Airline Economics · Loyalty

Net award revenue

When is an award seat worth it to the airline?

When a member redeems points, the loyalty program pays the airline for the seat. The seat can still cost more than that: the member might have bought a ticket anyway, or taken a seat a paying passenger needed. This simulator replays the same 1,500 flights with and without award passengers and compares the two, following an MIT thesis on the question, then tries a forecaster that sees buy-down.

Members
Simulating…

Where the money goes

Average flight, with award passengers vs. without

Net effect, piece by piece

Dollars per flight. Blue adds revenue, orange takes it away.

Passengers by class

Average per flight. The award passengers come mostly out of the cheapest fares.

The thesis experiment

The thesis tried five payments, from half the lowest fare to one and a half times it, each with RM valuing award seats at half, the same or one and a half times the payment. Click a cell to see it above.

This simulator

The thesis

Extension: a class-free forecaster that sees buy-down

The RM above forecasts by class, as if each customer only ever bought their own class. A class-free forecaster predicts how many people would pay at least each fare, so RM knows that opening a cheap fare lets everyone above it buy down. It can also see members who would redeem instead of paying.

Members buying down to an award

Class-based vs. class-free

Class-free: the thesis grid again

Change in total revenue vs. the same flights without awards, under the class-free RM. Click a cell to see it.

How it works

The flight. 150 seats and six fares, from $640 (Y) to $180 (K), the cheaper ones with advance purchase rules (21 days for K, 14 for Q, 7 for H, 3 for M). Leisure travelers book early and would pay $240 at the median, business travelers book late and would pay $540; each buys the cheapest open fare they are willing to pay. Demand varies by about ±25% between departures, so some flights fill up and others don't.

Members. A share of leisure travelers have enough points. Redeeming isn't free in their eyes (the points are worth something), so it feels like paying a multiple of the lowest fare. A member takes an award when one is open and it feels cheaper than the cheapest fare they would pay. Awards close 7 days before departure, as in the thesis.

RM. The award is one more class, worth the value RM is given. RM forecasts demand by class, award requests included, then either sets bid prices with a dynamic program (a class is open while its value covers the price of the next seat) or EMSR-b protection levels at the start of each booking period.

Class-free (extension). RM forecasts, for each booking period, how many customers would pay at least each fare and, when members' buy-down is seen, how many would redeem rather than pay if awards were open. A dynamic program then picks, for every moment and number of seats left, the cheapest fare to offer and whether to offer awards. In practice, fare adjustment approximates this by running bid prices on the marginal revenue of each fare (Fiig, Isler, Hopperstad and Belobaba, 2010).

Measuring. The same customers, with the same arrival times, budgets and points, book the flight twice: once with no award class, once with it. Comparing them passenger by passenger gives the thesis's three effects:

Award revenue = award passengers × what the program pays per seat Displaced = fares lost: members who would have paid but redeemed, and paying passengers who no longer got a seat Sell-up = change in the fare paid by passengers who flew both times Net = award revenue − displaced + sell-up

The thesis measured them on class averages across a network; on one flight each class has a single fare, so they are measured passenger by passenger here.

Limitations

Source

This is a personal project and isn't affiliated with any airline. For the business side of award seats, read Loyalty 101, chapter 10: Where Loyalty Meets RM.