Airlines have been at the forefront of Revenue Management (RM) since deregulation in the 1980s, and the advent of low-cost, low-fare carriers, forced them to innovate their approach to pricing optimization. Since then, algorithm after algorithm and forecast after forecast have worked to squeeze every dollar possible from passengers.
Airlines who acted on the opportunities presented by Revenue Management saw revenue increase up to 18% without adding capacity! With such promising results, one would expect that all airlines would be experts in the field of airline revenue management. However, many aren’t. Most airlines still don’t implement RM. Why? Because they believe that:
- Revenue management has become very complicated, and they cant effectively practice it
- The only way to benefit from revenue management is to buy an expensive system
- Lack of resources or skills
- You need massive amounts of data
Despite these hurdles, all businesses, not only airlines, can realize the value of implementing RM. Whether you’re an international airline or a lemonade stand owner, you should not let these misconceptions discourage you from practicing RM.
One of the most challenging areas of airline revenue management is inventory control—how many of each fare type to sell to maximize revenue. Ideally, airlines would sell 100% of the more expensive/flexible tickets. However, there may not be enough demand to support this. While basic/restrictive tickets sell more easily, it is not in the airline’s best interests to sell out the plane with basic tickets. Hence a balance is required to determine the optimal “mix” of premium and basic tickets to sell out the plane and maximize revenue. A revenue management strategy can help accomplish this objective. The three components of an RM system are shown in figure one.
1. Extract, Transform and Load Data
This step requires historical transactional data to be “cleaned” before forecast calculations. Cleaning data includes changing the desired format of your data to allow forecasting and removing any outliers that may comprise/contaminate any valid insights. Outliers can be removed by using statistical methods (confidence intervals) or user-defined parameters (i.e. group bookings or IROPS). An area airlines have difficulty with is differentiating which flight was actually flown versus what the passenger initially booked (i.e. changes to a booking; due to Irregular operations [IROPS] or self-driven). This can have a significant effect on how demand is predicted. It is important to track data as close to the “true” demand of the passenger as possible.
This step applies an algorithm to the data to create a base of the expected revenue by booking class. There are three critical things to consider when calculating a forecast:
a. The Level of Detail:
There will always be a trade-off between complexity, accuracy, and volume. What is more efficient – predicting hourly flights, day to day, or week to week? There is a tipping point where greater detail will add no value and may have negative implications.
b. The Algorithm:
Most airlines and businesses take a statistical approach and use averages and variances to predict. Averaging provides the simplest and “best bang for your buck” in forecasting. More complex forecasting methods include finite regression mixture models, time series and drift methods.
Forecasts are done at some level that does not represent the entirety of an airline (i.e. only day of the week and not all the flights in the network). To have a network-wide forecast (i.e. every flight), each flight must have an accompanying forecast mapped to it.
This step takes the base from forecasting and applies revenue maximization techniques to build an optimal inventory. Many airlines and businesses skip this step and thereby leave money on the table. The goal is to protect seats within each fare class based on marginal value and probability of realizing demand. There are two methods to do this.
a. Flight Leg or Product Maximization:
Here, we are trying to calculate the value of selling one additional seat (the last seat) within a fare class. If this value is higher than the value of selling a seat in a lower class, then a seat in the higher class should be protected (sold). But how do we calculate this expected value? Assuming the probability of selling a seat is normally distributed, where we use the mean and the variance of the forecast to predict if a seat will be sold.
b. Origin-Destination (OD) (Network) Maximization:
Many airlines do not have this capability but can take a simplified approach to accomplish the same outcome. Upon implementation, this method can yield revenue improvements of 2-5%. Here marginal network value on the last seat of each leg needs to be calculated. This can be achieved in two ways.
i. Displacement Costs:
The idea here is that connecting passengers displace revenue throughout the legs of their journey. As a result, these origin-destination fares are mapped to network value, where Network value = total origin-destination fare – displacement costs. Displacement costs can be estimated by using a leg-based approach found via expected marginal seat revenue.
ii. Bid Price:
The marginal network value of a leg can represent the bid price of that leg. For a single leg journey, the fare is accepted if it is higher than the bid price. For a multi-leg journey, the origin-destination fare must be greater than the sum of all the bid prices.
Implementation of an origin-destination management system is what holds many airlines back from practicing it. A way around this is by using a value-based class mapping. Here the fixed relationship between fare type and booking class is no longer applicable. Booking classes are defined according to revenue (i.e. fare), and availability is based on booking class. This article primarily focused on mid-sized airlines, although many of the principles can be adapted to other businesses. For example, EMSR can be used in manufacturing and CPG when deciding how price changes will affect sales. Network optimization and bid price can be used for B2B when deciding trade-offs between projects.
Whether you are a multi-million-dollar airline or a tier 3 manufacturer, all concepts in this article can be easily applied to beat your competition!
ABOUT THE AUTHOR Marc Carias is a Consultant at Revenue Management Labs. Revenue Management Labs helps companies develop and execute practical solutions to maximize long-term revenue and profitability.