In an environment where the whole basis for RM is optimal allocation of fixed capacity, is there another way? There just might be, writes Tom Bacon

RM forecasts demand, no-shows and allocates seats. But what if the forecasts are wrong and they discover the error after seats are already sold? Or what if the number of seats available shrinks suddenly due a change in equipment? Alaska Airlines and Volantio calls these scenarios ‘post-sale RM’ or ‘variable capacity’.

Questions that arise include:

• What if there are fewer no-shows than expected?
• What if demand is much more than expected and you are now spilling high fare passengers?
• What if supply changes with an aircraft downgrade or upgrade, suddenly changing the whole demand profile?

Despite the enhancements to the forecasting process introduced over the last decade, demand forecasts by fare range remain highly uncertain – the variance for demand around an average forecast is typically huge. Consequently, it is not uncommon that two to three weeks before departure, the demand forecast has changed dramatically from the forecast more than a month before departure, when there were few sales already on the books. Frequently, in hindsight, an analyst will wish he had ten more seats on the plane given the increased demand on one flight – and ten fewer seats on another flight.

Despite the enhancements to the forecasting process introduced over the last decade, demand forecasts by fare range remain highly uncertain

Revenue management is touted for how dynamic it is. Every night it updates forecasts based on the most recent information and re-optimizes inventory across thousands of flights. But it is focused on each flight independently and doesn’t consider such inter-flight opportunities – except in a crisis situation when dealing with oversales at the airport (which is handled by front-line airport employees, not the RM analysts). The result is higher cost of oversales, more customer disruption, and less revenue than if flights are treated more as a system earlier in the booking process.

The solution is ‘dynamic guest movements’. We are all familiar with moving passengers in a denied boarding situation – often it is cumbersome and costly, including an auction process that seems to escalate compensation levels especially at peak times. However, this is just one application. Guests can be moved much more efficiently – at relatively low compensation levels – days or even weeks before the departure date. This may be initiated in response to a known aircraft downgrade or in response to unexpectedly high demand for a certain flight, effectively increasing capacity on a flight a few weeks before departure! 

Critically, the process identifies ‘movable’ passengers before the flight. When oversales is handled at the airport, there is typically an auction process with compensation escalating to attract more flexible travellers. By identifying flexible passengers ahead of time, however, the level of compensation does not play as important a role. In fact, airline tech firm Volantio has found that, in the days or weeks before the flight, the level of compensation is no longer as predictive of customer willingness to move (WTM). Alaska Airlines and Volantio instead use machine learning to set compensation levels; the primary driver of WTM is the number of passengers on the passenger name record (PNR) – by the way, a single traveller is most willing to move. Other factors may include outbound versus inbound, fare paid, age, and flight day-of-week and time-of-day.

‘Variable capacity’ has been a dream for revenue managers: the whole basis for RM is optimal allocation of ‘fixed capacity’. When demand varies from forecast, being able to add or delete capacity has not historically been a tool RM analysts can take advantage of. But ‘dynamic guest movements’ offers a tool for effectively doing just that. 

Tom Bacon has been in the business 25 years, as an airline veteran and now industry consultant in revenue optimisation. He leads audit teams for airline commercial activities including revenue management, scheduling and fleet planning. Questions? Email Tom or visit his website