Revisiting Customer Centricity For Better Data Analysis
Airlines are ready to adopt new customer-centric practices — supported by efficient data management and enabled by operations research and analytics.
In the airline industry — which quite appropriately emphasizes customer service as a hallmark of the successful carrier — a customer-centric approach always makes good economic and business sense.
Security and government regulations have effectively provided airlines a potentially key advantage, which is represented by the expansive volume and depth of information that has been gathered about customers.
Over the years, airlines have collected and stored large amounts of detailed data describing the demographic, social, economic and behavioral characteristics of their customers.
But because of the complexity of data management and the challenges created by the sheer immensity of available data, the value of that data has not even yet been completely unlocked.
Therefore, airlines need records and decision-support systems that focus on customer centricity. Advanced reservations, profitability, evaluation and revenue management systems must use customer-choice modeling in designing optimal policies for airlines’ planning and operational processes.
Over the past several years, Sabre Airline Solutions® has invested quite significantly in these types of advanced systems, as well as in data-management technology.
For example, platforms for real-time PNRs management, travel-shopping data and aggregated customer information allow relatively easy access to a variety of data sources. They also enable the next steps in automation and optimization of business processes currently used by airlines, as well as creation of new opportunities for carriers to generate revenue and better serve their customers.
To create a coherent 360-degree view of customer data available to airlines, Sabre Airline Solutions has introduced the concept of a trip lifecycle, which consists of four main stages:
Customer Data Collection
Trip lifecycle includes all stages travelers go through from planning all the way through the in-flight experience. Airlines can align their services with customers’ expectations by enhancing the decision-making process with customer data collected throughout all trip stages.
- Planning — Customers at this stage haven’t yet begun looking for travel options, and, therefore, they haven’t revealed any of their attributes. An airline can identify segmentation of historical bookings and prompt customers to choose a particular trip by sending relevant suggestions based on their travel history, or by advertising characteristics of an offer that is particularly attractive for these specific customers. If, for example, a carrier determines strong economic or ethnic community ties between two regions, it can develop a schedule that ensures favorable connection opportunities.
- Shopping — Customers have now submitted requests and revealed a few simple attributes such as origin, destination and duration of a trip, as well as the channel the customers book through, the schedule or price sensitivity, etc. Based on these attributes, an airline can allocate a request to a particular segment, and guide customers toward the most attractive offers for that segment. These offers can be related to the primary trip choice (or, in case that trip is not available, to a secondary trip choice) and to ancillary services that match recommendations for that segment. Customers might be offered an alternative destination within the same trip theme, or an opportunity to enhance their travel experience with an upgrade or extended stay at a leisure destination.
- Pre-flight — Customers have already booked a trip and have, therefore, revealed many more attributes. They also have well-defined travel plans. Placing these customers into well-defined segments helps airlines cross sell ancillary products and services. A family flying to Alaska for a 10-day trip, for example, is likely to respond to a car rental offer, while a couple visiting New York on a weekend might be interested in theater tickets.
- Post-departure — Customers have checked in and started their trip, or have already completed it. These customers can be segmented not only on their trip attributes, but also on their experience. Segment information can be used to better serve them during the trip, or to manage their feedback afterward. Gate agents can be given access to this information so they can be more efficient in handling disruption situations. Connecting passengers can be offered services at an airport, ranging from assistance in shopping to providing expedited transfers to ensure that a tight connection is not missed. Passengers who experienced a disruption or baggage mishandling might be approached afterward with compensation options so they will be more likely to remain loyal to the airline in the future.
Travelers — during the course of their entire trips — are going through the stages described, with the post-departure stage eventually turning back into the planning stage, during which customers reapply their travel experience (recent and otherwise) in preparing for their next trip, thus completing the full 360-degree circle of the trip lifecycle.
Interaction between an airline and a customer at any one of the stages should not be limited to the data related directly to that stage only.
By supplementing and supporting the decision-making process with customer data collected throughout all trip stages, airlines will be able to better match their services with travelers’ expectations, and thereby be more efficient and profitable.
Customer Segmentation And The Cluster Approach
Customer segmentation is an important concept in any service industry because it allows the identification of clusters of customers who are interested in similar levels of service — so additional targeted offers can be created.
For years, airlines have used artificial rules such as advance-purchase restrictions, refundability, booking classes, etc., to create these clusters.
In a new approach, airlines can leverage numerous data sources and create flexible methodology combining those sources to bring to life the most complete picture of an airline’s customer base.
The full range of data sources might include PNR databases, loyalty-program information, corporate-sales records, sales and promotions history, ancillary-services records, social-network statistics and other sources.
Once data points are aggregated — and the data’s consistency is verified — the airline can apply various clustering algorithms implemented in a segmentation analyzer to optimally partition all customers into clusters so elements of each cluster are as similar to each other as possible, while the clusters themselves are separated.
A segmentation analyzer provides tracking as well as analyzing capabilities.
Characteristics of each cluster, such as the revenue it provides, the typical advance-purchasing behavior of its population, social activity, trip length, etc., can be monitored and compared across different markets or times.
A segment-transition matrix, for example, can provide visibility into changes in segment mix over time, and can be used to evaluate an impact of promotion campaigns, introductions of a new service and other customer-centric business practices.
It is important to differentiate between trip segmentation that identifies the PNR phenotypes an airline serves, as opposed to consumer segmentation that works on what is commonly labeled a “customer DNA” level.
The fundamental difference between the PNR phenotype and customer DNA is in the respective level of granulation.
While the PNR phenotype primarily identifies a purpose and attributes of an “anonymous” trip, customer DNA is directly associated with a specific traveler and includes the history of all trips the customer has taken.
A business traveler making a day trip or a family with children traveling for a two-week summer vacation, for example, are PNR phenotypes — while Raj, who has made multiple U.S. domestic trips during the past year using his student discount, has a customer DNA.
These segmentation types can then be used in both tactical applications that support real-time interaction with an individual customer and strategic planning tools focusing on processes and practices relevant for a large part of an airline network, customer community or long operational period.
And these applications can be used in product design, promotion management, revenue analytics, customer evaluation and retail practice, in addition to other areas.
Next, let’s look at a few examples of how information about customers at the different stages of the trip lifecycle can be used to improve efficiency, enhance revenue and further the customer connection and overall trip satisfaction.
Additional Targeted Offers
Customer segmentation is a framework that utilizes multiple data sources and enables a variety of application in different areas of an airline’s commercial practice. It enables airlines to identify clusters of customers who are interested in specific levels of service so they can create additional targeted offers.
Targeted Network Planning And Scheduling
An airline’s planning and scheduling department is responsible for creating a schedule — the primary product airlines offer to their customers.
Decisions that are made by this department include selection of the markets the carrier will serve, the frequency of service and the departure times for individual flights.
Today, these decisions usually rely on average passenger demand for each flight. But it is certainly no business secret that most of an airline’s profit is generated by premium traffic.
A recent industry study presented at AGIFORS showed, for example, that while only 14 percent of passengers travel between the United States and Europe in first class and business class, those two classes accounted for almost half of the revenue generated on these routes.
The same study predicted steady growth for premium traffic over the next several years at 3.5 percent annually.
Customers who fly first class and business class have specific preferences and exhibit behavior that is decidedly different from an average economy traveler.
The schedules that best fit each of these different types of customer are therefore different, as well. Demand for premium cabin, for example, is typically much lower than average on weekends, but morning and afternoon peaks are more pronounced.
To recognize and properly identify these patterns, airlines need to switch from flight-level to cabin-level demand forecasting.
Historical PNR information represents customers in the planning stage. In particular, it contains the booking class that can be mapped into the specific cabin. And it, thereby, allows the carrier to identify the typical demand split among first, business, premium economy and economy for each flight.
This approach serves to help improve not only network planning and scheduling, but also fleet planning and capacity allocation because these processes determine cabin configurations so they directly affect the amount of premium traffic an airline can accommodate.
Importance Of Premium Travel
Most of an airline’s profits are generated by premium traffic. A recent industry study showed, for example, that while only 14 percent of passengers travel between the United States and Europe in first and business class, they contributed almost half of the revenue generated on these routes.
Sales And Promotions
Travelers at the shopping stage are actively looking for travel options. And one of the mechanisms for an airline to capture these customers is through sales and promotions.
Airlines apply these practices widely to stimulate demand and generate significant additional revenue.
But decision support based on data analysis in this area is still rudimentary. Significant opportunity can be found in better understanding of the effect of these tactics on customer behavior and optimized selection of markets, time periods and fares for which a promotion is active.
First, historical booking performance should be evaluated with respect to the impact a promotion had on markets on sale, other markets that can be affected by redistribution of demand and customer segments that were targeted by the promotion.
Second, promotions should be calibrated from both a demand-elasticity and a demand-stimulation perspective to enable what-if analysis.
Finally, optimal promotion design that accounts for revenue dilution, advertising and distribution costs, regulatory restrictions and combined revenue opportunity is suggested by an optimization engine and adjusted by analysis according to the airline’s corporate objectives and marketing strategies.
Now let’s look closely at an example of how information about customers in the pre-flight stage can be used to improve performance.
One of the standard practices airlines use to secure additional revenue is overbooking. And the most important component in a successful overbooking mechanism is the accuracy of the no-show forecast.
Today, most airlines use historical information about flight performance to predict the difference between the number of bookings made for a flight and the number of passengers who will actually be present at flight departure.
But at any point prior to departure, extremely rich and relevant information is available in PNRs that have already been created for a flight for which the carrier is trying to zero in on an as-accurate-as-possible overbooking level.
Based on close analysis of the characteristics of these PNRs, the airline can estimate their “survival” probability.
Reactive promotions (in red), initiated as a response to decreasing market share, are usually created as a response to competition. Proactive promotions target specific time periods and a set of markets in which an airline can increase market share. Promotion tracking and what-if analyses allows an airline to evaluate an impact by comparing results of a promotion with a prediction on performance without it.
In a recent study, Sabre Airline Solutions used a regression-trees algorithm that analyzes up to 40 different PNR attributes. The results indicated that the PNR no-show forecast was up to 7 percent more accurate than the aggregate flight-level approach.
This increment of improvement can directly result in airlines being able to sell approximately 1 percent more tickets on full flights without increasing the risk of denied boardings.
Similar methodology can be used in revenue integrity, onboard provisioning, cargo revenue management and other processes that require an accurate passenger boarded count.
Adoption of this approach should quite properly be expected to drive additional benefits through consistency of information in various business areas.
Operations Research: An Accomplished Past, A Promising Future
Since the early 1960s, operations research has been helping the airline industry optimize performance. During the past 10 to 15 years, customer centricity has become one of the leading concepts in developing automated decision-support systems.
Recent changes in data-management technology have provided airlines access to data describing consumer preferences on a much more detailed level.
Now airlines have an opportunity to make full use of all these advances in data analytics and operations research to further improve their planning and operations practices, as well as to provide better service to their customers.
The overall result is a major win/win: significant revenue enhancement opportunities for the airline, and a more customized and satisfying travel experience for it customers.