Yield management, also called Revenue Management, is a pricing methodology used by all airlines that has spread to other business areas. Wikipedia defines it as: “Yield management is a variable pricing strategy, based on understanding, anticipating and influencing consumer behavior in order to maximize revenue or profits from a fixed, perishable resource.”
The interesting point is “perishable resource” meaning when you don’t use it you lose the potential revenue. Seats in a specific flight are a natural example. Once the flight takes off the opportunity to sell the empty seat is gone.
Isn’t it the same for the capacity of ALL RESOURCES?
A machine that is idle today – that part of its capacity is lost forever. We in TOC are well aware that capacity lost on the bottleneck means lost Throughput. Yield Management looks on seats in a flight as a potential constraint. This is a valid view of looking on the micro-capacity, like a flight or one day in a hotel, as an independent opportunity to fill the whole batch and gain more sales. We can compare the situation to using an oven in production wishing to fully utilize its capacity by putting several different products in the oven.
Most travelers need the flight at specific dates and time in the day. This provides an opportunity to increase the price when the demand seems relatively high, because changing the date of the flight is not a proper alternative for many, and thus they are forced to pay higher price for exactly the same service. On the other hand, when the expected demand is low then offering very low price could steal customers from other airlines and earn revenues that would be otherwise lost.
Yield Management is an exploitation scheme of every instance of the capacity of a critical resource. At the same time Yield Management exploits the market demand by reducing the price when low demand is expected. This basic objective of Yield Management is definitely in line with the five focusing steps!
The means of exploitation of the constraints are different than the regular TOC approach. TOC would first look to exploit the aircrafts, assuming they are the constraint, searching for ways to make more flights yielding overall more T. Yield Management looks on every micro-capacity of the constraint, every single flight, in isolation. Service often requires the customer to be physically at the specific time and location. This makes service organizations exposed to many more peaks and off-peaks than manufacturing. The other difference is the heavy use of dynamic pricing. Goldratt wished to use as much market segmentation as possible, certainly using response time as a justification for higher price, but not to the level used by airlines or hotels, which do not offer any justification to the customer for their crazy pricing.
A hidden assumption within Yield Management is that the truly-variable-costs (TVC) are zero. This is about right for airlines and hotels. One of the first articles on Yield Management by American Airlines (where Yield Management was first developed) did state this assumption and even add in parenthesis that adding a traveller to a flight has a certain cost, but it is pretty low and can be ignored. Applying Yield Management to renting cars requires using Throughput instead of revenues, because there are TVC, like the costs of spare parts that are proportional to the miles driven. Using Yield Management, but with T instead of R should be a major insight for many businesses, assuming dynamic pricing is acceptable by the market. As I have already noted – the capacity of every resource is “perishable”.
It seems to me that the main flaw of Yield Management is the crazy run for optimization. The airlines developed super-sophisticated algorithms not just to predict the probability of running out of seats, but also the chance of a seat to be sold at a specific higher price!Daily forecasting of a flight, two or three months before the take-off, causes frequent price changes in a manner not understood by potential customers and travel agents. This general dissatisfaction of the clients, and agents, provides an opportunity for a competitor to change the rules and gain leap in performance.
The assumption adopted by all airlines is that the market would not react too harsh against dynamic pricing. Many travellers know that the probability that their neighbour in a flight has paid the same price is quite low. No one likes it, but as all airlines use this dynamic pricing the travellers have to find ways to fight back, like extensive search for the cheapest flights and becoming more flexible with the timing. The fact that the current price can change in one day puts the customer under pressure. I think that if a certain airline would adopt a strict policy of pricing relative to the time until the take-off, the customers would be more content and the appeal of that airline will grow.
What keeps this reliance on forecasting and optimization algorithms from customers running away are two effects:
One is the simple fact that all airlines imitate each other in almost every detail. The entrance of the low carriers has rocked the boat of the airlines until a certain balance has been achieved and the massive imitation continues. I believe that if a significant airline will introduce a major change in the pricing rules it’d take time for all airlines to decide whether to adopt it or not, meanwhile the balance within the airlines will change.
The other effect is the impact of the frequent-flier programs that create loyalty of the best customers to a specific airline. This is an enormous marketing success of creating special value to customers and it neutralizes, to a certain degree, the negative effects of the crazy dynamic pricing. The readiness to give free tickets (not truly free – just much cheaper) to passengers is based on the realization that the cost is very low. We should all learn the lesson, certainly we in TOC who are able to distinguish between T and OE.
While the airlines understand the need to exploit the constraint of an individual flight, they fall into the trap of the flaws of cost-accounting when it comes to decisions about whether to fly or not to fly and what routes are profitable. So, they recognize the impact of the capacity constraint for a flight, but not for the aircrafts (or the terminals in certain cases), which are their critical resources.
Learning from the good and bad of the management practices of other methods or of other business areas should be a core value to TOC.
4 thoughts on “Learning the Good and the Bad from Yield Management, a methodology developed by American Airlines”
Nice article. In many industries other than the airline industry they view the residual value of a product or in some cases services provided decreasing the closer it gets to the expiration date. Wholesalers offer discounts to reduce inventory, increasing their cash position, and recognizing that the new and improved replacement product about to be released will decrease the residual value of the older product. In turn retailers have Sales for the same reasons.
The airline does not take any action to recoup the potential revenue from an empty seat. Perhaps they should develop a hybrid solution focusing on ‘Zero Empty Seats’. Their forecasting algorithms then trigger action the closer they get to closing the aircraft door.
Beautifully put logic as usual.
There is an interesting justification assumption that reinforces the current balance and local optimizations found in airlines today: that airplane capacity is rigid and hard to modify (elevate) which lowers the priority of strategic pricing changes as you mentioned.
Now after a quick reflection this assumption was invalidated by the airlines themselves with codesharing and other schemes, which can vary capacity much more rapidly than buying (or selling) an airplane.
This wraps nicely with your work on dealing with uncertainty: Yield management appears to ignore uncertainty by: trying to precisely adjust the prices daily to try to exploit seat occupancy and ignoring lines and airplane allocations and capacity adjustments.
Just imagining a buffer of negotiated capacity or flexible lines, or even better shared terminals and/or fingers (gates) sends my brain itching with the potential.
Throughput Accounting works great when the throughput is measured at the System Constraint. Yield Management works great to maximize the revenue per available seat, but the available seat(s) are not the System Constraint. A Flight(s) between a City-Pair(s) (IE – Market) is the Airline’s basic unit of measure for throughput. Add up all of the throughputs from all of the flights to determine airline throughput. Subtract all of the fixed and semi-fixed expenses for the airline to determine the net profit. Using Available Seats or worse yet, Available Seat Miles, is measuring throughput at a non-constraint; it is a poor metric with which to make airline decisions. Yield Management at the Airline Constraint Level (The Flight) is essentially, maximizing Throughput per the Constraint and would be an excellent metric with which to make decisions; but that’s what Throughput Accounting already does.
To calculate and visually display Throughput per Flight and Throughput per Flight Block Hour, I developed the Financial Flight Deck (iDashboards Display) to make it easy to calculate, model and understand Throughput per Flight (TP/Flt) and Throughput per Block Hour (TP/BHr).
Feel free to email me at;
Thanks for all your good TOC thinking and blogging.
United Airlines Captain
This is very interesting and encouraging that United Airlines considers Throughput per Flight and per blocking hours. I certainly like to know more and maybe to add some more ideas to support superior decisions. My email is: email@example.com also firstname.lastname@example.org.