8

Revenue Management

Revenue management (RM) (once called yield management) refers to the strategy and tactics used by a number of industries—notably the passenger airlines but also including hotels, rental cars, cruise lines, and others—to manage the allocation of their capacity to different fare classes over time in order to maximize revenue. Revenue management is applicable when all of the following conditions hold.

• The seller is selling a fixed stock of perishable capacity.

• Customers book capacity prior to departure.

• The seller manages a set of fare classes (also called booking classes), each of which has a fixed price (at least in the short run).

• The seller can change the availability of fare classes over time.

Revenue management can be considered a special case of variable and dynamic pricing with constrained supply. However, the final two conditions give revenue management its special flavor. Revenue management is based not on setting and updating prices but directly on setting and updating the availability of fare classes, where each fare class has an associated fare (price) that remains constant through the booking period. This distinctive feature is a legacy from revenue management’s origin. The passenger airlines that pioneered revenue management in the 1980s needed to utilize the capabilities they had at hand. This meant using the booking controls embedded in their reservation systems as the primary mechanism for controlling the fares displayed to customers at any time. Following success at the airlines, revenue management has been adopted by a number of other industries, including hotels, rental cars, freight transportation, and cruise lines—many of whom use the same (or similar) reservation systems as the passenger airlines.

We use revenue management industries as a generic term for industries meeting the conditions listed above and revenue management companies to refer to companies that make use of booking controls and fare classes. While these companies apply revenue management in its purest fashion, the techniques of revenue management can be applied to any situation in which a seller needs to determine how to allocate a fixed supply among different channels or customer segments paying different prices.

This chapter serves as an introduction to revenue management. It is a prelude to the next three chapters, which address specific aspects of revenue management in more detail. We start by surveying some of the history of revenue management and then look at revenue management strategy as an application of some of the price differentiation techniques introduced in Chapter 6. In Section 8.2, we discuss how revenue management operates at three different levels: the strategic, the tactical, and that of moment-to-moment booking control. Section 8.3 describes the customer segmentation strategy underlying revenue management. Section 8.4 describes the interactions between the reservation and distribution systems in the travel industry, and Section 8.5 describes how bookings and cancellations are managed within those systems. Understanding this material is a prerequisite for understanding the ideas behind tactical revenue management. Section 8.6 introduces the three elements of tactical revenue management: capacity control, network management, and overbooking, which are described in detail in Chapters 9, 10, and 11, respectively. Section 8.7 describes various metrics that are used to measure the effectiveness of revenue management. Section 8.8 discusses the role of incremental cost and ancillary profit in revenue in various industries, and Section 8.9 reviews the current status of revenue management adoption in various industries.

Since this chapter is introductory, it may feel a bit lacking in algorithms and computations. This is made up for in the following chapters, each of which addresses a specific aspect of tactical revenue management in detail.

8.1 HISTORY

Prior to 1978, the airline industry in the United States was heavily regulated. Both schedules and fares were tightly controlled by the Civil Aeronautics Board (CAB). Fares were held sufficiently high to guarantee airlines a reasonable return on their extensive capital investments in airplanes. In 1978, Congress passed the Airline Deregulation Act. The act specified that the industry would be deregulated over four years, with complete elimination of restrictions on domestic routes and new service by December 31, 1981, and the removal of all fare regulation by January 1, 1983. The result was a shock from which it took many years for the airlines to recover.¹

One of the reasons Congress passed the Airline Deregulation Act was to inspire and encourage creative new entrants into the passenger airline business. One of the first new airlines to arise after deregulation was PeopleExpress. PeopleExpress was not unionized, and it offered a bare-bones service, with passengers paying extra for baggage handling and onboard meals. As a result, its cost structure was significantly lower than those of American and the other major airlines, such as United and Delta. By offering fares up to 70% below the majors, PeopleExpress filled its planes with passengers from a previously untapped market segment: price-sensitive students and middle-class leisure travelers who were induced to travel by the existence of fares far lower than anything the industry had seen before. PeopleExpress built its business initially by entering underserved markets, where its competition was primarily bus or car travel. PeopleExpress experienced four years of phenomenal growth and, in 1984, began service on Newark–Chicago and New Orleans–Los Angeles routes—key markets for American Airlines.

The choice American Airlines faced was stark. If it matched People’s low fares, it might retain its customer base but would not be able to cover its costs. If it did not match People’s fares, most of its customer base would be siphoned off by the low-price competitors. And in the newly deregulated environment, nothing could keep a successful PeopleExpress from moving in on all of American’s core markets. American Airlines seemed doomed—its only choice appeared to be between a slow death and a rapid one. However, Robert Crandall, CEO of American Airlines, formulated a counterattack: American would compete with PeopleExpress on low fares and simultaneously sell some of its seats at a higher fare as well.

In January 1985, American Airlines announced its Ultimate Super Saver Fares program. Ultimate Super Saver Fares matched PeopleExpress prices, with two key differences:

• For a passenger to qualify for an Ultimate Super Saver discount fare on American, she would need to book at least two weeks before departure and stay at her destination over a Saturday night. Passengers not meeting this restriction would be charged a higher fare. In contrast, PeopleExpress put no restrictions on its discount fares—every passenger paid a low fare.

• American restricted the number of discount seats sold on each flight to save seats for full-fare passengers who would be booking within the last two weeks prior to departure. PeopleExpress allowed every seat to be sold at a low fare.

American’s two-pronged approach was carefully thought out and turned out to be the perfect counter to People’s challenge. The booking restrictions ensured that the vast majority of the discount passengers buying American were leisure passengers who were able to book early and who were more price sensitive. The later-booking passengers who paid full fare were primarily business travelers who were less price sensitive but needed seat availability at the last minute. Both groups of passengers preferred American’s superior service to People’s bare-bones approach. In effect, American Airlines had segmented the market between leisure and business travelers and used differentiated pricing to attack a competitor.

The impact of American’s actions was dramatic. American announced the new fares and new structure in January 1985. By March of that year, PeopleExpress was struggling, and by August it was on the verge of bankruptcy. In September, Texas Air bought PeopleExpress for less than 10% of the market value it had enjoyed a year earlier. PeopleExpress’s flamboyant CEO, Donald Burr, put the blame for his airline’s demise squarely on American’s superior yield management capability: “We had great people, tremendous value, terrific growth. We did a lot of things right. But we didn’t get our hands around the yield management and automation issues” (Cross 1997, 129).

This was the origin of revenue management (or yield management as it was called at the time). American’s success prodded other major airlines in the United States to develop revenue management capabilities of their own. Throughout the remainder of the 1980s and well into the 1990s, carriers such as United, Delta, and Continental invested millions of dollars in implementing computerized revenue management systems and establishing revenue management organizations. As world aviation markets were increasingly deregulated, carriers in Europe and Asia also began to adopt revenue management. Hotels and rental car companies followed the airlines in adopting revenue management. Marriott was a pioneer in hotel revenue management, and Hertz and National were pioneers in rental car revenue management. Vendors such as PROS and Talus Solutions developed and sold commercial revenue management software packages. The next wave of adopters included cruise lines, passenger trains, and various modes of freight transportation. Development and investment in revenue management continues in many of these and other industries today.

8.2 LEVELS OF REVENUE MANAGEMENT

Successful revenue management requires consistent execution at three levels, as shown in Table 8.1. Revenue management strategy is the identification of customer segments and the establishment of products and prices targeted at those segments. Once products and prices have been established, revenue management tactics require setting and updating limits on how much of a particular product can be sold at a particular fare to each segment for some period of time—say, a day or a week. Booking control is the moment-to-moment determination of which booking requests should be accepted, which should be rejected, and how capacity for different products should be updated in the face of bookings and cancellations. For airlines, hotels, rental car companies, and cruise lines, booking control is performed by the reservation system. Tactical revenue management is the brains of the process. It is where future demand is forecast, optimization algorithms are run, and booking limits are set and updated.

We first consider the strategy of revenue management and how it applies in different industries. We then turn to a somewhat detailed discussion of booking control—the real-time face of revenue management. With this background in place, the next three chapters deal with elements of tactical revenue management—capacity control, network management, and overbooking.

8.3 REVENUE MANAGEMENT STRATEGY

Revenue management strategy consists of the identification of customer segments and the establishment of products targeted at those segments. A fundamental element of revenue management strategy at many hotels, rental car companies, and airlines is the distinction between leisure customers and business customers first recognized by American Airlines. The different characteristics of leisure and business customers are shown in Table 8.2. The airlines used these characteristics to segment their market and create virtual products oriented toward the different segments. A typical approach is shown in Table 8.3. Here, an airline has identified five customer segments, three business and two leisure, with at least one product targeted toward each segment. The leisure products have various restrictions (early purchase, Saturday night stay) that make them unattractive or unavailable to many business travelers. This is a classic application of the segmentation technique described in Section 6.3.5 by which a seller artificially creates an inferior product to sell at a lower price in order to segment the market.

TABLE 8.1

The three levels of revenue management decisions

TABLE 8.2

Characteristics of leisure and business airline-passenger segments

TABLE 8.3

Typical airline market product segmentation

Product versioning is not the only price differentiation tactic used by the airlines. They also use most of the other tactics described in Section 6.3. Airlines sell products targeted to many segments such as government, senior citizens, groups, tour operators, and cruise lines. International airlines are great believers in regional pricing. They will sell the same tickets for different prices in different countries (even after adjusting for exchange rates) to exploit differences in price sensitivity. Airlines are also channel pricers; ticket prices may vary among online platforms and on their website.

Other revenue management companies followed the airlines in segmenting their market, creating virtual products and establishing a wide range of prices. Hotels and rental car companies establish and manage products and different prices oriented toward various groups: corporate, leisure, and business segments. As a simple example, many hotels in Hawaii maintain kama’aina rates for local residents.² These rates are often far lower than the standard rates, and hotel revenue managers make them available during slack periods to fill their properties. Only customers who can prove Hawaiian residency qualify for a kama’aina rate. This is a classic example of group pricing. By confining availability to locals, hotels ensure that the low kama’aina rates do not cannibalize their high-paying business from the mainland United States and Asia.

Cruise lines do not have late-booking business customers, but they do sell to the incentive segment—bulk purchases of cruise capacity by companies to be used as incentives for their employees. For example, a medical supply company might purchase 20 berths on a particular sailing to reward the 20 top North American sales managers. This incentive business is booked earlier and is sold at lower fares than individual cruise tickets. Cruise lines also charge different rates for customers from different cities. Managing availability by city is one way they can maximize contribution from each sailing.³

8.4 THE SYSTEM CONTEXT

American Airlines was able to outmaneuver PeopleExpress in part because it had developed a computerized reservation system called SABRE that allowed it to save seats for later-booking business passengers—a capability PeopleExpress lacked. When it was turned on in 1964, SABRE was a technological marvel. Not only did it replace the colored index cards in revolving trays that American and other airlines had used since the 1930s to manage their reservations; it gave American Airlines the ability to distribute its products and fares globally and to receive bookings from anywhere in the world.

Airline distribution has evolved considerably over the past few decades (and continues to evolve), but the distribution systems and reservation systems continue to play an important role. A high-level schematic diagram of airline distribution is shown in Figure 8.1. Consumers have a variety of channels for purchasing tickets ranging from online platforms such as Expedia and Orbitz, an airline’s own website, travel agencies, and in-house corporate travel management. The global distribution systems transmit price and availability information from the airline to each of these channels and send booking and cancellation information from the channels back to the airlines. Every airline has a reservation system that enables it to keep track of all of the current bookings that it has in the system and their status. As of 2019, four major global distribution systems are in use—Amadeus, World-span, Travelport, and China’s TravelSky. Most airlines distribute through all four systems because individual travel agencies (including corporate travel) are typically associated with only one of the systems. The same global distribution systems provide similar services for other travel-related industries such as hotels, cruise lines, and rental cars.

Figure 8.1 Airline distribution channels.

This system background is important because airlines did not choose to pursue a revenue management strategy and then design reservation systems to support that strategy. Rather, reservation systems and global distribution systems were developed first and revenue management capabilities were added later. SABRE went operational in 1964—14 years before the airlines were deregulated and 21 years before American first did battle with PeopleExpress. The computerized reservation systems were among the most sophisticated and complex transaction-processing systems of their day. But they were not very flexible. Since they were built on large mainframes using 1960s-vintage software, major modifications were (and are) expensive and time consuming. As a result, the way airlines and other travel companies manage their inventory is still dictated by the capabilities of these systems. And at the heart of these systems are the booking control mechanisms that determine moment by moment which fare classes are available for sale and which are not.

8.5 BOOKING CONTROL

Booking control is the real-time face of revenue management. The primary function of booking control is to determine whether each booking request received should be accepted or rejected.⁴ Secondary functions are to update availabilities for different products when bookings and cancellations occur. As shown in Figure 8.1, booking requests stream into the reservation system from many different channels. Within a very short time (typically less than 200 milliseconds), the reservation system needs to send a message saying whether or not that request can be accepted. Given the short time available, reservation systems rely on very simple, mechanical procedures for determining whether to accept a request.

When a booking request is received—say, for a seat for a future flight or a hotel room for some future dates—the request is assigned to a fare class. This assignment may be based on the time and characteristics of the request (e.g., whether it meets the qualifications for a deep discount), the channel through which the request was received, the market segment of the customer (e.g., group versus individual), or combinations of all of these. The fare classes are typically assigned letters, so we can speak of a B-Class or M-Class booking request.⁵

The reservation system includes a booking limit for each fare class on each product. When a booking request is received, the reservation system checks the booking limit for the associated fare class. If there is sufficient availability, the request will be accepted; if not, the request will be rejected.

Example 8.1

An airline receives a B-Class request for three seats on Flight 137 from Houston to Miami, departing in two weeks. The current B-Class booking limit for this flight is two. Because there is insufficient availability, the request is rejected.

At each time, the booking limits currently in place determine which booking requests will be accepted and which will be rejected. When a new booking is accepted or a previously accepted booking cancels, the reservation system automatically updates the booking limits for that product. This is all done in milliseconds, consistent with the need of the reservation systems to communicate availabilities to customers in real time.

8.5.1 Allotments

An obvious way to manage bookings would be to divide the available capacity into discrete chunks and to allocate each chunk to a fare class. This is known as the allotments approach, and the size of the chunk allocated to each fare class is called its allotment. Under this approach, bookings are accepted in a class until the allotment for that class is exhausted.

Example 8.2

A 100-seat aircraft is being managed using allotments: 30 seats have been allotted to deep-discount bookings (B-Class) with a $125 fare, 45 seats to full-fare coach (M-Class) with a $200 fare, and 25 seats to business class (Y-Class) with a $560 fare. Two weeks before departure, 25 B-Class bookings, 45 M-Class bookings, and 10 Y-Class bookings have been accepted. The remaining allotments are 5 seats for B-Class, no seats for M-Class, and 15 seats for Y-Class.

While the allotments approach is easy to understand, it has a major drawback: it does not work very well. In particular, it can result in high-fare customers being rejected while lower-fare customers are still being accepted. In Example 8.2, if the M-Class allotment closes before the B-Class allotment, the airline will be accepting customers paying $125 while rejecting $200 bookings. This is a prime revenue management sin—an airline cannot maximize revenue by rejecting high-fare customers to save seats for low-fare customers. For this reason, revenue management companies nest their inventory so that high-fare customers have access to all of the inventory available to lower-fare customers.

8.5.2 Nesting

Nesting was developed to avoid the situation in which high-fare bookings were rejected in favor of low-fare bookings. To describe nesting, we number fare classes so that 1 is the highest (most expensive) fare class and n is the lowest. We define b_i as the booking limit for class i. With nested booking controls, booking limits are always nondecreasing; that is,

Furthermore, at each time, every fare class has access to all of the inventory available to lower fare classes. This avoids the possibility of accepting low-fare bookings while rejecting high-fare bookings.⁶ The booking limit for the highest class, b₁, is an upper bound on the total number of bookings that will be accepted. If the airline is not expecting any no-shows or cancellations, then b₁ would equal the capacity of the flight. When there is a possibility that bookings will cancel or not show, then it is usually optimal to overbook—that is, to accept more bookings than available capacity, in which b₁, and possibly the booking limits for other classes, might be greater than the capacity of the flight. Chapter 11 addresses how to calculate overbooking limits.

We can also describe nesting in terms of protection levels. The protection level for class i is the total number of seats available to class i and all higher classes. Let y_j be the protection level for class j = 1, 2, 3, . . . , n – 1. The relationship among booking limits and protection levels with nested booking controls is illustrated in Figure 8.2. From this figure it should be evident that

and y_n = b₁; that is, the protection level for the lowest fare class is equal to the total booking limit for the flight. Combining Equation 8.2 with condition 8.1, we can see that protection levels decrease for higher fare classes; that is,

0 ≤ y₁ ≤ y₂ ≤ . . . y_{n – 1},

as can be seen in Figure 8.2.

8.5.3 Dynamic Nested Booking Control

What happens when we accept a booking when limits are nested? Define the availability of a fare class as the booking limit for that class minus the booking limit of the next lower class. The availability of the lowest fare class is its booking limit. Thus, if a flight has two fare classes with booking limits (b₁, b₂) = (100, 30), the availability for class 2 is 30 and for class 1 is 70. The standard practice in the airline industry is to decrement the availability of the booking class from which the booking occurred, from which the booking limits can be reconstructed. Thus, if the airline with booking limits (100, 30) accepted a five-seat booking in fare class 2, its new booking limits would be (95, 25). If it accepted a five-seat booking in fare class 1, its new booking limits would be (95, 30).

Figure 8.2 Nested booking limits and protection levels.

This approach is called standard nesting and it is the most widely used among the airlines. However, an alternative is to decrement the availability of lower fare classes as well when a higher-fare booking limit is accepted. In this case, accepting a five-seat booking in fare class 1 would result in new booking limits of (95, 25). This approach is called theft nesting and it is sometimes found in practice.

Example 8.3

A flight has a total booking limit of 100 and five fare classes with booking limits of (b₁, b₂, b₃, b₄, b₅) = (100, 73, 12, 4, 0). Note that this flight is currently not accepting any bookings from class 5. Any booking class with a limit of 0 is said to be closed. We can derive the corresponding protection levels using Equation 8.2: (y₁, y₂, y₃, y₄, y₅) = (27, 88, 96, 100, 100). The availabilities for the fare classes are (a₁, a₂, a₃, a₄, a₅) = (27, 61, 8, 4, 0). Assume that a booking request arrives for 3 seats in class 3. Because the booking limit in class 3 is 12, this request will be accepted. Under standard nesting, only the availability in class 3 will be changed, so the new availabilities will be (a₁, a₂, a₃, a₄, a₅) = (27, 61, 5, 4, 0), which corresponds to booking limits of (b₁, b₂, b₃, b₄, b₅) = (97, 70, 9, 4, 0). Under theft nesting, the total seat capacity drops from 100 to 97, and the booking limits are adjusted to (b₁, b₂, b₃, b₄, b₅) = (97, 70, 9, 1, 0), which corresponds to protection levels of (y₁, y₂, y₃, y₄, y₅) = (27, 88, 96, 97, 97).

Note that the protection levels for the higher classes are preserved when a booking is accepted for a lower class. Even after the request for three seats was accepted, 27 seats are still protected for class 1 demand under both theft nesting and standard nesting. Although both theft nesting and standard nesting are in use, most systems use standard nesting. If bookings strictly occur in fare class order—that is, booking class n books before booking class n – 1—then theft nesting and standard nesting will give the same results. Some evidence indicates that when bookings do not occur in strict fare order, standard nesting tends to perform better than theft nesting.⁷ Table 8.4 shows the dynamics of bookings accepted and rejected under standard nesting and theft nesting. Note that theft nesting rejects more lower-fare bookings than standard nesting in this case, meaning that more capacity remains for later, hopefully higher-fare bookings.

8.5.4 Managing Cancellations*

Since almost 30% of airline bookings are canceled before departure, the booking management process at a passenger airline needs to update booking limits when cancellations occur. The obvious way to do this would be to treat a cancellation as an increase in capacity; that is, when a single-seat booking cancels, increase all the positive booking limits by 1. This is, in fact, the way that many reservation systems treat cancellations.

TABLE 8.4

Dynamics of booking limits and protection levels under standard nesting and theft nesting

There is a problem with this approach, however. Say we have a booking limit of three seats in the lowest booking class and we accept a booking for four seats in some higher class. Then, the lowest class would close. Now, assume that 10 seconds later the booking for four seats is canceled (perhaps the original request was an error). Under the approach described earlier, the four seats would be added back to the open booking classes and the lowest booking class would remain closed. This seems wrong. After all, we have the same available capacity and the same forecasts for future demand that we had 10 seconds before—why should our booking limits change?

Another way of viewing the issue is that, under the process described so far, the closure of a booking class is irreversible, at least until the revenue management calculates new booking limits.⁸ For this reason, this procedure is called an irreversible process. Recognizing the shortcomings of an irreversible process, some reservation systems enable a reversible process. The trick in creating a reversible process is to allow the booking limits to go below zero. In a reversible process, new bookings are subtracted from all booking limits and cancellations are added back to all booking limits. As before, a booking request is accepted only if the corresponding booking limit is greater than the number of seats requested. This means that a class with a negative booking limit is closed. With a reversible process, a cancellation can result in availability going from zero (or less than zero) to positive, opening a previously closed class.

TABLE 8.5

Dynamics of booking limits and protection levels

These two approaches to treating cancellations are illustrated in Table 8.5, which shows a sequence of booking requests and cancellations and the corresponding evolution of the booking limits for four booking classes under both approaches. The initial set of booking limits is (73, 12, 4, 0). The second entry in each row of Table 8.5 is either a booking request or a cancellation. The other entries in each row are the booking limits just prior to the event under both approaches and the action (if any) that the airline would take. Since an airline cannot reject a cancellation, no actions are associated with cancellations.

From Table 8.5 we can see that the reversible approach sometimes opens previously closed booking classes. This means that it often accepts more low-fare bookings than the irreversible approach. While this may seem appealing, the other side of the coin is that the irreversible approach saves more seats for later high-fare bookings. Generally, it is felt that the reversible approach provides higher revenue, since a booking request that immediately cancels will not change the booking limits. (For an example of this, see event numbers 2 and 3 in Table 8.5 and the resulting changes in booking limits under the two approaches.) However, which approach is used is often determined by the capabilities of its reservation system rather than by considerations of relative benefit.⁹

8.6 TACTICAL REVENUE MANAGEMENT

The job of tactical revenue management is to calculate and periodically update booking limits. These booking limits are transmitted to the reservation system, which then uses them to determine which booking requests to accept and which to reject based on the logic described in Section 8.5.

To introduce the tactical revenue management problem in its most general form, it is useful to define resources, products, and fare classes.

• Resources are units of capacity managed by a supplier. Examples of resources are a flight departure, a hotel room night, and a rental car day. Each resource is typically constrained—the flight departure has a limited number of seats, the hotel only has a limited number of rooms it can sell, and a rental car company has only so many cars it can rent out.

TABLE 8.6

Revenue management in four different industries

• Products are what customers seek to purchase. A product may require use of one or more resources. A seat on Flight 130 from St. Louis to Cleveland on Monday, June 30, is a product that uses only a single resource. A two-night stay at the Sheraton Cleveland Airport for a customer arriving on March 19 and departing on March 21 is a product that uses two resources: a room night on March 19 and a room night on March 20.

• One or more fare classes are associated with each product. Each fare class is a combination of a price and a set of restrictions on who can purchase the product and when. Different fare classes can be used to establish different virtual products, for group pricing, for regional pricing, or for combinations of all of these.

Table 8.6 shows how the resource and product approach applies in four different revenue management industries. A resource unit is the smallest unit of capacity that can be sold: a single seat for an airline, a room night for a hotel. A company may sell several different types of resource: airlines sell coach, business, and first-class seats; rental car companies offer subcompact, compact, midsize, and luxury cars; and a hotel may sell five or six different room types. While resources are the assets a company needs to manage, products are what customers actually want to buy. An airline’s products are direct and connecting flight combinations. A hotel’s products are combinations of arrival date and length of stay. A customer reserving a two-night stay starting next Wednesday is buying a different product than one reserving a two-night stay starting next Thursday.

The resource, product, and fare class terminology enables us to formulate the tactical revenue management problem in a very general manner. A seller faces a tactical revenue management problem when he:

• Manages a set of resources with fixed and perishable capacity

• Controls a portfolio of products consisting of combinations of one or more of the resources

• Can set and update the price of each product. In classical revenue management, prices are set and updated by opening and closing discrete fare classes. In other applications, prices can be set and updated directly.

The tactical revenue management problem discussed in this and the next three chapters is how a seller should choose which fare classes should be open for sale and which should be closed for each product at each moment to maximize expected total net contribution.

The fact that revenue management is based on opening and closing fare classes does not make a big difference from the customer’s point of view—customers just see the fare changing over time. The focus on opening and closing fare classes is partly due to the design of the reservation systems. It is also based on the competitive dynamics of the airline market. Specifically, most airlines believe they need to match the advertised, or headline, fares offered by their competition in key markets to drive demand for bookings. However, as bookings arrive, revenue management enables them to shape demand to the limited capacity of each flight. Revenue management supplements rather than replaces pricing.

8.6.1 Components of Tactical Revenue Management

As we have seen, the point of tactical revenue management is to determine which fare classes should be open and which should be closed for all products to maximize return from a fixed set of resources. Tactical revenue management can be decomposed into three constituent decisions.

• Capacity allocation. How many customers from different fare classes should be allowed to book?

• Network management. How should bookings be managed across a network of resources, such as an airline hub-and-spoke system or multiple-night hotel stays?

• Overbooking. How many total bookings should be accepted for a product in the face of uncertain future no-shows and cancellations?

A general view of the relative importance of each of these three decisions in different industries is shown in Table 8.7.

Capacity allocation is important whenever a company sells the same unit of constrained capacity or inventory at two or more different prices. In this case, the company needs to determine how many units to sell at the lower price(s) and how many to reserve for sale at a higher price. The major passenger airlines need to determine how many seats to sell to early-booking groups and discount customers and how many to reserve for later-booking business customers. Rental cars and hotels face similar challenges. Sporting teams need to determine how many tickets to sell as season tickets or on discounts or promotions and how many to hold to sell at full price. Air freight carriers have agreements with different customers that specify different rates—the carriers want to make sure that high-paying customers do not find themselves shut out because capacity was entirely reserved by lower-paying customers. On the other hand, resort hotels, discount carriers such as Southwest, and passenger trains may maintain only a small number of prices for the same inventory. For these companies, capacity allocation tends to be less important than other aspects of revenue management.

TABLE 8.7

Relative importance of revenue management problems among industries

* = of little importance, ** = can be important in some cases, *** = usually important, **** = of great importance, N/A = not applicable

Network management is important for companies that sell products consisting of combinations of resources. It is the single most important component of revenue management for passenger railways and for rental cars and business hotels, where managing length of stay usually has more impact on total revenue than managing rate classes. It is important for airlines that operate hub-and-spoke systems but not as important as capacity allocation or overbooking. It is of little or no importance in industries in which each product sold uses only a single resource, such as cruise lines, point-to-point airlines, sporting events, and theater.

Overbooking is important whenever bookings are allowed to cancel or not show with little or no penalty. Since airline tickets were historically sold in this way, overbooking has been a major component of airline systems. Overbooking has become somewhat less important as the fraction of nonrefundable tickets has increased, but it is still an important component of revenue management at most airlines, as it is at business hotels and rental car companies. On the other hand, most resort hotels and cruise lines do not overbook because of the high perceived cost of denying service to a customer who may have planned her entire vacation (or even honeymoon) around staying at a particular resort or departing on a particular cruise. Sporting events and theaters generally have nonrefundable tickets and usually do not overbook.

Over the course of the next three chapters, we examine each of these three constituent problems independently, but in fact they are closely entwined and, for maximum profitability, they need to be solved together.

8.6.2 Revenue Management Systems

The primary job of a revenue management system is to calculate and update the booking limits within the reservation system. Typically, a revenue management system is separate from the reservation system but linked to it, as shown in Figure 8.3. The revenue management system receives a feed of bookings and cancellations from the reservation system and, in turn, calculates booking limits that are transmitted periodically to the reservation system. The revenue management system will then send messages to update the booking limits in the global distribution systems.

Figure 8.3 Schematic overview of a typical revenue management system.

The revenue management system includes a database with information on current bookings on all flights. It also includes a database incorporating fares for all product–fare class combinations, capacities on all flight legs, and passenger variable costs by product. This information is typically extracted from other computer systems, specifically the pricing system, the scheduling system, and various accounting systems.

The forecasting module generates and updates forecasts for all product–fare class combinations for all future dates. Generally, an initial forecast is generated about one year prior to departure—the time when most airlines first allow bookings for a future flight. This forecast is updated periodically as bookings and cancellations are received over time. Typically, a forecast is updated monthly when a flight is six months or more from departure and more frequently as departure approaches. Forecasts are typically updated daily for flights that are within two weeks or so of departure.

The forecasts generated by the revenue management system are probabilistic; that is, they predict both a mean and a standard deviation for future demand. Uncertainty plays a major role in the calculation of optimal booking limits.

Example 8.4

A typical forecast generated by a revenue management system would be that the number of expected future B-Class bookings on Flight 47 from Chicago to Grand Rapids, Michigan, departing at 9:30 two weeks from today is 13 passengers with a standard deviation of 9, while the number of expected M-Class bookings is 22 passengers with a standard deviation of 10.

These forecasts are based on current and recent bookings for this flight, historic booking performance for this flight, and general demand trends for all flights. Most systems also forecast cancellations and no-shows to support overbooking calculations—a topic discussed in Chapter 11. Standard forecasting techniques are used to generate the initial forecast and to update the forecasts as bookings and cancellations are received over time.

The probabilistic forecasts are key inputs used by the optimization module to generate the booking limits. As described in Chapter 9, booking limits are calculated by estimating the economic trade-off between accepting more discount bookings now versus having more capacity available to serve future bookings. Typically, booking limits are automatically recalculated whenever the demand forecasts are updated.

Revenue management systems recommend booking limits for each product–fare class combination. Effective revenue management requires that these recommendations be continually monitored and reviewed. For example, Delta Airlines employs a group of more than 30 revenue management analysts who continually monitor the forecasts and booking limits generated by Delta’s revenue management system. Monitoring and adjusting forecasts are particularly important parts of the process. For example, the revenue management system will not know that the Super Bowl will be held in New Orleans this year. It is the job of the revenue management analyst to adjust the forecasts for January flights in the New Orleans market to reflect the additional demand generated by the Super Bowl.

8.6.3 Updating Booking Limits

Most RM companies set initial booking limits based on resource capacities, the demand forecasts for each booking class, and the economic trade-offs among the classes. Most reservation systems begin to accept bookings about a year prior to a flight departure or room rental date. At this time an airline sets initial booking limits for fare classes on those departures. The initial booking limits are loaded into the reservation system and decremented as bookings are accepted. Periodically, demand is reforecast and new booking limits are calculated based on the new forecast of demand and remaining unbooked capacity. This is known as a reoptimization or an update. Updates can be triggered in three different ways.

• Periodic updates occur at scheduled intervals. These intervals are typically long (say, monthly) when the flight is far from departure and booking requests are rare. As the flight approaches departure and the booking pace increases, periodic updates are scheduled more frequently—daily or even more often during the last two weeks prior to departure.

• Event-driven updates are triggered by events such as a booking class closing, a change in aircraft, and an unanticipated spike in demand.

• Requested updates may be launched at any time by a flight controller or revenue manager based on competitive actions, changes in fares, anticipated changes in future demand, or any other reason.

Each time booking limits are updated for a flight, the new booking limits are loaded into the reservation system and the booking management process resumes immediately starting from the new limits.

8.7 REVENUE MANAGEMENT METRICS

Before deregulation in 1983, airlines could only change their fares after a lengthy regulatory review process. Without direct control over fares, the point of most airline marketing was to fill the planes. The Civil Aeronautics Board always set fares sufficiently high that more passengers meant more money and therefore more profit. Furthermore, little or no price differentiation was allowed, so there was no scope to perform revenue management. In this environment, the airlines used load factor as their primary performance measure. Load factor is defined as the number of seats sold on a flight divided by the number of available seats. A 150-seat flight that departs with 105 passengers on board has experienced a 105/150 = 70% load factor. Load factor was a sensible performance measure in the regulated world. Prices were regulated at a level such that if an airline filled about 75% of its seats on each flight—the break-even load factor—it would meet its total costs. Each passenger above the break-even load factor represented additional profit. A marketing program was considered successful if it led to increased load factors.

Following deregulation, finding appropriate performance metrics for a flight became more complicated. In a multifare world, load factor could no longer pass muster as the most important performance metric for a flight. The reason is easy to see—in response to the rise of discount carriers, the major airlines had introduced deep discounts that induced vast amounts of leisure demand on many flights. It would be easy to maximize load factor for these flights—simply allow deep-discount customers to book all the seats. But this would be counter to the entire purpose of tactical revenue management, which is to maximize revenue by reserving seats for late-booking business passengers. Since load factor ignores revenue, it is the wrong metric for a revenue management world. In fact, it is worse than that: policies designed to maximize load factor would lead to planes filled with deep-discount passengers and massive denial of availability to business passengers—the airlines’ highest-paying customers.

To get a better view of the revenue picture, airlines began to look at yield in addition to load factor. Yield is simply revenue per passenger mile. Like load factor, it can be measured at any level from flight to market to entire airline. It was sometimes stated that the goal of managing bookings should be to increase yield—hence the term yield management, which was the original name for what is now more commonly called revenue management. Unfortunately, yield alone is a very imperfect metric since it ignores flight capacity. The strategy for increasing the yield from a flight is the exact opposite from that for increasing load factor—reject all early-booking demand and accept only late-booking, high-yield passengers. This policy would maximize yield but would be disastrous for the airline, since it would result in flights that were empty, except for a few high-paying business passengers.

What the airlines really needed was a metric that combined the capacity focus of load factor with the revenue focus of yield and recognized that the goal of revenue management is to maximize the return from resources. For an airline, these resources are seats, and the airline needs to maximize revenue from available seats. Since prices (and costs) are at least somewhat proportional to distance, it helps to normalize by distance and to measure revenue per available seat mile, or RASM. An airline that flies a 100-seat aircraft on an 1,800-mile flight (about the distance from Chicago to San Francisco) with a total revenue of $50,000 achieved an RASM of $50,000/(100 × 1,800) = $0.28 for that flight. Note that RASM is indifferent to how the revenue is distributed among passengers. Fifty passengers paying $1,000 each (a 50% load factor) would result in the same RASM as 100 passengers paying $500 each (a 100% load factor). This is the right focus from the revenue management point of view—a policy that has resulted in $51,000 from the flight is more successful than one that resulted in $50,000, even if the first policy resulted in a lower load factor.¹⁰ However, load factor can still be useful: if a flight is achieving a high RASM with a low load factor, it would be good to look for ways to find more customers for the flight even at low fares, as long as existing sales are not cannibalized. And if the load factor is extremely high for a flight, the airline might experiment with raising some of the fares to see if it can continue to fill the flight at higher average fares. If successful, these tactics would lead to increased RASM.

Revenue per available seat mile can also be expressed in terms of net yield and load factor:

RASM = net yield × load factor

Revenue per available seat mile is the leading metric currently used by airlines to measure effectiveness of pricing and revenue management. Similar approaches are also employed at other RM companies: revenue per available room night (often abbreviated REVPAR) at hotels, revenue per available rental day at rental car companies, revenue per berth for cruise lines, and revenue per available seat hour (RevPASH) for restaurants. In each case, the metric emphasizes that the goal should be to maximize the revenue—or, more accurately, contribution—throughput per unit of capacity.

The RASM metric enables comparison of revenue performance across markets and within a single market over time. Speaking broadly, flights that are achieving high RASM are generating a higher return on the company’s assets than those with low RASM. And RASM also provides a useful performance benchmark among different airlines. Those airlines achieving high RASMs are generating more revenue from the utilization of their assets than those airlines with lower RASMs.

While RASM is an important high-level metric for measuring the overall effectiveness of revenue management, it does not address the need to measure the individual effectiveness of the different components of revenue management, such as capacity allocation and overbooking. To do this, airlines typically use a portfolio of metrics in addition to RASM, which is likely to include the following.

• Voluntary and involuntary denied-boarding rates measure the fraction of passengers who show up for a flight but are denied boarding either by volunteering (voluntary) or by being chosen (involuntary). The calculation of these rates is described in Section 11.8.

• Spoilage rates measure the number of empty seats at a flight departure that could have been filled but, because some booking requests were denied, went empty. This is a measure of the effectiveness of overbooking and is described in Section 11.9.

• Dilution rates measure the extent to which early low-fare bookings take seats that could have been sold to future high-fare bookings—an example of cannibalization. Dilution rate is a measure of capacity control effectiveness, and it is described in Section 9.6.

• Revenue opportunity metrics measure the effectiveness of capacity control against a hypothetical “perfect seat allocation” alternative, as described in Section 9.6.

In each case, tracking trends is usually more important than the absolute numbers. It is often difficult to say if the revenue opportunity metric or RASM achieved by a particular flight departure is good or bad given the multitude of market and competitive factors that can influence it. But a consistent downward trend in revenue opportunity metrics or RASM usually signals a situation that requires attention.

Revenue management metrics are still evolving. By using a portfolio of metrics, a revenue management company can get a good view of its overall performance in generating revenue from its markets relative to competition and how that performance is changing over time. It can evaluate the performance of its revenue management program and determine areas for improvement. It can evaluate individual revenue managers by measuring the number of times their interventions improved overall performance versus the times when intervention degraded performance. However, a considerable amount of work remains to be done on developing improved revenue management metrics. Areas of current research include developing metrics to evaluate the performance of network management and developing metrics that separate the effects of pricing decisions from those of revenue management decisions.

8.8 INCREMENTAL COSTS AND ANCILLARY REVENUE IN REVENUE MANAGEMENT

Chapter 5 notes that the goal of pricing optimization is usually to maximize expected contribution, incorporating both the incremental cost and expected ancillary revenue associated with a customer commitment. The same is true of revenue management—decisions on which bookings to accept and reject should be made on the basis of expected contribution. Revenue management was developed in the passenger airlines industry at a time when the incremental cost of carrying a passenger was extremely low relative to the fares. The earliest generation of airline revenue management systems ignored incremental costs (or assumed they were zero) and focused on maximizing expected revenue—hence the name revenue management. As a result, much of the revenue management literature is couched in terms of maximizing revenue instead of maximizing contribution.

For airlines, this can be a reasonable assumption—while it depends on the length of the flight and other factors, the incremental cost of an additional passenger is on the order of $10 to $20 in the United States, which is typically (but not always) much lower than the fare. Nonetheless, incremental costs can be an important consideration when fares are low, and they can be quite important in other revenue management industries. Table 8.8 shows some of the most important elements of incremental cost and a subjective estimate of the importance of incremental cost in several revenue management industries. Incremental costs are relatively high—and thus important to consider—in cruise lines and container shipping. Food is a major expense for cruise lines, and the amount of food they purchase for a sailing depends on the number of passengers booked. At the other end of the spectrum, incremental costs are relatively low (and therefore generally unimportant) in theater and sporting events. It generally costs little more to stage a play in front of a full theater versus an empty one—not counting, of course, the psychic cost to the actors of playing before an empty house. In between these two extremes fall hotels, rental car companies, and passenger airlines. In all cases, it is most important to estimate and incorporate incremental costs into revenue management decisions when the incremental cost is high relative to the fare.

TABLE 8.8

Incremental cost elements in some revenue management industries

TABLE 8.9

Ancillary products and services in some revenue management industries

Ancillary revenue can also be important in revenue management. Over the past decade or so, the airlines have unbundled and begun to charge separately for services that were previously provided for free, such as luggage and onboard meals. As a result, global airline ancillary revenue grew from $32 billion to $59 billion from 2010 to 2016, representing about 9% of airline revenue. Ancillary revenue has become an important part of the business model of some airlines—it generated 45% of the revenue for low-cost Spirit Airlines in 2018.¹¹ In fact, ancillary revenue plays an important role in most revenue management industries: Table 8.9 shows the major sources of ancillary revenue associated with different revenue management industries and their relative importance in profitability. Casino hotels are an interesting extreme case: in many cases, the potential gambling revenue can substantially outweigh the room rate for a customer, which means that it can be profitable for the casino to offer rooms for very low rates (or even free), with the knowledge that they will make up any lost room revenue in gambling revenue (Kuyumcu 2002).

In Chapters 9, 10, and 11, revenue management problems are often formulated in terms of maximizing revenue rather than maximizing contribution. This is a remnant of the origin of revenue management in the airlines where incremental costs were historically much lower than fares and could be ignored in pricing and booking acceptance decisions. However, for a company to maximize profit, it should adopt revenue management policies that maximize expected contribution defined as fare (or price) plus expected ancillary profit minus incremental cost.

8.9 REVENUE MANAGEMENT IN ACTION

By most measures, revenue management has been a great success. From its origins in the 1980s, revenue management has advanced to the point that it is now de rigueur for all major passenger airlines as well as most major hotel chains, rental car companies, and cruise lines. For most major airlines, it makes no sense to ask, “What benefit are you realizing from your revenue management system?” because the alternative of returning to setting and updating booking limits manually is unthinkable. In the airlines, hotels, rental car companies, and a number of related industries, revenue management has gone from a strategic investment that provided a step change in capabilities and profitability to a fundamental capability necessary to compete—the table stakes.

It is a sign of the effectiveness of revenue management that low-cost carriers such as Southwest and EasyJet have invested in developing (or purchasing) automated revenue management systems and have built substantial revenue management departments. While they continue to compete primarily on the basis of lower costs, they have learned that revenue needs to be maximized as well to survive in the fiercely competitive passenger airline industry.

Furthermore, revenue management has expanded well beyond the airlines to other service industries that share the characteristics of constrained and immediately perishable inventory. These industries include spas and health clubs, parking garages, toll roads, self-storage units, theme parks, trucking, air cargo, passenger and freight railways, parking lots, and multifamily rental units (apartments). In each of these industries, there has been academic research into the application of revenue management, one or more companies providing revenue management software, or both.

In any case, the most important development in most revenue management industries has been the rapid growth and ultimate dominance of the internet as a channel for sales. The internet has influenced the development of revenue management and pricing optimization in four important ways:

• The internet has provided unprecedented fare visibility to consumers. Instead of relying on travel agents or 800 numbers, price-conscious shoppers surf the net for bargains on their own time, 24 hours a day, 7 days a week. This increases the pressure for airlines and others to continually manage their prices and availabilities.

• While the internet has been a disaster for the traditional travel agency distribution channels, it has led to the rise of new online intermediaries, such as Expedia, Travelocity, and Priceline. These intermediaries and their various competitors are all striving to be the dominant retailer of travel products on the internet. Naturally, the airlines are leery of the prospect of any intermediary gaining a dominant position, so a consortium of US airlines created Orbitz to be their own online retailer. For the foreseeable future, airlines will need to manage pricing and availability through a variety of online retailers as well as through their own websites and the traditional channels.

• The internet has also created new opportunities for the airlines to create and sell more “inferior” products with fewer features included. For example, Priceline allows customers to bid for travel without knowing the exact departure time or airline they are purchasing. Priceline has explicitly marketed this as an inferior product that the airlines can safely sell at a high discount without cannibalizing their mainstream products. Other airlines and hotels are experimenting with selling distressed inventory (i.e., empty capacity close to departure) at deep discounts. The need to manage these additional discount products adds further complexity to revenue management.

• Revenue management was initially shaped by the design of the global distribution systems the airlines developed in the 1960s. These systems maintain a relatively small number of fare classes for each product, and the focus of revenue management has been on which of these classes to open and close at different times. Travel agents were trained in the parlance of booking classes and product restrictions that the airlines had established. But the internet is much more suited to real-time pricing than to availability management. Online shoppers do not care how many Y-Class or M-Class seats are available; they want to know how to get from point A to point B as cheaply and conveniently as possible. While many industries—notably online retail—are able to adjust prices in a fully dynamic fashion, many revenue management industries (hotels, airlines, rental cars) are still stuck in a world where the need to manage booking classes limits their flexibility to perform fully dynamic pricing.

In any case, revenue management is here to stay. Until the flexible airplane and hotel are invented, airlines and hotels will need to maximize net contribution from fixed and perishable capacity. The capability to determine who gets to purchase at what fares will be an important driver of market success for the foreseeable future.

8.10 SUMMARY

• Revenue management is the name given to the ways by which a number of industries maximize expected contribution from their constrained resources. It is applicable to a seller who meets the following conditions:

• The seller is selling a fixed stock of perishable capacity.

• The seller offers a set of fare classes, each of which has a fixed price (at least in the short run).

• The seller can change the availability of fare classes over time.

• Customers book capacity prior to usage.

• High-fare customers book later than low-fare customers.

Revenue management is used in its purest form by passenger airlines, hotels, rental car companies, cruise lines, passenger trains, apartment rental managers, and various forms of freight companies. Many of the core concepts also apply in other industries.

• Revenue management needs to be implemented at three levels. At the strategic level, it requires identifying customer segments and creating virtual products and other ways of differentiating prices. At the booking-control level, it requires determining in real time whether or not booking requests should be accepted or rejected. At an intermediate level, tactical revenue management periodically recalculates and updates the booking limits used for booking controls.

• The primary revenue management segmentation at the passenger airlines is between early-booking, price-sensitive leisure travelers and later-booking, less price-sensitive business customers. The airlines, hotels, and rental car companies use time-based product differentiation to create products oriented toward each of these segments. In addition, revenue management companies typically have products tailored to other segments, such as groups and large corporations. Finally, they may also use channel pricing and regional pricing to further segment the market. This can lead to pricing structures of great complexity.

• In general, a revenue management company controls a set of constrained resources that can be combined to create products. Each product has a range of discrete fare classes with different associated fares. These fare classes reflect the price differentiation tactics being used by the seller and the set of products and fares the seller has established. Tactical revenue management requires updating booking limits on these fare classes over time.

• Tactical revenue management consists of three interrelated problems:

• Capacity allocation: How should fare class booking limits be set for a single-resource product?

• Network management: How should bookings for multi-resource products be controlled?

• Overbooking: How many total bookings should be allowed for a resource?

These three problems are not independent. However, they have different levels of importance within different revenue management industries.

• The main job of a revenue management system is to set initial booking limits and to perform the periodic reoptimization of the booking limits. A revenue management system includes a forecasting module that calculates probabilistic forecasts of future demand and an optimization module that uses those forecasts along with other information to determine the optimal booking limits.

• Early programs for revenue management aimed to maximize expected revenue. In an environment where incremental costs are low relative to prices, this is a reasonable assumption. However, as prices have fallen and revenue management has been adopted by new industries, it has become increasingly important to estimate incremental costs and include them in the objective function. Further, ancillary products play an important role in many revenue management industries, and their expected contribution needs to be incorporated into revenue management decisions.

• The best measure of overall revenue management performance is contribution per available unit of resource. The metric most typically used at the airlines is RASM—revenue per available seat mile. Equivalent metrics exist in other revenue management industries. For example, REVPAR—revenue per available room—is the preferred metric for revenue management performance in the hotel industry.

• Revenue management has been a highly effective tactic at the major airlines since the mid-1980s. However, the major airlines are increasingly finding themselves challenged by low-cost carriers. While the major airlines have been able to maintain higher revenues per available seat mile than the low-cost carriers, this revenue advantage has not been able to overcome the cost advantage of the low-cost carriers. It is likely that, at least in the airlines, revenue management will increasingly evolve toward dynamic pricing, with less emphasis on capacity allocation.

The next three chapters address each of the three tactical revenue management problems in more detail. Chapter 9 deals with capacity allocation, Chapter 10 with network management, and Chapter 11 with overbooking.

8.11 FURTHER READING

Robert Cross’s book Revenue Management: Hard-Core Tactics for Market Domination (1997) includes the story of PeopleExpress versus American Airlines, as well as other examples of revenue management adoption in other industries. Boyd and Bilegan 2003 discusses the role of distribution systems in the development of airline revenue management. Barnes 2012 gives an account of the history of pricing and revenue management in the airline industry, and Boyd 2007 provides a first-hand account of the early days of revenue management in different industries.

Forecasting is a key element of most revenue management systems, but I do not address the topic in this book. Chapter 9 of Talluri and van Ryzin 2004b describes some of the leading forecasting techniques used to support revenue management.

The travel distribution ecosystem is complex and continues to evolve. Barnes 2012 discusses historical developments.

The following are some references on applications of revenue management in industries other than airlines and hotels:

• Rental cars: Carroll and Grimes 1996; Geraghty and Johnson 1997; Anderson, Davison, and Rasmussen 2004

• Casino hotels: Kuyumcu 2002

• Cruise lines: Lieberman 2012

• Golf courses: Kimes 2000; Kimes and Schruben 2002

• Restaurants: Kimes 2004; Kimes, Phillips, and Summa 2012

• Spas: Kimes and Singh 2009

• Theme parks: Heo and Lee 2009

• Toll roads: Göçmen, Phillips, and van Ryzin 2015

• Trains (passenger and freight): Kraft, Srikar, and Phillips 2000; Armstrong and Meissner 2010

8.12 EXERCISE

A restaurant has 25 tables, each of which can seat up to four people. Typically, it can do three seatings per table during the dinner period from 6:00 to 10:00 p.m. The restaurant has a fairly predictable demand pattern, with Fridays and Saturdays the busiest nights and Sundays and Mondays the least busy. The restaurant is always booked to capacity on Fridays and Saturdays and sometimes on Wednesdays and Thursdays but virtually never on Sunday or Monday. It has implemented a revenue management program under which it will sometimes turn down booking requests for one or two people on Fridays and Saturdays if it believes it can fill the restaurant with bookings for larger parties of three and four people. What performance measure should it use to evaluate the effectiveness of its revenue management program?

NOTES

1. Some would argue that the global airline industry has still not fully recovered from the shock of deregulation; see, for example, Berry and Jia 2010.

2. Kama’aina is the Hawaiian word for “native.” As another example of this type of segmentation, during the off-season, Disney World offers discounts to Florida residents who can produce a driver’s license as proof of residence.

3. Lieberman 2012 discusses the “air-sea cruise line management process” in detail.

4. Airlines, hotels, and rental cars need to deal with both booking requests and availability requests. The booking request is a request for an actual booking. An availability request is a check on what fares are available for a particular product. Many availability requests are from customers who are just shopping. For simplicity, the book speaks of booking requests.

5. As a consequence of the fact that fare classes are typically labeled with a letter of the alphabet, some reservation systems limit the number of booking classes per product to a maximum of 26.

6. This is true at least when fare classes have independent demands. If a lower fare class can cannibalize demand from a higher fare class, then it may be optimal to close the higher fare class earlier than the lower one. Section 9.3 discusses this possibility.

7. Discussion of theft nesting and standard nesting can be found in Talluri and van Ryzin 2004b and Gallego and Topaloglu 2019.

8. There is one exception: most reservation systems will reopen the highest class when cancellations occur on a totally booked flight.

9. Once again, different reservation systems use slightly different approaches to update booking limits and protection levels in the face of cancellations, and there is no consensus on the best approach.

10. From an overall pricing and revenue optimization (PRO) point of view, net contribution per available seat mile (NCASM) would be an even better metric. For the example flight, assume there is a per-passenger operating cost of $50. Then, 100 passengers paying $500 each would result in an operating margin of $50,000 – 100 × $50 = $45,000, for an NCASM of $0.25, while 50 passengers paying $1,000 apiece would result in an operating margin of $50,000 – 50 × $50, for an operating margin of $47,500 and an NCASM of $0.26. The higher operating margin per available seat mile (OMASM) accurately reflects the superior operating margin of the second case, which is not reflected in RASM. However, for RM companies where the fares are much higher than the operating costs, RASM is close enough to NCASM to provide an effective metric.

11. Ancillary revenue plays a particularly important role for low-cost airlines and, in many cases, is a core element of their business model.