand in 2004 followed by Air Canada (Vinod 2008). Many airlines followed and initially adopted branded fare products with pre-defined bundles. Some airlines, led by US majors and regional carriers, promoted unbundled pricing with a la carte sale of ancillaries. However, today a large number of airlines have evolved to a hybrid model. In this hybrid model, branded fare products are retained and include support for standalone sales of additional ancillaries that are not included in the branded fare product. Other varia- tions also exist; some airlines offer bundles of ancillaries (e.g. Delta Air Lines’ ‘‘Lift Package’’ provides price sav- ings for priority boarding when combined with a 1000 mile frequent flyer booster). Figure 2 shows the evolving trend in merchandising and retailing. For example, one of the early adopters, Air Canada1 originally had three branded products in economy (Tango, Tango Plus, Latitude) and two branded products in the premium cabin (Executive Class Lowest, Executive Class Flexible). Each branded fare product is a combination of an airfare with bundled attributes, such as access to the Maple Leaf lounge, priority baggage check in, fare refundability, Fig. 1 Ancillary sales growth in Sabre channels 1 Air Canada currently has the Air Canada Altitude Program https:// altitude.aircanada.com/status/home. B. Vinod et al.
  • 3. advance seat selection, frequent flyer miles accrual, and change fees. In addition, even with the purchase of a lower valued branded fare product, customers can pay for specific ancillary services such as advance seat selection. The dis- tribution infrastructure of suppliers and the GDSs is maturing to simplify branded fare filing and settlement. The Airline Tariff Publishing Company (ATPCo) in con- junction with the International Air Transport Association (IATA) has provided a service fee solution. The fee types are OA (booking fees), OB (ticketing fees) and OC (op- tional service fees). Airlines that fail to adopt an a` la carte pricing model will likely lose customers and potential revenue (Nason 2009). In addition, the S8 record from ATPCo that links the branded fare products to ancillaries is being adopted by a large number of airlines such as Delta Air Lines, Virgin Australia, and Kulula. Significance of offer management According to IdeaWorks (2016), airline ancillary revenue statistics compiled from Air Transport World, Airline Business and airline websites totaled $59.2 billion world- wide in 2015 (projected to grow to $67.4 billion in 2016). With this growth in airline ancillary sales, offer manage- ment is increasingly important; it consists of creation of dynamic custom personalized offers based on customer traits, customer value score, flight and schedule attributes, ancillaries and non-air products through the airline’s pre- ferred channels of distribution. The initial focus has been to sell ancillaries and sell-up of branded fares through the direct and agency channels for online and codeshare part- ner flights (Smith et al. 2006; Vinod and Moore 2009; Zouaoui and Rao 2009). The initial focus of airlines in the area of merchandising and ancillaries has been on execution; these execution- oriented systems provide the ability to sell ancillaries and sell-up of branded fares through the direct channel and agency channels for online and codeshare partner flights (Vinod 2015). An offer management platform should include decision support capabilities as well complement real-time execution systems for generating targeted offers to customers that maximize revenues. The new distribution capability proposed by IATA is a travel industry-supported program for the development and adoption of a new data transmission standard for airlines to communicate with retails sales outlets such as travel agents, corporations and travelers. Towards this end, application program interface (API) standards have been defined for various functions as shopping, booking, ancil- laries and fulfillment. The offer management solution for both the direct and indirect channel are NDC compliant in the send that they would use the same NDC API’s as a means to communicated between the airline and the retail outlet. Offer management is important for IATA’s New Distribution Capability (NDC), and it also extends across all channels of distribution. This paper provides a framework for offer management decision support including the process of determining the composition of the optimal bundles for customer segments and their respective price points (Fig. 2). A bundle consists of both a base fare and targeted ancillaries with a total price Table 1 Comparison of ancillary sales by EMD type for airline and agency channels Type Airlines (%) Travel agencies (%) EMD-A (air e-ticket) 92.3 99.1 EMD-S (standalone) 7.7 0.9 Fig. 2 Trends in airline price bundling An approach to offer management: maximizing sales with fare products and ancillaries
  • 4. that is promoted by an airline to a customer segment or an individual passenger. The remainder of this paper involves the non-traditional revenue management components as shown in Fig. 3 and is organized as follows (the direction of red arrows in Fig. 3 indicate the step by step flow in the proposed offer man- agement framework). Trip-purpose segmentation The main requirement in making effective bundled offers to customers is to understand their wants and needs for their upcoming trip. Determining the trip purpose (e.g. business or leisure, duration, origin and destination, etc.) has a strong bearing on the customer’s ancillaries pur- chased and price sensitivity. So the need for trip-based segmentation for bundling of offers cannot be understated. Much of the marketing literature describes customer profiles based on historical purchases and declared pref- erences by specific customers (e.g. using their name, credit card, phone number or frequent flyer ID); although this information is important, it may not be applicable for future flights. This is because the typical traveler has multiple profiles depending on the purpose of the trip (e.g. business, traveling with family, visiting friends and relatives, etc.). Also in most travel purchase situations (e.g. booking via a travel agent or a website without cookies enabled), shopping for flights is done on an anonymous basis. In the authors’ experience, we find trip-based segmen- tation to be the most practical initial method of classifying customer types; it can be augmented with customer- specific information in a later step when the actual cus- tomer is declared such as when the flight purchase is completed and full customer details are obtained. Classi- fying customers can be based on trip characteristics such as: • origin and destination, • how far in advance they book, • distribution channel (e.g. airline direct or via a travel agency), • number in party, • children or infants, • length of stay, • departure and return day-of-week, • Saturday night stay-away classification, • season, • holiday versus non-holiday period, • and presence or absence of special qualifiers on the low-fare search, etc. Fig. 3 Offer management framework B. Vinod et al.
  • 5. These are all examples of variables that can be used to help ascertain the trip purpose and classify customers into variants of leisure and business types. Table 2 illustrates an example of trip-purpose segments for air. This particular rule-based, ten cluster classification scheme was based on 1 year of a multi-agency and multi-airline Sabre ticketing data sample (including both domestic and international markets), but it is important to note that best-performing rules would vary depending on the particular airlines, agencies and networks considered. As a pragmatic matter, our definitions of these segments are mutually exclusive (based on fixed classification rules); although that can potentially worsen the clustering performance, it does make it easier to apply the algorithm in practice. Although, we used data analysis in finding good breakpoints for each dimension, we rounded them to nearby values which cor- respond more closely with standard airline pricing restric- tions to make them more intuitive to airline users. We were concerned that using fixed, rules-based clas- sification schemes could lead to unacceptably high volatility among items mapped within each cluster when compared to more standard clustering methods. As a quick validation test, for each data point in our sample, we computed its Euclidean distance to the centroid of the cluster it was assigned to. Euclidean distance is the square root of the sum of the squared differences across the dif- ferent dimensions considered; in our case, the dimensions used were all numeric (i.e. advance purchase period, length of stay and number of passengers). We then summed the Euclidean distances across all the data in our fixed classi- fication scheme and compared those totals to two other clustering approaches: Ward-based hierarchical clustering with nine clusters (the tenth one was removed because it only had a few observations mapped to it) and a random clustering approach (purely to provide another baseline reference). Ranking the approaches from least to greatest total Euclidean distance we found: 1. Fixed classification (smallest total difference). 2. Hierarchical clustering. 3. Random assignment (greatest total difference). Although there were many other validation tests we could have performed, we concluded from the Euclidean distance test that the rule-based assignments were not unreasonable in terms of within-cluster minimization performance when compared with a commonly-used automated approach. In actual practice, our plan is to make additional dis- tinctions. These include short haul (800 miles) versus long haul (C800 miles) markets since purchase of ancil- laries is influenced by length of haul. If Saturday Night Stay is added, it is a mechanism to address day of week by saying that the segment either includes or excluded Saturday Night Stay. Selling channel is another obvious dimension to consider (direct vs. indirect). Hence, if short vs. long-haul, Saturday Night Stay and channel dimensions are added, then the number of proposed air trip-purpose segments would increase. Market basket analysis Ancillaries promoted by airlines fall into two broad cate- gories: dominant ancillaries and non-dominant ancillaries. Examples of dominant ancillaries are baggage, pre-re- served seats, premium seats (extra leg room, aisle, window, etc.). Non-dominant ancillaries are the extra features that are nice to have but are usually not a necessity (e.g. meals in coach, lounge access, newspaper, etc.). Due to differ- ences in elapsed flight time, seating configuration, etc., each flight leg can have a unique relationship between the dominant and non-dominant attributes. Figure 4 shows an example of ancillary type associa- tions in two different markets where a pre-reserved seat was a dominant ancillary and the other non-dominant ancillaries were co-purchased in each case. The number of connection lines between the dominant and the non-dom- inant ancillary in the two types indicate the strength of the association or higher number of co-purchases. Notice in this example that, even though the pre-reserved seat is the dominant ancillary in both markets, the non-dominant ancillaries vary in each case; this situation is typical in practice. In Fig. 4, the acronym FFP stands for frequent flying programs. Table 2 Air trip purpose segment definitions Air segment Id Type Business/ leisure Advance purchase Length of stay A1 Individual Business 0–6 days 0–1 days A2 Individual Business/leisure 0–6 days 2? days A3 Individual Business/leisure 7–13 days 0–3 days A4 Individual Leisure 7–13 days 4? days A5 Individual Business/leisure 14–20 days Any A6 Individual Business 21? days 0–3 days A7 Individual Leisure 21? days 4? days A8 Couple (2) Leisure 0–20 days Any A9 Couple (2) Leisure 21? days Any A10 Family ([2) Leisure Any Any An approach to offer management: maximizing sales with fare products and ancillaries
  • 6. Willingness-to-pay An active area of research is to quantify a customer’s willingness to pay for ancillary services to determine their value (Ratliff and Gallego 2013). Fortunately, during the past few years, EMD data have become available, thus allowing a consumer’s willingness to pay to be calibrated for dynamic pricing of ancillaries. Established methods include multinomial logit choice analysis (Ben-Akiva and Lerman 1985; Train 2003; Bal- combe et al. 2009), the van Westendorp pricing model (van Westendorp 1976; Hague 2008) and conjoint analysis (Green et al. 2001; Hair et al. 1984) which considers tradeoffs between various combinations of price and pro- duct features. A stated preference model (Martin et al. 2008) can be used to provide empirical evidence of esti- mated valuations air passengers have on quality of service attributes such as comfort, food, ticket change fees, fre- quency and reliability. Survey tools can be used to estimate a customer’s willingness to pay based on stated prefer- ences. This approach is closer to a customer’s shopping experience, and hence it is an improvement over single dimensional surveys such as the van Westendorp (1976) pricing model. The outputs of the model are customer willing to pay for ancillary products and the relative importance of each. This information is important in cor- rectly setting prices for ancillaries and branded fare bun- dles. Experimental design (described in ‘‘Controlled experimentation’’ of this paper) is another emerging area for willingness-to-pay estimation which is becoming more mature (Besbes and Zeevi 2015). An alternative approach which can be used to indirectly estimate ancillary willingness to pay is based on the (seemingly mild) assumption that ancillary sales have a similar price elasticity to air travel in general. A simple heuristic approach for estimating air travel demand price elasticity using commonly available airline revenue man- agement history data can be found in (Ratliff and Vinod 2016); this approach uses a cumulative sales history across differently priced fare products together with log–log regression methods to model the demand curve. For situ- ations where relatively poor historical data are available for ancillary willingness-to-pay estimation and/or direct experimentation of different ancillary prices is impractical, then assuming that ancillary price elasticity is similar to air travel in general seems to be a reasonable alternative. Competitive positioning Perhaps the most important (but often ignored), component of an effective retailing system is competitive positioning. The benefits of effective customer retailing and accurate availability information are well understood by the airlines, but ongoing monitoring poses challenges owing to the data required. In the context of offer management, such com- petitive information includes both airfares as well as ancillary prices. In this section, we will begin with airfare competitiveness (for which some automated data sources exist) followed by ancillaries (for which commercially available comparison data are lacking). Applications which help airlines to automatically respond to changing competitive circumstances for airfares are emerging; one recent reported example is the market adaptive pricing (MAP) module (Ratliff 2016) which is a component of an airfare dynamic pricing engine. Tradi- tionally, most airlines have used the Airline Tariff Pub- lishing Company (ATPCo) to file fares and compare prices. Fig. 4 An example of dominant ancillary with its non-dominant association B. Vinod et al.
  • 7. While very useful for many purposes, unfortunately these filed fare data do not consider availability. MAP processing relies on real-time (or recent) low-fare search results (which do consider availability) by market, point of sale (POS) and date to understand an airline’s current compet- itive positioning (considering both price and schedule quality aspects). Customer choice and optimization models are used which consider price, schedule quality and cus- tomer segmentation to estimate probability of selection and expected revenue of available itineraries. MAP provides a real-time, ‘sense and respond’ capability for maximizing expected itinerary profits (i.e. expected revenue net of estimated flight opportunity costs); see Ratliff and Vinod (2005) and Cary (2004). Understanding your airline’s current air availability and price competitiveness is at the core of market adaptive pricing. The low-fare search results by market, date and POS from GDS and third-party robotic vendors are a good source for this type of information because they provide results across multiple airlines. Ideally, these shopping data would be comprised of streaming, real-time information. However, given the large number of items (i.e. markets, departure and return date combinations) that most airlines want to monitor (often [1 million), cached shopping results are typically used instead (Ratliff 2016). Real-time, MAP optimization requires accurate estimates of marginal cost (i.e. flight bid prices by itinerary); see Gallego and Hu (2014) and Choubert et al. (2015). The potential benefits of MAP are large; Choubert reported simulation results indi- cating net revenue improvements in the 4–7% range. Accurate product availability information also affects the execution of an airline’s retailing strategy (Ratliff et al. 2003). Incorrect availability can lead to the wrong (or in some cases no) product might be offered to the customer, or inconsistencies can result between the products dis- played during the shopping process versus those which are actually bookable. Inasmuch as airfare availability and shopping data pose data collection challenges, at least there exist GDS (global distribution system) and web-based tools for such infor- mation; in fact, there are even some commercial vendors providing shopping data solutions (e.g. Infare and QL2). Unfortunately vendor solutions for gathering and compar- ing competitor ancillary information are still immature, and most airlines rely on manually conducting such competi- tive comparisons. At present the best automated source of ancillary price information are the ATPCo optional ser- vices records: • S5—sub code services record (detailed service defini- tions to allow cross-airline comparisons). • S6—concurrence record (for flight-related or rule buster services to ascertain partner airlines’ concurrence). • S7—provision record (to identify the travel, passenger, geographic, flight, fare and sales requirements for each specific sub code service in the S5). Carriers often ask ‘‘how important is it to be competitive with respect to specific ancillaries?’’ In our opinion, total price (i.e. airfare ? ancillary purchases) is the main con- sideration by customers. Because of the difficulties in gathering current airfare and ancillary information, direct comparisons at the detailed level are not easy. Fortunately, although there are hundreds of different possible types of ancillaries, the majority of the sales are comprised of only a few major items. By knowing which ancillary items are the most important ones by market, a carrier can simplify its competitive comparisons by focusing its monitoring efforts on airfares and top-selling ancillaries. Dynamic pricing of bundled offer Most airlines today have ancillary prices that are the same (or nearly so) for all their markets in a given region (e.g. $25 baggage fees on all domestic flights). In the future we expect a broader trend of airlines experimenting with market and date-specific ancillary prices, possibly even further varying prices in other contexts such as time to departure or scarcity of the ancillary products (e.g. extra legroom seats). Ancillary pricing at these more detailed levels will necessitate expanded data collection efforts and further complicate an airline’s understanding of its com- petitive position. This variable ancillary pricing by market for services such as pre-reserved seats based on competitive market conditions and distance is frequently referred to as dynamic pricing of ancillaries (Ødegaard and Wilson 2016). In recent years there has been considerable research work in the area of dynamic pricing for airfares (Choubert et al. 2015), and an ATPCo working group was formed in 2016 to create common input/output specifications for dynamic pricing engines for air travel. Our expectation is that sim- ilar capabilities will emerge in regard to ancillary dynamic pricing in the near future. Dynamic pricing of the bundle is based on a consumer’s willingness to pay, which is calibrated by trip-purpose segment (previously described in ‘‘Trip-purpose segmen- tation’’). Figure 5 below illustrates how ancillary prefer- ence rankings vary by trip-purpose segment for a large, multi-airline data sample we tested. An approach to offer management: maximizing sales with fare products and ancillaries
  • 8. Personalization Today’s customers have high expectations for personalized services. Successful customer experiences promote rev- enue growth and shall influence the future product invest- ments from suppliers, intermediaries and software vendors to ensure the loyalty and retention of profitable customers. While trip-segmentation itself can be used to determine the composition of the initial bundle to be offered to cus- tomers, the composition of the bundle can be augmented and refined once the customer’s identity is known. Past travel history of the customer, preferences, loyalty program affiliation and a customer’s preferences for the specific trip all play a role in fine tuning and presenting the final trav- eler preferred offer. The trade-off analytics algorithm is a technique for ordering preferences by similarity to the ideal solution. It ranks the bundles based on a traveler’s relative trade-offs between the various attributes of the ancillaries and the budget constraint. The algorithm scores all bundles based on the trade-off evaluation based on the relative importance of the ancillaries and displays the bundles in the order of the scores. The resulting improvements to the composition of the customized bundled offer and display order should help improve sales conversion rates. Branded Fares is also an important consideration in creating the final offer. When a bundled offer is deter- mined, it should be mapped to the attributes included in each branded fare product to ensure the lowest cost option is offered to the customer. Hence for example it may be economical to offer a customer a cheaper branded fare product and an ancillary priced separately than offering the higher valued branded fare product with the ancillary included. Hence offer management provides a range of upsell opportunities tied to the various branded fare prod- ucts that are available for sale, and the recommendation can be based on the least total cost to the customer. Controlled experimentation Controlled experimentation is often useful either as an alternative to model-based tools or when used in con- junction with model-based tools as a validation and fine- tuning mechanism. In recent years, direct experimentation applications have become more widespread (e.g. Airbnb, Amazon, Expedia, Facebook, Google, Microsoft, Staples, and Walmart). Business experimentation provides a direct approach to running controlled tests to learn customer behavior. Unlike model-based decision support tools, business experimen- tation is in many respects simpler to use and implement. Models range from simple to sophisticated. However, all of them are only as good as the quality of the underlying input estimates (e.g. demand forecasts, operational schedule, weather patterns, guest checkout patterns, etc.). Big data can only provide clues about the past behavior of cus- tomers (i.e. out-of-sample predictions are questionable). Also, there are often situations in which no historical data exists (e.g. for new markets or products), and business experimentation may be the only viable approach in the short term. The idea of business experimentation is to directly test new prices, product designs, systems or processes com- pared to current ones to see if there is performance Fig. 5 An example of ancillary preferences ranked by revenue in each trip segment B. Vinod et al.
  • 9. improvement. Experimentation can help to answer com- mon business questions such as: • What is the best price point for maximizing revenue? • Which particular algorithm or business approach works best? • How are my revenues and conversion rates impacted by changes to my display rules? • Which bundled offer maximizes revenue? Thus, even with reasonably advanced models, estimates of trip purpose, willingness-to-pay, ancillary preference order, etc. are still imperfect. Direct, controlled experiments are one way to estimate willingness-to-pay for ancillaries and bundles. Such controlled experiments involve randomly allocating different experiments (e.g. prices) across various test subjects (which can be organized by session, departure dates, markets, agencies, etc.). Randomization reduces bias by equalizing other features that have not been explicitly accounted for in the experimental design. Simple binary tests (e.g. old vs. new) are commonly referred to as A/B testing. One of the more advanced (and widely used) experimental methods today is the multi- armed bandit (MAB) algorithm, and there are a number of good reasons it is used in our proposed offer management framework. MAB implementations are more general, are adaptive (i.e. learns over time), and can support running multiple experiments simultaneously (e.g. A/B/C/D, etc.). We chose the multi-armed bandit methodology because it is statistically sound and open-source software is available. The multi-armed bandit method continually learns as experiments are conducted; the proportion of experiments being run can be changed dynamically depending on cur- rent results, so instead of fixed proportions, as test results are obtained, more weight is given to those experiments which show better performance (and vice-versa). Advo- cates of MAB describe it as balancing ‘experimentation’ (e.g. testing different prices) with ‘exploitation’ (by dynamically increasing the proportion of those experiments showing better performance). An example of this dynamic reweighting of the experiments over time is shown in Fig. 6. There are various heuristics used to compute the dynamic reweighting proportions. The authors recommend Bayesian approaches which use randomized probability matching (e.g. Thompson Sampling). The underlying principle behind these methods is to randomly allocate each experiment according to its probability of being optimal (Scott 2010). Fig. 6 Multi-armed bandit dynamic reweighting example Fig. 7 End-to-end offer management business process An approach to offer management: maximizing sales with fare products and ancillaries
  • 10. The end-to-end view: putting it all together Synthesizing all the various components described in ear- lier sections of this paper, we provide an illustration in Fig. 7 which shows the end-to-end, at-booking, offer management process. Notice that offer management is an extension of existing systems; a key input in creating the bundled offer price for seats and ancillaries is the tradi- tional revenue management control mechanism (e.g. leg/ segment, O&D or dynamic pricing). The recommendation engine and offer engine rely on historical booking data as the source of data to recommend relevant content and create the customer specific bundle respectively. The rec- ommendation engine determines the relevant content that is applicable to a segment based on historical behavior. The offer engine takes the results from the recommendation engine together with user specific preferences to generate the 1-to-1 bundle based on a trade-off analytics model which ranks bundles under consideration based on a trav- eler’s relative trade-off between the various bundles sub- ject to a budget constraint. The model scores all bundles based on the trade-off evaluation using relative weights which imply importance of the specific ancillary as part of the bundle and displays the bundles in order of the scores. As the figure indicates, the workflow begins with a user executing a shopping request, selects an itinerary which is followed by a ranked list of default bundles which are by trip purpose segment. The user proceeds to select a default bundle and modify the composition of the bundle- ancil- laries and relative weights based on importance of the preference before the trade-off analytics model recom- mends a bundle specific to the traveler. The offer management process for the post-booking model is very similar, with the exception that the workflow begins after the air booking has been made. Conclusions Decision support for offer management requires an understanding of customer preferences and generating pertinent targeted responses to customer requests during the sales process life cycle. Generating a multitude of offers to customers who have to select from a myriad of options slows down the sales process and leads to aban- donment. A key to customer centricity is the ability to offer the right product or bundle to the right customer at the right price at the right time based on derived customer prefer- ences. Displaying targeted offers that resonate with cus- tomers requires an investment in a data infrastructure and advanced analytics to understand consumer behavior and preferences. This approach in turn generates incremental revenues with targeted offers and ensures repeatable, profitable customers. Development of new business pro- cesses and management tools to support total revenue management will continue to be an important growth area. References Alexander, K.L. 2006. Paying more for small extras. Washington Post, January 31. http://www.washingtonpost.com/wp-dyn/con tent/article/2006/01/30/AR2006013001282.html. Balcombe, K., I. Fraser, and L. Harris. 2009. Consumer willingness to pay for in-flight service and comfort levels: A choice experi- ment. Journal of Air Transport Management 15: 221–226. Ben-Akiva, M., and S. Lerman. 1985. Discrete choice analysis: Theory and application to travel demand. Cambridge, MA: MIT Press. Besbes, O., and A. Zeevi. 2015. On the (surprising) sufficiency of linear models for dynamic pricing with demand learning. Management Science 61 (4): 723–739. Cary, D. 2004. A view from the inside. Journal of Pricing and Revenue Management 3 (2): 200–203. Choubert, L., T. Fiig, and V. Viale. 2015. ‘‘Amadeus Dynamic Pricing’’, presentation at the AGIFORS Revenue Management and Distribution Study Group meeting, Shanghai, China. Gallego, G., and M. Hu. 2014. Dynamic pricing of perishable assets under competition. Management Science 60 (5): 1241–1259. Green, P.E., A.M. Krieger, and Y. Wind. 2001. Thirty years of conjoint analysis: Reflections and prospects. Interfaces 31: S56– S73. Hague, N. 2008. The problem with price. B2B International. http:// www.b2binternational.com/library/whitepapers/pdf/the_pro blem_with_price.pdf. Hair, J.S., R.E. Anderson, and R.T. Tatham. 1984. Multivariate data analysis with readings, 2nd ed. New York: Macmillan. IdeaWorks. 2016. Airline ancillary revenue projected to be $67.4 billion worldwide in 2016. Press release, 29 Nov. 29, 2016. No. 115. http://www.ideaworkscompany.com/wp-content/uploads/ 2016/11/Press-Release-115-Global-Estimate.pdf. Martin, J.C., C. Roman, and R. Espino. 2008. Willingness to pay for airline service quality. Transportation Reviews 28 (2): 199–217. Nason, S.D. 2009. The future of a la carte pricing in the airline industry. Journal of Revenue and Pricing Management 8 (5): 467–468. Ødegaard, F., and J.G. Wilson. 2016. Dynamic pricing of primary products and ancillary services. European Journal of Opera- tional Research 251 (2): 586–599. Ratliff, R.M., A. Walker, B.C. Smith, and T. Brice. 2003. Availability based value creation method and systems, U.S. Patent 20030191725, filed March 26, 2003, and issued October 9, 2003. Ratliff, R.M. 2016. A business overview of market adaptive dynamic pricing. Sabre white paper. Ratliff, R., and G. Gallego. 2013. Estimating sales and profitability impacts of airline branded-fares product design and pricing decisions using customer choice models. Journal of Revenue and Pricing Management 12 (6): 509–523. Ratliff, R.M., and B. Vinod. 2005. Airline pricing and revenue management: A future outlook. Journal of Revenue and Pricing Management 4 (3): 302–307. Ratliff, R.M., and B. Vinod. 2016. An applied process for airline strategic fare optimization. Journal of Revenue and Pricing Management 15 (5): 320–333. Rickey, D. 2014. Total revenue management. Ascend 13 (4): 20–22. B. Vinod et al.
  • 11. Scott, S. 2010. A modern Bayesian look at the multi-armed bandit. Applied Stochastic Models in Business and Industry 26: 639–658. Smith, B.C., R. Darrow, J. Elieson, D. Guenther, B.V. Rao, and F. Zouaoui. 2006. Travelocity becomes a travel retailer. Interfaces 37 (1): 68–81. Train, K.E. 2003. Discrete choice methods with simulation. New York: Cambridge University Press. van Westendorp, P.H. 1976. NSS—price sensitivity meter (PSM)—a new approach to study consumer perception of price. Proceed- ings of the ESOMAR Congress, Venice. Vinod, B. 2008. The continuing evolution: Customer centric revenue management. Journal of Revenue and Pricing Management 7 (1): 27–39. Vinod, B. 2015. The expanding role of revenue management in the airline industry. Journal of Revenue and Pricing Management 14 (6): 391–399. Vinod, B., and K. Moore. 2009. Promoting branded fare families and ancillary services: Merchandising and its impacts on the travel value chain. Journal of Revenue and Pricing Management 8 (2–3): 174–186. Zouaoui, F., and B.V. Rao. 2009. Dynamic pricing of opaque airline tickets. Journal of Revenue and Pricing Management 8 (2–3): 148–154. An approach to offer management: maximizing sales with fare products and ancillaries