Air Berlin’s Fare Ladder: A story of success and missed opportunities
As a former member of the Team Operations Research & Systems at Air Berlin*, we were tasked with finding ways to optimize the airline’s fare prices and increase revenue. As specialists in mathematical optimization, we were able to bring a new approach to the pricing strategies at the airline. Using data and Operations Research, we implemented an optimization that significantly increased the airline’s revenue — the Fare Ladder Optimization.
Despite the (unfortunately too late) efforts we made in pricing optimization, the airline sadly became bankrupt due to high operating costs, due to cost of fuel and salaries, as well as financial mismanagement by the company’s leadership. Nevertheless, the impact of this optimization strategy remains as a testament to its effectiveness in the airline industry.
We share our story …
What is Revenue Management?
Booking a flight can be frustrating, as prices seem to fluctuate randomly upwards as they get closer to departure date. However, there is a math concept behind this — revenue management. This is a crucial factor for airlines to succeed, as it involves setting the right price at the right time for each flight. Airlines aim to sell all the seats on a flight, with some seats priced lower than others. A pricing process involves defining a sequence of fare prices, also known as the fare ladder, and assigning seats to those prices. The cheapest seats are sold first, with the goal of selling all seats before departure. The remaining unsold seats at current and higher prices are called availabilities. It is crucial to note that the fare ladder is a strategic instrument that remains usually static over several months, while the availabilities can be dynamically adjusted in the short term or on a daily basis.
Heuristic Fare Ladder and Common Inefficiencies
Air Berlin employed a crucial team of pricing managers tasked with developing fare ladders. The airline’s approach to pricing was reactive; when a competitor dominated a specific route, typically the prices were matched. However, this approach was not consistently successful, and also default fare ladders were often used. Despite the concept’s simplicity, numerous discussions were held among the team regarding the optimal number of fare prices to use for competitive matching. There was a need for a more proactive and strategic pricing approach to be developed.
Before developing this new approach, we conducted an analysis of existing fare ladders and identified four common inefficiencies. It was also discovered that these inefficiencies were prevalent across the industry and were not exclusive to Air Berlin:
- Constants Upsells: Constant upsells in a fare ladder refer to the pricing strategy where each fare price in the ladder is a constant increase from the previous one. As an illustration, suppose the lowest fare price is set at 40 €, the following price level could be 114 €, and subsequently, 188 €, and so forth (refer to the visualization below). While this approach ensures that each ticket sold earns the airline more revenue than the previous one, it can be challenging to manage availabilities when there are fewer customers willing to pay higher prices. As prices increase, demand may decrease, leading to unsold seats and lost revenue. This can be can be especially challenging in markets with numerous competing airlines.
- Decreasing Upsells: In contrast decreasing upsells involve a pricing strategy where fare price changes in the ladder can decrease from the previous one. For example, if the previous upsell was 50 €, then the next price level may be 30 €, then 142 € as one can see in the visualization. This strategy runs into the risk of selling tickets too cheaply, resulting in missed revenue opportunities. Managing availabilities became challenging due to the need to sell more seats at higher prices, even if the demand for those seats was lower. This made it difficult to predict and adjust the availability of seats in a way that maximizes revenue.
- Wasting Fares: Another inefficiency observed in the fare ladder strategy is the existence of fare prices that yield little or no revenue for the airline due to their lack of significant demand. For instance, consider the following fare ladder visualization, where the initial fare price is set at 40 €, followed by the next price levels at 90 € and 157 €, but with an abrupt jump to 400 €. The excessive emphasis on selling seats at a premium price level resulted in missed opportunities in areas with lower demand.
- Cutting Demand: One major inefficiency of fare ladder strategies is when the initial price is set too high. This is intended to limit the risk associated with poor availability steering by providing a fallback option. For example in the visulization below, if the lowest fare price is 90 €. However, such an approach carries a significant risk of flights remaining unfilled, even if the entry-level price is maintained throughout the booking horizon.
Optimal Fare Ladder based on a Demand Curve
After thorough investigation, we discovered that an efficient fare ladder must be based on the unique demand curve of each route.
The demand curve in the airline industry is a graphical representation of the relationship between the quantity of airline seats that passengers are willing to buy and its price. According to the law of demand, when the price of an airline ticket increases, the quantity demanded decreases, and when the price decreases, the quantity demanded increases.
For example, if an airline increases the price of its economy class ticket from 500 € to 600 €, the number of seats demanded by passengers may decrease from 100 to 80 per flight, causing a movement along the demand curve. This is because passengers may not be willing to pay the higher price for the same seat.
In the airline industry, a step function can be observed in the demand curve when there is a sudden change in the quantity demanded at certain price points. For instance, if an airline offers a promotional fare of 100 € for an economy class ticket, the demand may increase sharply to 200 seats per flight from the regular demand of 100 seats. This creates a step function where the demand remains constant at each price point until a new price threshold is reached.
Since flights typically operate at regular intervals on multiple weekdays and time slots, we estimated the demand curve using historical bookings, which is typically represented by an exponential function. By utilizing a solver, we were able to determine the optimal fare ladder for each demand curve, maximizing the opportunities for availabilities steering. In other words, given the number of fare prices, the objective is to maximize a step function (green) in a given area (red), as illustrated below.
During our discussions with the pricing team, we explored various degrees of freedom involved in creating fare ladders. These included considerations such as:
- The permissible variation in the lowest fare price.
- The degree of fluctuation expected in the highest fare price.
- The number of fare prices that should be implemented.
- The minimum upsell required.
- The extent to which the demand curve should be adjusted using a buy-down compensation parameter.
- The alignment of fare prices with booking classes and product types.
- The impact of fare prices on passenger demand (pax) and average price (yield) with respect to flights of varying demand levels (SLF — seat load factor).
- etc.
All of these business requirements were incorporated as constraints and parameters into a mathematical program. By running a solver, the optimal fare ladder was identified for the specific route and proposed to the pricing team in digital form, as illustrated below. The proposal gave a comprehensive view on the business impact, in terms of passengers (pax), average fare price (yield) and revenue, if the proposed fare ladder would be implemented. In addition, it also demonstrated how the demand distribution depending on the fare prices should look before and after implementation.
It is important to highlight that we strongly advocated for a complete implementation of the proposed fare ladder, whereby pricing managers were only able to negotiate on the parameters rather than the fare prices themselves. We emphasized the significance of this message to the pricing managers, explaining that altering one fare price would result in a suboptimal fare ladder. We also emphasized that any modification to a single parameter would necessitate the reoptimization of the entire fare ladder and leads to a new proposal by us.
Impact and Results
After discussing with pricing managers and senior management, we conducted several successful test runs using the fare ladder pricing strategy. Since each new fare ladder was applied to many flights scheduled months in advance on different days and at various times, we had to prepare extensively by calibrating the availabilities. We then created a plan to implement the fare ladder across the entire network. Although the strategy proved successful on specific routes, unfortunately, it was implemented too late to have a significant impact on the airline’s overall situation.
Nevertheless, to assess the impact of the fare ladder adjustment, a three-week period before and after the cut-off date was examined, and the results were plotted as shown in the charts below. The light blue areas indicate new revenue opportunities that were realized during the analyzed period.
As one can see in the image above, the initial entry fare remained the same, resulting in no change in the total number of passenger bookings (+0.4%). However, there was a significant increase in both the average fare and revenue (+12.5% and +13% respectively). This increase was primarily due to a much-improved utilization of the low and mid-range fare prices.
We implemented another optimization example where we allowed a decrease in the entry fare, as shown in the image above. This led to a significant increase in passenger bookings (23.8%), but at the same time, it caused a decrease in the average fare (-5.6%). However, despite the decrease in the average fare, the revenue increased significantly (16.9%). As shown in the chart, the revenue increase was mainly due to the decrease in the entry-level fare and a better utilization of the mid-range fare prices.
Conclusions
As former Team Operations Research & Systems at Air Berlin, we were proud to have contributed to the optimization of the airline’s pricing strategies through the implementation of the Fare Ladder Optimization. Our data-driven approach, with the help of Operations Research, enabled us to create a pricing strategy that resulted in a substantial boost in revenue on some Air Berlin routes (the roll-out had so far only taken place in about 25% of the short and medium-haul routes).
Despite our efforts, the airline was unable to overcome the challenges it faced, ultimately resulting in bankruptcy. However, we believe that the Fare Ladder Optimization can serve as a valuable lesson in effective pricing strategies for the airline or other industries with perishable inventories. Addressing the ongoing debate in revenue management on whether to focus on fewer sales at a higher price or more sales at a lower price. We outlined common inefficiencies.
In sharing our story, we hope to inspire others to consider similar optimization techniques and contribute to the growth of Operations Research.
References
Currie, C. S., & Simpson, D. (2009). Optimal pricing ladders for the sale of airline tickets. Journal of Revenue and Pricing Management, 8, 96–106.
In memory of the good old Operations Research & Systems team at Air Berlin: Stefan Heuer, Martin Kuras, André Siggel, Anne Rochow, Katalin Sváb and Bertalan Juhasz
* Note on Air Berlin: It was a German airline headquartered in Berlin, Germany. It was founded in 1978 as a charter airline, and later grew to become Germany’s second-largest airline after Lufthansa. AirBerlin operated both domestic and international flights to destinations in Europe, North America, South America, Africa, and Asia. In August 2017, Air Berlin filed for bankruptcy and ceased operations, with its last flight taking place on October 27, 2017.
