Revenue Management, Then and Now
From the basement to the boardroom: The analytics revolution is returning us to the Marrakech markets of old
When you purchase services online such as an airline flight or hotel room, you enter into a fascinating and somewhat frustrating world of dynamic pricing. While the price fluctuations from moment to moment or consumer to consumer seem random, there are mathematical methods to the madness. The sophisticated modelling that drives these systems allows firms to maximize profits from their limited capacity by charging different prices to different customers at different times. This can look like “profit squeezing” but it can allow customers with smaller wallets to get a service or product that they would otherwise have to forgo if the price was fixed. Jeff McGill, Smith School of Business professor in management science, has been studying revenue management for more than 25 years. In this conversation with Smith Business Insight, he shares his unique perspective on the practice.
How did revenue management get started?
The idea of different prices for different customers goes back to the origins of commerce but the airlines were the first to use revenue management, or yield management, in its modern form. It started with BOAC (now British Airways) in the early 1970s, but other airlines quickly caught on. For example, American Airlines became heavily involved in the 1980s and, by 1992, reported a $1.4 billion increase in revenues over three years from using their system. Revenue Management (RM) systems became strategic assets for airlines, and details of the systems were considered trade secrets. In fact, I was involved as an expert witness in a lawsuit by one airline against a smaller rival for alleged raiding of employees who were experts in RM systems. It was an interesting sideshow.
These systems were changing the profitability profiles of pretty large corporations, so it grew from there. Similar applications have spread into the accommodation industry, cruise lines, passenger rail, broadcasting, rental cars, and the list goes on. These systems are really early examples of “analytics” applications that are currently generating so much excitement.
Why were airlines the first to get into this?
Revenue management used math and computer algorithms that had been around since the 1950s, so why did it take hold first in the airlines? I have two guesses. They may have been more technologically receptive because of the technical backgrounds of the people in airlines; for example, pilots and engineers who moved into management. The other is they already had good computer and reservation systems. For safety and security reasons, their reservation systems had to be pretty solid. They had to keep track of the passenger name records, so they had something to build on.
Revenue management first took hold in service industries with very high fixed costs and low marginal costs per customer. In the airline’s case, they had to fly their planes on scheduled routes under regulations, so they were basically this multi-billion dollar industry with near zero marginal costs. The costs of flying an additional passenger from New York to Los Angeles was, like, $20 — some paperwork, a few teaspoons of jet fuel, and a plastic sandwich. That gave them a tremendous range of prices that they could still make money on.
We're seeing dynamic pricing in oil, banks, and insurance companies. How do you price insurance policies? How do you price mortgages?
This area of dynamic pricing is still booming. It already has been used somewhat in financial services and there's a great deal of potential growth in that area. We're seeing it in oil, banks, and insurance companies. How do you price insurance policies? How do you price mortgages?
How do consumers feel about this?
The ethical side is interesting. It has the emotional grab of discrimination, of one person paying one price and someone else paying a different price. But if you look at it from the point of view of economic utility, they in fact might be paying the same amount in perceived value. Moreover, travellers who book on short notice may pay more than they used to, but they're actually getting a different product because they're able to buy it closer to departure. I say “may pay more” because if there are many seats unfilled near departure, the price may actually be lower — a game of probabilities for the customer and the firm. It's the same seat when you sit on the aircraft, but in fact it was a different product bundle, and that allows airlines to charge lower prices for people who book earlier, or when there is excess capacity, or for people who fall in different market segments, like student travellers.
Where are these systems going?
Think about how business was done centuries ago. A customer walked into a carpet shop in Marrakech and started haggling with the owner. Three hours later, they came out with a carpet at a good price for them. Then the next customer came in and got a different price. In contrast, for most of the last century with mass production and mass marketing, we’ve seen frozen prices, one price fits all.
The growth of the World Wide Web from the early 1990s has created the potential for second-by-second monitoring of customers, and we are moving back towards that old model where your market segment is the individual customer. You're haggling with individual customers again, but with a very different set of tools.
We have highly technical systems that actually are right in the ballpark of what management does and can’t be hived off
On the research front, there are many interesting new problems. How can you incorporate into revenue management systems mathematical models of how individual consumers choose products and services at different prices? How might Big Data and related analytics methods influence RM developments? How do you engineer RM systems that can quickly adapt to shifting corporate and marketing strategies?
Customers are not passive. When they become aware of how prices are changing over time, they can become strategic; for example, timing their purchases until prices are low or ‘hoarding’ discounted consumables. Is it possible to actually incorporate strategic consumer responses into RM systems?
Are revenue management systems now considered strategic assets by large firms?
For matters that used to be in the “basement” — operational or tactical decisions — if you automate these processes, they suddenly become strategic. The ability to execute tens of thousands of small decisions with greater precision becomes a strategic matter. People high up need to pay attention because if they don’t get it right, they may not be in business.
The other side is if you’re making tens of thousands of decisions, you’re spending a lot of money, and those cheques are signed at very high levels. It is incumbent on senior management to understand what is actually going on with what used to be considered technical stuff and perhaps beneath their dignity.
There’s a bit of a communications and credibility challenge. You have people who were formerly systems analysts and programmers suddenly designing systems that have strategic importance. They have to convince senior management that they are doing it correctly or that the work should be done at all. This is certainly reflected in modern analytics. The most common complaint of senior management is that they can’t communicate with analysts, and vice versa.
The divide between technical expertise and management expertise has been around for a long time. In days gone by, technical specialists could operate in isolation. But what’s happening here is that we have highly technical systems that actually are right in the ballpark of what management does. They can’t be hived off.
— Interview by Alan Morantz