Your Plane Ticket Is A Prediction
You might think an airline ticket has a price. It doesn't. It has a whole pile of possible prices, and behind the scenes, computers are constantly trying to figure out which fares should still be available. And strangely enough, much of that started before most Americans had even seen a computer.
It Started With A Really Annoying Problem
Airlines had one unusual problem: their product expired incredibly quickly. If a plane left New York at 2:00 with 12 empty seats, those seats weren't going into a warehouse to be sold tomorrow. At 2:01, their value had dropped to exactly zero.
Sjoberg Bildbyra, Getty Images
So Airlines Started Studying Us
That meant airlines desperately wanted to know how many passengers would book, how many would cancel, how many wouldn't show up and how many might appear at the last minute willing to pay a higher fare. Long before anyone called it predictive analytics, airlines were already turning passenger behavior into a math problem.
At First, Reservations Were Surprisingly Primitive
In the early postwar years, airline reservations could still involve handwritten cards and physical filing systems. As passenger numbers grew, keeping track of which seats had actually been sold became increasingly difficult. Flights could end up overbooked, underbooked or simply tangled in clerical errors.
Abdiel Hernandez Villegas, Pexels
Then Two Guys Named Smith Sat Together On A Plane
In 1953, American Airlines president C.R. Smith happened to meet IBM salesman R. Blair Smith on a cross-country flight. They began talking about American's reservation problem. That chance conversation eventually helped produce one of the most important computer systems the travel industry had ever seen.
American Airlines And IBM Built SABRE
The result was SABRE, a massive computerized reservation network developed by American and IBM. Instead of relying on disconnected records, agents could access seat availability through terminals linked to central computers. By the mid-60s, SABRE was reportedly processing around 7,500 reservations every hour.
But Keeping Track Of Seats Was Only The Beginning
Once airlines had enormous amounts of booking information sitting inside computers, another possibility appeared. They didn't have to use the data simply to know whether seat 14A was available. They could use past passenger behavior to make educated guesses about what future passengers were going to do.
Boris Babanov / Борис Бабанов, Wikimedia Commons
The No-Show Problem Was Worth A Fortune
Airlines knew some passengers with reservations wouldn't arrive. Leave a seat empty for them and the airline lost revenue. Sell that seat twice and there was a chance both passengers would appear. Researchers began developing mathematical models to balance those two risks as early as the 1950s.
Yes, This Is Where Overbooking Comes From
Overbooking wasn't simply an airline crossing its fingers and selling too many tickets. Increasingly, it became a calculated prediction. Airlines could study historical no-show behavior and estimate how many extra reservations a flight could safely accept while keeping the risk of having too many passengers manageable.
Then Someone Asked A Much More Profitable Question
By the early 70s, researchers were moving beyond simply asking, “How do we fill the airplane?” British Overseas Airways Corporation researcher Ken Littlewood tackled a more interesting question: how could an airline decide which fares to accept if the goal was maximizing revenue rather than passenger numbers? BOAC would later merge with British European Airways to form British Airways in 1974.
Jon Proctor, Wikimedia Commons
Filling Every Seat Wasn't Necessarily Best
Imagine an airline sells every seat months early for $100. Sounds great. Except 20 business travelers might have happily paid $400 a few days before departure. The airline had filled its plane perfectly...and potentially left thousands of dollars sitting on the table.
San Francisco Chronicle/Hearst Newspapers, Getty Images
So Some Seats Had To Be Protected
The solution was surprisingly clever. An airline could sell discounted seats to attract price-conscious travelers, but stop selling those cheap seats once the computer determined it should protect the remaining inventory for passengers who might arrive later and pay considerably more.
Then Washington Changed Everything
For decades, U.S. airlines operated in a heavily regulated environment where the government controlled significant aspects of routes and fares. The Airline Deregulation Act of 1978 substantially removed those controls. Suddenly airlines had far more freedom to compete on price, routes and service.
And The Price War Was On
New competitors could enter with dramatically different business models and cheaper fares. Established airlines now faced a nasty problem. They needed low prices to compete for bargain hunters, but they couldn't afford to let every passenger who would have paid full fare buy the cheap ticket instead.
American Airlines Became Obsessed With The Problem
American had actually begun researching ways to manage revenue from its seat inventory back in the early 60s. But deregulation made the work vastly more important. Its operations-research teams increasingly attacked the problem using forecasting, optimization and the enormous booking database made possible by SABRE.
Roger Smith, FSA/OWI Collection, Wikimedia Commons
Enter Robert Crandall
American executive Robert Crandall became a major champion of what came to be known as yield management. The basic idea was simple enough to explain and incredibly complicated to execute: sell every seat for as much money as the airline could realistically get for it.
The Wings Club, Wikimedia Commons
The Computer Didn't Need To Know Your Name
This wasn't originally about a computer deciding that Bob from Cleveland personally looked rich enough to pay another $80. The models were forecasting demand: how people on particular flights tended to book, which fare classes filled first and how much high-paying demand might still appear.
Michael Ball, Wikimedia Commons
Every Flight Became Its Own Little Prediction Market
A Tuesday afternoon flight to Orlando could behave differently from a Monday morning flight to Chicago. Booking patterns changed by route, date, season and customer mix. Historical data gave airlines a way to estimate how demand for each future flight was likely to develop.
Charles O'Rear, Wikimedia Commons
Cheap Seats Became A Controlled Substance
Airlines could still advertise eye-catching fares. The trick was controlling how many seats were actually available at those fares. Once enough discounted seats were sold, the cheap inventory could disappear while more expensive fare categories remained available on the exact same airplane.
People Express Put The System To The Test
In the early 80s, low-fare People Express became a serious competitive threat. American couldn't simply abandon higher fares and match a discount carrier across every seat. Instead, its revenue-management approach helped determine where cheaper fares could be offered without unnecessarily giving away seats to passengers likely to pay more.
American Unleashed Its Ultimate Super Savers
In January 1985, American introduced deeply discounted Ultimate Super Saver fares, including in markets where it competed with People Express. Behind those cheap tickets was something its rival couldn't easily see: American had sophisticated systems controlling exactly how much discounted inventory it released.
The Algorithm Had Become A Competitive Advantage
American wasn't merely charging a low price. It was trying to predict where that low price would produce extra passengers instead of replacing customers who would have paid more. That distinction turned the airline's giant reservation database into something much more powerful than a glorified electronic seating chart.
Hunter Desportes, Wikimedia Commons
And It Was Making Serious Money
By 1992, American Airlines researchers reported that their yield-management work had produced an estimated $1.4 billion in quantifiable benefits during the previous three years. They expected it to continue contributing more than $500 million in revenue annually. Suddenly, everyone had a reason to pay attention.
Other Industries Wanted In
American's analytics operation eventually began taking similar technology outside aviation. Its teams worked with rail companies and introduced revenue-management ideas into hotels, rental cars and cruises. The logic translated remarkably well anywhere a company had limited inventory that disappeared if it wasn't sold in time.
Hotels Had The Exact Same Problem
An empty hotel room tonight is a lot like an empty airline seat at takeoff. Tomorrow morning, nobody can go back and sell last night's unused room. That made hotels a particularly natural home for the same combination of demand forecasting, controlled discounts and changing availability.
The Airline Industry Had Helped Build Something Much Bigger
What started with reservations, empty seats and no-show passengers eventually became part of a much larger way of doing business. Companies could collect enormous amounts of historical behavior, find patterns inside it and use those patterns to make increasingly sophisticated predictions about future demand.
Euthman Ed Uthman, Wikimedia Commons
And Now You See It Everywhere
Hotels, rental cars, cruises, trains and countless other businesses use versions of the same basic idea today. Demand changes. Inventory is limited. Customers behave in patterns. And computers are exceptionally good at finding those patterns faster than any human with a clipboard ever could.
So About That $187 Plane Ticket...
When you see a fare disappear or return later at a different price, there may not be one employee somewhere deciding to squeeze more money out of you. You're seeing the descendant of a system airlines spent decades building to answer one enormously profitable question: what is this seat likely to be worth before the plane leaves?
Airlines Didn't Set Out To Predict People
They were trying to solve something much less futuristic: empty airplane seats were costing them money. But once they started collecting enough information about how passengers behaved, they discovered something far more valuable. Past behavior could help them predict future demand...and those predictions could be worth billions.
You Might Also Like:
Retirees Shared The Small Towns They'd Move To In A Heartbeat—Would You Live In Any Of Them?
