Revenue Management in Aviation Industry
Revenue Management is considered the backbone of the aviation industry. Predicting the right demand, timing, and pricing are responsibilities of the revenue management system. Technology has brought radical changes in the way revenue is realized by the airline industry. However, technology has a lot more to offer compared to what is being implemented. In this blog, Kelwin Kuriachan, Senior Principal Product Consultant, talks about the future scope of technology and its role in revenue management, the challenges involved in revenue management, and how AI can help with continuous pricing.
What are some of the challenges faced in airline Revenue Management (RM)?
There are many challenges in airline revenue management, from its deep-rooted association with legacy systems and processes to the manual interventions required by analysts. However, before diving into these challenges, I think it is important to understand the mission objective of the revenue management team at the airline.
The mission of this team is to find the optimal price for an airline ticket with which the airline is able to operate and maximize its profits. Ideally, an optimal price of a ticket is a sum of the following components -
- Opportunity cost of selling the service (ticket),
- Optimal margin for the customer based on their willingness to pay and
- Marginal costs of carriage (including the taxes and interline settlements).
There are challenges that exist in deriving each of these components.
- Finding the
opportunity cost of selling the service.
In today's revenue management practice, a forecast of future demand is estimated based on historical data. Airline ticket prices today are based on published fares and inventory held in the form of RBDs. This is a legacy methodology that the airline industry is grappling with, as it does not reflect true demand-based pricing. The airlines are left with no option other than to look at the history of bookings that are made using the concept of published fares and the RBDs. The problem with this concept is that in the historical data points, there is only a limited set of static price points and limited allocations of inventory.
Traditional revenue management systems have become outdated.
Primarily, the legacy systems do not have the inherent capability to adjust the forecast based on real-time demand (the current RM systems at many airlines failed to perform during highly volatile conditions of covid), and secondly, most of these systems do not make use of forward-looking data from non-airline sources.
The next important thing is to find the optimal margin for the customers based on their willingness to pay. The biggest challenge in this component is primarily due to the current RM system's inability to determine accurately a customer's willingness to pay in real time. This forces the airlines to settle for fixed price points resulting in fare jumps, losing price-sensitive customers, and consequently losing revenue between two set price points.
- Finding the marginal cost of carriage associated with operating airlines.
In today's environment, it is the most challenging variable because of the high volatility that exists in fuel costs and foreign currency-denominated costs. Moreover, the data for the cost of carriage is not reviewed frequently and is not available for use with the current revenue management systems. A number of analysts are involved in attaining this objective. These analysts are required to constantly monitor the routes and update strategies in the form of rules and make updated to inventory etc. Even in airlines with a well-defined structure, there is often a lack of coordination between teams, and the structure itself creates duplication of effort. All of this leads to inefficiencies due to high levels of manual involvement.
According to you, what is the way out of this legacy problem of static price points? Do you think continuous pricing is the way forward?
Yes, the airlines have to move away from the concept of static price points to a continuous one gradually. Since the current RM systems and processes are deep-rooted in legacy, it is not going to be an easy path. The transition will be slow and an airline has to take a step-by-step approach. Continuous pricing has been in the airline industry's spotlight for all the right reasons. Almost everyone in the airline industry understands that continuous pricing is not about generating new revenue but rather about reducing the lost revenue through greater efficiencies.
Recent advances in artificial intelligence and machine learning have paved the way for a future in which each product – whether a ticket, an ancillary, or even a bundle can be priced optimally, in real-time. These enable airlines to make better pricing decisions by considering multiple parameters to accurately estimate current demand as well as a specific passenger's willingness to pay. Pilot implementations across various airlines have demonstrated that this methodology results in revenue increases, explaining why many airlines are prioritizing continuous pricing.
Why AI for continuous pricing?
Currently, airlines use RBDs and fare basis codes to segment demand, which only allows a limited and insufficient number of static price points in a market. A fare basis code consists of up to eight characters and contains information such as fare class, season, weekend/weekday, advance purchase, minimum stay, point of sale, and so on. This is used to categorize customers into various demand segments. However, due to the limitations of manually managing many combinations, airlines cannot fully utilize these. Most network airlines work with a maximum of three hundred to four hundred fare basis code entries for a given origin and destination pair, whereas the possible segmentations can be much higher, reaching hundreds of thousands or even millions in some cases. Current systems are incapable of handling such a large number of segmentations. Even if systems were capable of doing so, managing such volumes would be impossible for humans. This is where Artificial Intelligence (AI), or more specifically, Machine Learning (ML), can come in handy.
ML can identify demand patterns in historical data and, using probability, determine the demand segment for a booking. This capability is used to calculate price points based on demand segments without the use of complex rules such as footnotes and fare basis codes. This enables airlines to create virtually infinite price points, breaking free from existing constraints. Furthermore, better segmentation can be achieved with ML, even with existing data sources, before bringing in other external data sources. True continuous pricing can only be achieved by calculating fares in real-time as and when a booking request is made, as this is the only method capable of accurately capturing the current demand context. Once an airline has achieved real-time price determination capabilities, implementing self-learning AI algorithms help optimize fares generated by the system, so that revenue is maximized based on conversions.
Further, in today's RM teams, the majority of the work done by analysts is manual, repetitive, and time-consuming. They have to go through a number of reports and rules in order to determine the current fare or inventory levels or to determine the various forecasts. Some of these processes are automated in existing revenue management systems but the complexity of such rules and their maintenance itself invites additional manual work. Leveraging AI technology to better automate these routine tasks is a major use case for an airline starting to implement continuous pricing.