American airlines deliberately sell more tickets than there are seats on some flights. The practice is a response to no-shows and to the fact that an empty seat has no value after pushback.
A departed seat is worthless
An airline's product expires at the moment of departure. Unlike unsold merchandise, an empty seat cannot be stored, discounted later, or sold to anyone else.
Almost all the cost of the flight is incurred regardless of how full it is. Fuel, crew, gate fees and maintenance change only marginally with an extra passenger.
That makes the marginal revenue of filling a seat close to pure contribution, which is why carriers work aggressively to avoid departing with empty rows.
No-shows are predictable in aggregate
Some ticketed passengers never appear. Connections fail, plans change, and refundable and flexible fares are cancelled or altered at the last minute.
Individually this is unpredictable, but across thousands of departures it forms a stable pattern that varies by route, time of day, day of week and fare mix.
Revenue management systems forecast that pattern for each flight and authorize a booking level above capacity accordingly, adjusting as departure approaches.
Fare rules make the problem larger
Business-oriented fares are sold with flexibility because travelers pay a premium for it. Flexibility means a higher rate of last-minute change.
A flight carrying many such tickets is more likely to have absent passengers, so the authorized overbooking level on that flight is higher.
Leisure routes sold mostly on restrictive nonrefundable fares behave differently, with fewer no-shows and correspondingly less overbooking applied.
Aircraft swaps create a separate problem entirely. When a smaller plane is substituted for a mechanical reason, a flight sold exactly to capacity becomes oversold in an instant.
Denied boarding follows federal rules
When more passengers appear than seats exist, carriers first seek volunteers, typically offering travel credit or vouchers that rise until someone accepts.
If volunteers are insufficient, passengers may be denied boarding involuntarily under criteria the airline must publish, and federal rules set compensation based on delay length and fare.
Those rules also govern what must be disclosed at the gate. The compensation framework is a matter of federal regulation, not airline discretion, though details change over time.
The economics still favor the practice
Compensation is expensive but happens on a small fraction of departures, while empty seats would occur on many more if booking stopped exactly at capacity.
Better forecasting has narrowed the error band, and carriers now buy volunteers earlier through prompts during online check-in rather than at the gate.
Some airlines have reduced overbooking for reputational reasons after high-profile incidents, choosing a modest revenue loss over the cost of a public confrontation.