Revenue management models need to adapt to the market change brought about by the rise of low-cost competition and new online sales channels
Structural shifts are clearly taking place in the air travel market, raising the question of how revenue management techniques will now need to change in order to keep pace. Network carriers, in particular, face intense pressure on revenues due to new pricing regimes launched by low-cost competitors and the fare simplification efforts that others have had to field in response. The growth of budget competitors and the rise of internet sales, together with the erosion of traditional fare-class fences and the structural shifts taking place in flyer behaviour have all combined to make the old revenue management models vulnerable.
In their place powerful new modelling techniques are emerging that will allow airlines to adjust prices for market conditions as they really are. Some leading travel companies are already seeing early results in the form of better decision-making. The challenge for airlines is to incorporate such models and adapt their organisations and technical infrastructures to the new techniques, without eroding their brand equity or disrupting short-term revenue streams.
The revenue management tools that have developed over the past couple of decades tend to assume a much more static marketplace than exists today, with a rigid hierarchy of fare rules, a multitude of prices and constantly shifting inventory. There are at least five basic assumptions that now need to be challenged:
* fares themselves are static and the thing that changes is the availability of fare classes;
* demand is distinct for specific products or groups that are "fenced off" from one another to prevent less price-sensitive segments from moving to lower-fare products;
* passengers who do not get the fare they want book travel on another carrier or do not travel at all;
* demand forecasts are based on historical patterns, not up-to-the-minute market conditions;
* competitive actions from other carriers or other unexpected events are captured by the manual adjustment of forecasts.
Since the late 1990s, shifting travel-buying behaviours have rendered these assumptions largely obsolete. Leisure travellers, especially in the US market, are taking shorter but more frequent holidays and often combine vacations with their business travel. Consumers are also becoming much more sophisticated in shopping for the best deals, with latest US surveys suggesting that nearly 40% of leisure travellers are now using the internet to book travel. One recent US study suggests that the average online buyer visits 5.5 websites before purchasing.
Booking behaviour is changing for business markets too. Corporations are not only keeping up the pressure on budgets, but also encouraging their business travellers to shop around for the best deals. Online sites have begun to tap the marketing, such as Orbitz for Business, which claims to have been adopted by 89% of its corporate users in the first quarter of 2004.
Online channels
Online channels have clearly given consumers the power to view and compare prices. With websites such as FareChase, SideStep, BookingBuddy, Mobissimo, Qixo and (for hotels) Travelaxe, travellers can search the web automatically for the best deals. A new start-up, Hamlet Inc, is recruiting for data mining gurus to commercialise a technology which promises to advise consumers on the right time and price to buy an air fare.
Small wonder then that traditional revenue management assumptions have broken down. But the sophisticated tools built up by the major carriers in very different market conditions have generally not been adapted to the new realities. The typical revenue management is based on quantifying the opportunity cost of the next seat to be sold in each cabin on every flight leg. Such models do at least four things:
* forecast demand based on the time series of demand from previous years and from similar days of the week;
* weigh the value of the next seat sold against the cost of potentially displacing future demand from higher-value fare classes;
* make fare products available when the revenue generated by those products exceeds this opportunity cost;
* adjust for network displacement costs for multi-leg itineraries;
* consider a variety of other important considerations such as overbooking and upgrading.
In particular, if historical demand suggests that the probability of selling the last seat on an aircraft is zero (or nearly so), all fare classes will be open (subject to the applicable fences). However, when passengers might buy up, or when competitive pricing affects demand, this may not be the right response.
Instead of considering only opportunity costs, revenue management in today's environment needs to consider other factors of price elasticity. If the price is right then passengers may well buy alternative products rather than stay neatly within the fare class that the airline lays out for them. The fares being offered by a competitor may also influence how inclined the buyer is to accept the offer in front of them. Different sales channels may also show different levels of elasticity and indeed passengers may switch in response to price.
A RESPONSE-BASED MODEL
In short, what is needed are tools that clearly model the market's response to an airline's own pricing decisions and its response to those of competitors. Such a "market response-based" revenue management concept rests on four key elements:
1. Market awareness
To start, fare intelligence must be collected on a more systematic basis and extend beyond the usual published sources to include channel-specific prices from carrier websites and online travel agencies such as Expedia, Orbitz and Travelocity. Indeed, airlines are now beginning to trawl for such fares, either using in-house capabilities or outside web scraping services.
Carriers also need intelligence on booking volumes at various price points. Knowing their own volumes is relatively simple, although getting timely sales information may be a challenge for some channels, but getting access to competitive volumes is a more a different matter. MIDT information tapes from the global distribution systems (GDSs) may provide only a partial view since they fail to cover some key channels, but useful competitive insights may be gleaned even from this incomplete view.
2. Statistical modelling
From the volume and price trends, the elasticity of demand can be inferred although it is a subtle and delicate science. This is where the traditional dark art of revenue management comes in, though various statistical techniques are available to model the response.
3. Optimisation tools
A whole range of fare optimisation tools are brought into play, starting with analysis of the fares that should be offered in a given origin and destination (O&D) market based on price elasticities and current competitive conditions. This is where the probability of customer "buy-up" is considered. Given the propensity of customers to book in a higher-fare class when a lower-fare class is unavailable and when competitive pricing is as currently prevails, should the airline close the lower-fare class to force the buy-up?
This is separate from the decision of whether to close a particular fare class because there are not enough seats on the flight to meet all the demand. In contrast, it can be largely made on an O&D-specific basis, without too much concern for network effects. This decision process may result in closing certain fare classes, even though they may be above the "bid price" that would be calculated by a traditional, network-based revenue management optimiser.
In some cases, an airline may consider offering different fares through different channels where the elasticity varies. Naturally, making such a strategic decision would go beyond mere tactical revenue management, but a new style market response-based model could support the analysis required to do so.
Such a new model still needs to be integrated with traditional forecasting and decision-support tools - the need for traditional revenue management based on opportunity cost does not go away. Indeed, the O&D trade-offs made by the complex optimisation models are still critical, at least for itineraries that include capacity-constrained flight legs. One practical approach is to view market response-based revenue management as a pre-process to a more traditional revenue management system.
It is possible to derive a "first principles" approach that might entail more of a ground-up, all-inclusive solution, but many carriers have substantial sums invested in their traditional revenue managements and very specific interfaces to the GDSs.
4. Test and learn
A final element of the response-based model centres on a systematic "test and learn" capability. That includes a systematic approach for generating and leveraging market response insights. In some cases, the natural variety of prices and availabilities in the market may be sufficient to develop an informed market response model. In other cases, it may be necessary to derive additional insights which can be done through a "design of experiments" approach.
It is also important to understand the "physics" of market response, which experience suggests typically obeys certain general rules, akin to natural laws. Just as the gravitational attraction of two bodies in space depends upon mass and distance, so elasticity is determined by the relative size of competing airlines and the gap between their products.
Elasticity is generally directly proportional to the competitor's market share and inversely proportional to an airline's own market share. And the greater the "competitive distance" between two different airline product offerings, the lower the cross-elasticity demand. "Competitive distance" is the degree of incomparability between different products. For example, two competing flights that leave 1h apart will have lower cross elasticities than the same flights leaving 30min apart. Other sources of competitive distance for airlines include differences in:
* the geographic distance between different airports;
* service levels such as non-stop versus direct versus connecting;
* distribution channels;
* brands and perks such as frequent flyer programmes;
* the physical product itself.
Developing and leveraging market response insight will also require changes to the way work is organised in a typical revenue management department. Market testing teams will be responsible for the design of experiments for certain key O&D markets. They will leverage what they learn across other markets that behave similarly in terms of the "physics". These insights will be used to calibrate the fare optimisation tools.
Many airlines are already starting to challenge the way their traditional revenue management models work, and, in the process, trying various pieces of this approach. Their analysts try to gauge the appropriate fare open/closed strategy in these O&D markets based on competitive fare availabilities and gut feeling. These extensive manual overrides actually required them to hire more revenue management analysts - at a time when headcounts were being reduced across the board. Of even greater financial impact, however, are the inevitable mistakes that occur in such a manual process.
Other carriers have tried to implement more automated approaches by "tricking" their sophisticated O&D optimisation systems into closing out lower-fare classes by heuristics that introduce artificial biases to either demand forecasts or to flight capacities. These have helped to improve revenues, but tend to introduce other unwanted side-effects and result in recommendations that no-one could assert are truly optimal.
Almost no carrier today can say that it has solved the market response-based revenue management problem. Ironically, it sometimes seems as though revenue management could learn a thing or two from good, old-fashioned pricing theory, including learning from other industries such as retailers and hotels that have recently made significant advances in applying new market-based concepts.
An implementation plan must address the need to generate and learn from early results. Waiting for the development of a major new technology before experimenting is probably not the best option. Looking at other industry sectors, a lot of benefit can be delivered through less technically sophisticated solutions. In addition to learning from these short-term wins, the financial benefits they throw off may well be able to fund the development of the ultimate solution on a "pay-as-you-go" basis - something critically important to airlines these days.
The revenue management discipline has come a long way since its early days. Yet, in some sense, it must come back to its roots in classical pricing theory to take the next steps forward. Market response-based revenue management is where revenue management meets pricing (again) and knows it for the first time. The potential is clear, the early results are promising, and the issues are manageable; it should be a profitable reunion.
BY SCOT HORNICK AT MERCER MANAGEMENT CONSULTING IN CHICAGO
Source: Airline Business
