Two passengers sit side by side in economy, same flight, same legroom, same sandwich. One paid several times what the other did. Neither was cheated, and neither was tracked by a cookie. They simply bought different products that happen to occupy identical seats.

An airline seat is the most perishable thing in retail. The moment the door closes, every empty seat is worth nothing, and no airline can store it for tomorrow. So since the 1980s airlines have built some of the most sophisticated pricing machinery in any industry to sell each seat to the right customer at the highest price that customer will pay.

Here is how airlines price tickets: the fare classes and buckets behind every booking, the American Airlines system that started the revolution, the low-cost airline it destroyed, and why clearing your browser cookies will not save you money.

One seat, many products

An airline does not sell a seat. It sells fares, each with its own rules. A cheap fare might demand booking weeks ahead, forbid changes and refunds, or require a Saturday night away. An expensive fare on the same flight can be bought at the last minute and changed or refunded freely.

Those rules separate travellers by how much they need the flexibility. A business traveller who books two days ahead and might change plans cannot use the cheap fare, so they pay more. A holidaymaker booking months ahead can.

Each fare maps onto a booking class, also called a reservation booking designator or RBD: a single letter code such as Y, B, M or Q in economy. The letters vary between airlines, but the logic is the same everywhere. Each letter is a bucket of seats sold at a particular fare level.

Wendover Productions explains how airlines price flights, from fare classes to the forecasting behind them.

Fare buckets and nesting

The revenue management system's job is to decide how many seats to make available in each bucket, and to keep adjusting that number until departure. It forecasts how many passengers will want each fare and protects seats for the late, high-paying customers it expects.

The buckets are nested. A higher fare class can always sell a seat that a cheaper class could have used, but not the other way round. As a flight fills, or as demand runs ahead of forecast, the cheapest buckets close first.

This is why prices seem to jump. The fare you saw yesterday has not been raised; its bucket has sold out or been closed, and the cheapest open fare is now the next letter up. If demand drops, the system can reopen a lower bucket, and the price falls again.

Overbooking: selling seats that do not exist

Some passengers with tickets never show up. Airlines forecast these no-shows for each flight and sell more tickets than there are seats, so the aircraft still leaves full. When more passengers turn up than expected, someone has to stay behind.

Airlines first ask for volunteers. If nobody accepts, passengers are bumped involuntarily and compensation is set by law. In the United States, the Department of Transportation requires 200% of the one-way fare, capped at $1,075, for domestic arrival delays of one to two hours, and 400%, capped at $2,150, beyond that.

In the European Union, Regulation 261/2004 sets fixed compensation of €250, €400 or €600 depending on flight distance. Overbooking was one of the three problems, alongside discount allocation and traffic management, that American Airlines' pioneering system was built to solve.

American Airlines and the birth of yield management

The modern system was born in a price war. After US deregulation in 1978, new airlines were free to enter markets and set their own fares. The most famous was People Express, which began flying from Newark on 30 April 1981 and by 1983 was selling seats to London for $149 each way.

American Airlines could not match People Express on cost. Instead, under Robert Crandall, often credited with popularising the term yield management, it decided to match it on price only where that made sense. On 17 January 1985 American launched Ultimate Super Saver fares, heavily restricted discounts that had to be bought well ahead and came with steep cancellation penalties.

The trick was capacity control. American's computers forecast, flight by flight, how many seats would otherwise fly empty and sold only those at the low price. A business traveller booking late still paid full fare.

People Express, with its simple low fares, had no equivalent tool to respond. The airline had other problems too, including rapid expansion, but its founder blamed American's discounts for the sudden turn in its fortunes.

People Express was sold to Texas Air in 1986 and ceased to exist as a separate airline on 1 February 1987, when it was folded into Continental Airlines. American, meanwhile, kept refining its system, known as DINAMO, which became fully operational in 1988.

The results were documented in the academic journal Interfaces in 1992. American estimated that yield management had produced quantifiable benefits of $1.4 billion over three years and expected more than $500 million a year to continue. The work won the Franz Edelman Award for operations research in 1991.

From buckets to continuous pricing

For decades the system was limited by its own alphabet. A fare had to fit one of a fixed set of booking classes, traditionally no more than 26, and every price had to be published in advance. Delta, for example, says its fares are filed through the industry clearing house ATPCO multiple times a day.

Newer technology is loosening those limits. Under IATA's New Distribution Capability, or NDC, airlines can build offers in their own systems and send them directly to travel sellers. Lufthansa Group describes continuous pricing as an evolution of dynamic pricing that allows an airline to provide indefinite price points, instead of jumping from one bucket to the next.

Artificial intelligence has become the newest flashpoint. In 2025 Delta's use of AI in pricing drew letters from US senators worried about individual surveillance pricing. The airline replied that its prices are set for markets, not people.

Delta said its AI tools recommend adjustments in both directions and that its prices are dictated by market dynamics and competition. Every customer, it said, sees the same fares based on objective criteria such as route, advance purchase, length of stay and refundability.

That is the key distinction. Dynamic pricing changes the price of a flight over time as demand changes. Personalised pricing would charge two people different prices for the same product at the same moment, which is exactly what Delta denied doing.

The cookie myth

The most persistent travel tip on the internet says airlines and booking sites track your searches and raise the price when you come back, so you should search in incognito mode. Regulators have looked for evidence of this.

In 2013 the French data protection authority CNIL and the consumer watchdog DGCCRF opened a joint investigation into whether travel websites used IP tracking to raise prices for repeat visitors. Their preliminary results, published in January 2014, found no prices that varied according to the user's IP address.

What they did find were the ordinary drivers of airline pricing: the number of seats left, the date of purchase, booking fees that varied by time of day and the comparison site a customer arrived from. A price that rises between two searches usually means a bucket closed, not that someone was watching you.

CNBC Explains walks through how airlines price tickets and why fares change so often.

Why the system will not go away

Fare classes, buckets and overbooking can feel like a game rigged against the passenger. But they are also why cheap seats exist at all: the flexible traveller pays more so that seats which would otherwise fly empty can be sold cheaply.

The story of People Express shows the other side: a simple, cheap airline that was outsmarted on price rather than undercut on cost.

Andrew's Aviation tells the rise and fall of People Express, the low-cost airline undone by American's yield management.

Understanding how airlines price tickets will not guarantee you a bargain. It does explain why the cheapest fare is usually the one bought early, with the most restrictions, on the least popular flight.

Sources: Smith, Leimkuhler and Darrow, 'Yield Management at American Airlines', Interfaces (1992) and INFORMS Franz Edelman Award citation, RWTH Aachen (Käte Hamburger Kolleg) on the history of dynamic airline pricing, MIT OpenCourseWare (The Analytics Edge), Wikipedia (Yield management; People Express Airlines), Delta Air Lines News Hub (letter of 31 July 2025), Lufthansa Group 'Dynamic Offers' (2021, via IATA), US Department of Transportation, EU Regulation 261/2004, Hogan Lovells on the CNIL/DGCCRF IP-tracking inquiry