It has been exactly 32 years since I have been involved with Revenue Management (RM) in the Airline industry. I have seen the evolution of Airline Revenue Management over the last three decades, especially, in Air Cargo Revenue Management. Given my evidence-based learning, I can group the evolutionary aspects into these four major categories:
- Business Focus and Scope
- Inventory Control – How access to inventory is decided for a booking request
- Accounting for additional revenue drivers
- Newer and better methods to forecast supply and demand
While the evolution over the last three decades has been centred around the original concepts and initial framework of Revenue Management, the future seems radically different. The advent of big data and advances in hardware have enabled efficient storage of large amount of data and faster processing of data. The evolution of AI/ML–based methods, and more importantly the desire and need to simplify the RM process are pushing the boundaries of Revenue Management into a radically different space. This space no longer looks anywhere close to the traditional Revenue Management approach – that has remained intact for over half a century.
It looks like many of the standard revenue management modules or components and the sequence in which they are executed to arrive at controls to access inventory may become obsolete and may not be required.
I would like to expand a bit more on the evolution so far and then take you through my conceptual visualization of the future of RM.
The Evolution of Airline Revenue Management
I am not planning to list every single feature, functionality, methods, or mechanism that has become part of RM now or has evolved over time. My intent is to highlight some of the key ones.
The Business Focus and Scope
The level at which we optimize between demand and capacity shifted from leg to segment to network (Origin – Destination). The focus changed from maximizing revenue to maximizing profit or margin. The solution also incorporated some of the operational and commercial constraints along with user interfaces that closely represent and support optimal user experience through better look and feel and streamlined process flow.
Access to inventory is the goal of revenue management. The classic definition of revenue management is to sell to the right customer, at the right time, and at the right price.’ This is typically done in a couple of ways. One way is to allocate inventory to various demand groups and nest them appropriately. Demand groups can be defined in multiple ways with the objective of products/customers/commodities/passenger that have similar price/rate and consumption of space per unit of request. When a request comes if space is available for that group, then it is accepted. The other approach is to use bid prices as bid prices represent the value of inventory at any given time. Low-cost carriers have a similar approach but set up in an easier way to understand and execute prices change, typically go up as a function of time or depletion of inventory.
Accounting for Additional Revenue Drivers
Over a period, airlines started differentiating products using additional variables or services, in some cases the services that were included for free. For example, airlines started leveraging the need for early check-in/seat selection and boarding priority, legroom, checked bag, and onboard food to differentiate products and make more money. On the cargo side, these can be in terms of service time, cold storage, door to door pick up or delivery, etc.
Newer and better forecasting methods
The type and extent of data, assumptions on underlying distributions of the data, and the methods used for forecasting have advanced significantly. Innovation in scientific methods and advances in technological platforms enabled the solution to go to the next level. Some of us at RTS have been quite fortunate to be part of the entire journey of air cargo revenue management right from the start to till now. Our experiences with several carriers in this area has helped us to understand the barriers to realizing the benefits from implementing an air cargo revenue management. And has enabled us to identify key success factors for implementing the right revenue management solution.
Conceptual Visualization of the Future of Revenue Management – My Take
What are the driving forces that could reshape the future of RM. From the business and user standpoint, the expectation is to reduce the number of steps and models in revenue management and simplify the processes and minimize user intervention and monitoring.
From a business process and organizational standpoint, there has always been debate about the separation of pricing and revenue management functions. The future of revenue management is likely to combine the concepts of pricing and revenue management and combine them into single step in terms of decision making. The decision will simply be one of two things: what prices and itineraries to offer to a customer when a request comes or accept or reject an offer (price and itinerary) from a customer. This new framework will combine the concepts of bid price, willingness to pay, and value of a customer.
Bid price provides information on the value of space at a given time. Today, it is generally determined based on demand and supply. In the future, bid prices are likely to be forecasted or determined directly. Willingness to pay considers price elasticity and probability of a customer accepting a price. Customer value provides an indication of how offered price may have to be adjusted based on how important the customer is for the airline. So, access to inventory when space is available is primarily determined based on the above three factors in a single process or step.
The determination of available space will be done differently in the future. Available space for sale will not be determined using a separate overbooking module. Overbooking will be performed at booking request level rather than at flight capacity level. Traditionally, capacity is increased by a certain amount to account for cancellations and no-shows. rather than overbooking a flight and increase available inventory. In the future, the aspect of overbooking will be incorporated at the booking request level by predicting if the booking will materialize and how much of it will materialize in the case of cargo.
The future will ‘disrupt’ the current framework and methods of revenue management.
What Happens Next?
The future of revenue management, as I painted here, is one of the few possible scenarios though I am almost certain that other scenarios will still be a minor variation of this. We at RTS have created a conceptual framework for the next generation RM. Our current dynamic pricing methodology which is more like a continuous pricing approach. In dynamic pricing, prices change with time but they are still picked from pre-defined set of prices. In continuous dynamic pricing, prices are not limited to pre-defined set of prices; it can be any price based on bid price, willingness to pay, customer value, etc. Our pricing solution already has advanced analytic hierarchic process coupled with regression based methods to determine customer values. We are continuing to research and incorporate sophisticated machine learning methods to forecast demand in our RM solution. We have created a predictive analytic model to forecast show up behavior at booking request level. There has been some work done in this area but has not been adopted by the industry. We are now working on incorporating and testing it to reflect a new way of accounting for overbooking in a more robust manner. The next step will be to create use cases, and discuss the approach with airline thought leaders, revenue management researchers, and technology leaders.
We have researched and developed prototypes of several components of future RM solution. We need to extensively test them, fine tune them, and evaluate the feasibility of incorporating them into a process that is simpler and more streamlined than what is in place today. Making the leap into what I have described here requires a lot of things to come together – industry buy-in, significant development of AI models, and technology to handle big data.