New Generation of Revenue Management Systems and Processes

Published January 2017 in Airline Leader: Issue 38

For airlines, the increasing complexity in their operations and the frustrations as well as changing behaviour of their customers are calling for a need for new generation of revenue management systems.

Consider, for example:

  • the dramatically changing distribution landscape;
  • the development of different partnership types and code share agreements within and outside of alliances;
  • the transition from leg-based systems to O&D-based systems creating a need for the use of different data, different analytics, different human skills, and different KPIs;
  • the concern relating to the use of social media for customers to post reviews that has been impacting user-generated content;
  • the increase in customers’ frustration relating to the time spent shopping, the rapid change in availability of a fare, and, more recently, the price paid by a customer may not end up being the final price if some additional charges for ancillary products and services show up that were not clear at the time the reservation was made, and;
  • some customers’ frustration with the industry’s ways of maximising revenue, leading them to look at ways to ‘reverse engineer’ the industry’s revenue management process and find ways to optimise their own booking processes.

Although revenue management systems and processes have evolved significantly in the past four decades there is still considerable upside potential given the availability of much more comprehensive and real-time data, analytics to harness customer and business intelligence, and enabling technologies to work with the non-linear processes to make smarter decisions. Examples include the use of a wide range of information on customers’ preferences (and purchases as well as consumption behaviour), the use of web analytics to improve conversion rates, and the availability of algorithms for generating the best set of itineraries within an alliance network in response to a shopping request. Now systems are able to handle much more complex data in almost real time. Also, in the context of consumers’ preferences, weighted according to the parameters provided by consumers.

The general framework of the new generation of revenue management systems and processes are being influenced by, at least, the following six assumptions:

  • Customers now have much more information during shopping and booking patterns are changing.
  • Customers are looking for, to varying degrees, personalised services with focus on experience throughout the journey. (See Figure. 1)
  • Airlines want to offer the right content to the right customer through the right channel at different stages of the trip. (See Figure. 1)
  • Airlines want to maintain and or increase profit margins from the sale of ancillary products and services;
  • From the back-office viewpoint, heads of different functions within an airline want to see the ramifications of various decisions to ensure optimisation at an enterprise level, and;
  • Airlines, under pressure for higher profit margins, want to eliminate or reduce yield leakage.


While the framework still works around the two basic dimensions – pricing and inventory – there is now a focus around a third dimension – real time. The object now is to manage both prices and inventory in real time. Within the existing framework pricing has remained fairly static relative to inventory control. Even with respect to inventory which has been forecast by booking class and then varied by flight, the variation has been done manually. Consequently, the use of manual processes has limited the number of changes. If a large airline has 5,000 flights and 30,000 itineraries per day, with booking classes made up for 330 days ahead, how many flights can be monitored manually?

Figure. 1

Ironically, while some advanced revenue management systems do exist that can suggest movements in inventory automatically, analysts have tended to override suggestions made by these systems even though they can only monitor few flights and take into account only a few implications. Presumably, analysts do not trust the automated system nor feel confident in the data presented. Two key reasons behind the lack of trust include the latency of data used for batch processing, and the fact that recommendations are provided without rationale. Consequently, there is no transparency relating to the recommendations.

Interestingly, even when revenue management analysts have attempted to make dynamic changes, they have focused on internal considerations since they do not have full knowledge of what is happening in the marketplace with respect to actions of competitors – pricing and inventory tactics – as well as aspects of the shopping activity. 

Manual decisions can be acceptable for a small airline or even a large one, such as a low-cost carrier if it has limited connecting traffic. But for a large airline, analysts require systems that are contextually sensitive. These systems are becoming widely available, and will create even more value for both airlines and their customers with the further incorporation of machine learning algorithms to model consumer choice.

One key decision-making criterion in the revenue management process is the segmentation process. Previously, it was undertaken with the development of booking classes that, in turn, were controlled by vectors such as the booking time prior to departure and the conventional elements of trip purpose. Recently, questions have been raised about the viability of customer-based segmentation. However, given the complexity of customer-based segmentation, technologists are initially exploring segmentation for the purpose of the trip based on ‘revealed preferences’. A ‘revealed preference’ parameter could relate, for example, to the brand of the airline or a preferred airport for making a connection based on the type of trip – a fast connection through airport A vs. a leisurely connection for shopping through airport B. Consideration could also be given to the airline-airport on-time performance.

The trip purpose envisioned can go well beyond the conventional attributes (business, leisure, and VFR). For example, is the business trip a day-return or a trip requiring overnight stay for one or more days? Does it include a personal travel component at the beginning or end, and does it have one or more travellers in the group? A trip-based segmentation technique can work when the customer is anonymous or identified. And answering requests based on trip segmentation can also take into account compliance with corporate policies and rules to obtain higher conversion rates.

The advanced versions of revenue management systems being developed will also enable pricing analysts to develop and manage branded fares and bundled fares much more efficiently and effectively. In the case of branded fares, the optimisation process can relate to not just travel components and prices, but attributes related to personalisation and experience as well. In the case of bundled fares ancillary products and services, the airlines’ own products (seats and lounge access, for example) or products produced by travel partners (hotels, car rental companies, cruise lines, etc.), can be included. In fact, it may be possible to let the customer develop the bundle based on not just the components desired but also the weight given to each consideration.

The new generation of revenue management systems are being designed to incorporate algorithms for generating the best set of itineraries in response to a shopping request – inbound and outbound with considerations of fares and fare conditions – to meet customer needs and maximise conversion rates. Take, for example, a request for an itinerary between New York’s JFK Airport and Singapore. The OTA website of Expedia shows approximately three dozen airlines offering services (05-Dec outbound and 13-Dec inbound) involving one or two stops with fares ranging between USD807 and USD6,226. A consumer can rank the order of available itineraries by price, trip duration, time of departure, or time of arrival. While there are four ranking possibilities, a consumer can only select one of the four options at a time. The new systems enable the inclusion of not only more preference-based parameters, but weigh the importance of each parameter to suit an individual customer’s needs as well. Examples of additional parameters include the importance of the connecting airport and the refundability or non-refundability of the fare.

While airlines clearly have made significant progress in managing their revenues there is still more upside potential with the use of emerging systems and processes. Just as the switch from leg-based systems to O&D-based systems was a game changer, there is now an opportunity for another game changer given the capability to manage both prices and inventory dynamically. Moreover, the offer management process, particularly involving online activity, could itself be a step-changing process once developers harness the power of artificial intelligence and its role in the development of machine-driven natural language recognition, understanding, and translation.

At what level of sophistication – technology and human skills – should an airline manage its revenue? It depends on its vision, its business model, and its brand. Just from the distribution perspective alone, does the airline want to control the distribution of its products and services (including controlling and monetising its data), or does it want to focus on the development of its products (network, schedules, on-time performance, etc.,) and let someone else perform its revenue management? Or, does it want to be a wholesaler and sell its seats at various net prices to various distributors?



© 2010-2017 CAPA - Centre for Aviation