Where the buy-back solution originated: revisiting the reverse auction model for refunds


6 years ago, ChangeYourFlight.com (CYF) launched with the objective of creating a marketplace for airlines to retrieve unused flight coupons in exchange for a voucher for future travel.

Since its launch, the model has issued over 250k vouchers for future travel across 3x airlines, ultimately enabling over 20k new segments to be booked with an average value of 5x the originally unused coupons.

With post-sale optimization now becoming a core part of many airlines strategy, we thought we’d take a closer look at the earliest example of the concept.

What’s more, if you are interested in the maths behind the original model, we’ve taken the decision to share the paper that supported the original model with any airline who’d like to have it. Our view is that this can be effective anywhere and the original architect of the algorithm, Kaylan Talluri (current professor of RM at LSE), is happy for his work to be made public. We see the idea of CYF as being a force for good in the industry and we’d rather more airlines had it. After all, the concept is one thing, but the implementation is quite another.

Original Problem
Non-refundable tickets are a customer service nightmare and associated no-shows often exacerbate revenue spillage and denied boarding compensation.
Customers don’t like non-refundable tickets

Airline pricing places a premium on flexibility and refundability: the fewer passengers change their reservations the better airlines can manage their inventory. This is now an accepted facet of air travel and the overwhelming majority of tactical pricing, the fares we see advertised, come restricted to just the dates and times we have originally booked.

cyf tweet.png

Accepted and understood, but not liked and often disputed. A recent survey of our airline partners customer care teams cited fare conditions as being in the top 3 of most common sources of customer friction.

Spilled revenue

The cost to airlines of spilled revenue, in instances where they are unable to accommodate high-value customers at a later stage of the booking curve, can be cripplingly high. Airlines use overbooking models to overcome the cost of missed opportunities. However, the risks associated with overbooking can be counter-effective.

The reverse auction model for airline refunds
What if the airline could determine the potential resale value of customers unwanted seat and offer them a partial refund based on that mathematical probability?

Through an auction (the most economically sensible method to determining value), customers who’s plans had changed and who were left with an unused reservation could submit their partial refund request with the probability and moment of acceptance made clear at the outset.

The algorithm behind the program was designed to take into account 2x core dynamic factors to determine the likelihood of re-utilization of the seat ‘liberated’:

  • Recapture probability: how likely is it that the seat liberated will be resold
  • Recapture value: what are the current market prices vs. the original fare paid by the customer
  • By dynamically linking these factors, the reverse auction is continually optimized so improve the return on investment for the airline. A partial refund provided on a flight with a final take of seat load factor of less than 100% can be judged to be unsuccessful when looked at in isolation, however when considering the prevailing recapture value, typically 5x.

    The results: measuring success

    In the 6 years that CYF has been in place, many thousands of customers have submitted requests to refund.

    What this has translated into is over 20,000 seats recycled, equivalent to an extra 111 full A320’s.

    How this model has succeeded in iterating into post-sale optimization

    CYF.com provided the understanding that there is consumer demand for an ex-post marketplace. That understanding paved the way for proactive optimization models to be created that dynamically take on the challenge of spilled revenue whilst providing an opportunity for airlines to negate the cost of overbooking.

    What’s needed to bring a reverse auction for refunds to life?

    We are happy to pass on the theory behind the model. By utilizing it, airlines can deliver a model that brings about a customer-centric product that generates incremental capacity and further opportunity for high-yield sales.

    Implementing a basic version requires nothing more than cohesion around the concept and alignment around the values. Both of which we are happy to share our experience in. Moving beyond that, with the associated data, the model continues to learn, ensuring that there is future value.

    You can start by downloading the original Kaylan Talluri white paper above. Our team of partner success managers will happily talk you through it, too.

    The iterations in practice

    The iterative developments have been borne out of 6 year journey: a continued sophistication of both the technology and the data science behind the results.

    Seat Resale

    The first proactive approach to post-sale optimization.

    Taking the reactive CYF approach and instead monitoring flights to determine in advance the requirement to generate capacity meant that the offers were matched immediately to the opportunity and the customer


    Where Seat Resale proactively generated capacity, Exchange proactively generates sales. Understanding the potential ex-post marketplace allowed us in collaboration with customers such as LATAM to take the proactive concept further. Exchange, using AI and conscious analytics understands the potential for oversell and re-sale from each flight segment. By interacting with forecasting systems such as PROS and AirRM, Exchange measures demand and the potential willingness to move of existing customers: together, that algorithm aligns the potential reward from oversale and the willingness to move from existing customers to determine the correct AU and the reward or incentive required for customers to move.

    What to do next

    If you’d like to explore how a reverse auction model can generate extra capacity and provide an easy to implement customer-centric product, click here for the original white paper.

    If you’d like to understand from experience what a credible, long-lasting and profitable buy-back solution looks like, feel free to get in touch. We are more than happy to share how this technology can be implemented and discuss the best approach for your airline.