Advanced Forecasting Techniques

Revenue Management is a means of controlling price and capacity that are generally, if not absolutely, known and demand which is much less well known. Hence we need to forecast demand to facilitate optimal control.

Forecasting techniques are used to predict future demand based on historical data. Generally this will focus on a combination of data from the same period in previous year and current sales. Advanced forecast techniques can be used to strip out the effects of price changes and capacity constraints.

The key to successful revenue management is forecasting demand. The revenue improvement increases with forecast accuracy so it is important to use the most appropriate forecasting technique of the many available.

Unfortunately, system suppliers often ìtweakî their standard forecasting technique to fit your data. This will still increase your revenue but if youíre investing in Revenue Management its worth investing the thought and effort to select the right forecasting technique for you and increase your revenue further still.

This selection is not straightforward and depends on a variety of factors, the key ones being:

• customer buying behaviour;
  
  
• the competitive intensity of your marketplace;
  
  
• the historic sales and price data available;
  
  
• the length of the data processing window

Working with you and potential system suppliers we can use our expertise in forecasting to identify and develop the most appropriate forecasting technique for you.

During implementation we can then help you to identify how to integrate this forecasting into your business planning processes and your resources planning to improve your margins even further.

Yield Class Concept & Business Model

The term Revenue Management System is generally used to cover a combination of a forecasting system, as set out above, and an optimisation system. Optimisation is essentially the process of balancing supply and demand - profitably.

A forecasting system can generate significant return on investment. It is the logical first step for an organisation moving into revenue management but the benefits cannot be maximised unless the forecasting system is harmonised with an optimisation system. Deployment of such a system is a much more complex task than forecasting, which is likely to be used by small specialist departments. Facilitating the business in adopting RM principles is a major task. Typically the full roll out of a revenue management system including optimisation will involve many different departments and an element of business re-engineering. Usually changes are needed not only to the customer proposition but also to operational services, procedures and training.

There are a number of optimisation approaches that a business can follow. The best established being that known as the ëyield classí method.

Yield classes sometimes exist in a business, even if they are not referred to as such, but if not they need to be constructed as part of the customer proposition. Organisations need to segment their inventory other than by price; for example by setting out differential terms, conditions and extras and then allocating it to pre-determined yield class ëbucketsí in the volumes as determined by the forecast system for each bucket.

Yield classes should be ënestedí into a hierarchy that will allow inventory in the highest class to be reserved for those customers willing to pay for it, noting that their purchases often occur at the last minute, and ensuring a premium price for the associated enhanced terms, conditions and extras.

Simultaneously inventory that has been forecast as unsaleable other than at low rates can be made available in the lowest yield class at the time that the product goes on sale rather than waiting to the last minute to offload product at bargain basement rates and compromising product values.

Thus profitability can be maximised by selling the right product to the right person at the right time.

The number of yield classes varies from industry to industry but normally falls between four and eight initially. This tends to grow to 15 to 25 for long standing practitioners of RM as they seek to further segment their markets.

Optimisation is a process that needs to run every night to ensure that each days sales and the attendant fine-tuning of forecast demand are presented to the analysts the next morning.

Modern revenue management systems now have very user friendly front ends that facilitate decision support. Analysts and yield executives become much more effective when they have at their disposal a reliable reference tool that helps them understand the relationship between historical and future sales volumes.

Rather than the traditional approach of investigating what the key issues are and then acting on them a good RM system will produce structured exception reports which allow analysts to use their skills resolving inventory problems and managing prices.

At the same time, management benefits from an effective revenue management system gaining a decision support resource that can be used by both the marketing and operational functions. This is invaluable when planning marketing campaigns, and pricing and schedule changes.

Pricing Opportunities

The age-old business problem is selling the right product to the right customer at the right time for the right price. Anything else is either a lost customer or cannibalisation ñ both equate to a missed revenue opportunity.

Do you know how many of your customers would have paid more, or how many of your prospective customers went to a competitor after contacting you? How much revenue would you estimate has been missed? Big isnít it?

The introduction of revenue management opens the door to you to minimise these missed revenue opportunities and increase your revenue. Essentially there are two steps:

• optimise your pricing structure;
  
  
• optimise your price levels
  
  

Any economist will tell you that this can be calculated from the price elasticity of demand. Unfortunately itís a bit more complex than that. Different customers have a different price elasticity, which can change over time and depending upon their reason for purchase. To further complicate matters, many, if not all, will compare your price with that of your competitors.

The answer lies within your revenue management system. Here you have stacks of segmented data on how demand changes with price. Link this with your competitorsí prices and you have the minimum data required to create and update an effective segmented price elasticity model. This model will enable you to accurately:

• optimise your price levels;
  
  
• evaluate new price actions before implementing them;
  
  
• assess the impact of competitors price actions and determine the best response.
  
  

Data Warehouse and Information Provision

Revenue Management, like all decision processes, requires information. That information has to be accurate, available and accessible. Whilst most of this data normally exists in organisationsí operational systems, it is seldom available or accessible in the form necessary to support a Revenue Management system.

Automated Revenue Management systems are very processor and data intensive systems, requiring regular feeds of internal and external data.

This is why a Data Warehouse is an integral part of a Revenue Management solution. It will be designed for the Revenue Management system, but built from the ëbest of breedí Data Warehouse and Database toolkit products.

By collating the source information in a Data Warehouse, the Revenue Management system does not need to continually access, and impede the operational systems.

Whilst most organisations have Management Information systems, and some also have Data Warehouses, traditional Information Provision (IP) systems tend to suffer from some, or all, or the following characteristics;-

• a pre-dominantly financial focus, with very little reporting of non financial measures
  
  
• a range of disparate, and incompatible MI systems, that have evolved across the organisation
  
  
• considerable time spent on manual reporting and spreadsheet preparation
  
  

The result of the combination of disparate systems and manual reporting can be ìmultiple versions of the truthî. Of those companies that already have Data Warehouses, few are likely to have the content, speed or flexibility to support a Revenue Management solution.

A Revenue Management system has very specific data requirements to support the activities of advanced forecasting, optimisation and pricing. This information has to be made available by the data warehouse, early enough in the overnight cycle to enable the Revenue Management system to then complete itís own processing.

An existing data warehouse is typically very unlikely to be able to provide this data sufficiently early in the overnight cycle.

So where do we start to build a Data Warehouse to support a Revenue Management solution?

Firstly we need to assess the existing IP systems, and their respective source operational systems, to determine what functionality, tools and platforms can be extended. We look to re-use and enhance what already exists, to maximise cost effectiveness, and exploit existing in house expertise.

Secondly, weíll design the Data Warehouse, in collaboration with the actual needs of the Revenue Management system. This way, we focus primarily on extracting and loading data which will really be used, rather than ëtalking up, and enlarging the projectí to include every piece of data available

By focusing only on the data that is both available from the source systems, and actually required by the Revenue Management systems, and then designing the data warehouse based on dimensional modelling, the result is a cost effective and performant design that is both flexible and scalable.

This presentation layer must also be appropriate for its use.   There are many MIS presentation systems which offer 'On Line Analytical Processing' (OLAP), but to manage yield requires specialist reports, or an RM specific Visualisation system, which will focus, and prioritise, the most critical unsold inventory.

Finally, the implementation of the data warehouse has to be synchronised with the implementation of the Revenue Management system. This requires a hybrid Project Management approach, able to co-ordinate the activities of both the business and IT, by the use of joint working parties and joint project steering committee. This ensures that all parties are fully bought in to the process and work in harmony.

Operational Research /Performance Monitoring / Audit

Once you have your revenue management system, or if you already have one, how can you be sure that youíre really getting the best from it? What should you be measuring to assess its performance? Can you improve its performance by new or different forecasting and optimisation techniques? Are other parts of your organisation making full use of the information within the system? Do your service delivery or procurement functions use the demand forecasts within their planning to minimise cost and ensure customer satisfaction? Does your marketing team use it effectively to determine the most appropriate campaigns? Does it support your business analysis and planning?

These secondary benefits from revenue management systems are rarely exploited fully yet they can deliver dramatic increases in profit through better/faster decision-making, more effective marketing, improved operational efficiency and increased productivity.

To ensure that you get, and continue to get, the best from your revenue management investment you need to establish correct performance monitoring supported by periodic audits. These audits should address both the system and how the valuable data and knowledge is used throughout the organisation at different levels.