There are many commercial revenue management systems (RMS) available today. Although there are a growing number of new and exciting developments in Revenue Management analytics, there are many problems that remain unsolved. At the center of any good revenue management solution is its core analytical capabilities. Which business factors are considered for great forecasting, optimal pricing and overall business optimization? Which are not? These inputs directly contribute to the value a solution can bring, and also directly influence the degree to which the key outputs designed to grow revenue and enhance profitability are reliable and acceptable to the users. In this post, we aim to bring insights into some of the key advanced analytical capabilities that bring benefits in an RMS solution with a particular focus on the hospitality industry.
Understanding Marketing Segments
Many hotel systems (including RMS) aggregate data based on pre-defined fields (e.g. market segments or room types) from the hotel transactional data. Unfortunately, important information is often lost as a result of this grouping process – in particular, the information regarding how rates are priced and controlled. Today’s revenue management systems are able to automatically and analytically build the most effective groupings of available data to provide the most robust basis for forecasting but more importantly – for business optimization. Considering the tasks required in order to price and/ or yield segments appropriately to drive better revenue.
Analytical market segmentation allows an RMS to separate the transactions based on factors that influence their particular demand types and behaviours. Meanwhile, forecasts can still be made available to system users in a manner consistent with their segments and organizational reporting requirements.
Let’s look at an example of why this is important – imagine we have two corporate rates: One is contracted at a fixed price with Last Room Availability (LRA) for standard rooms only. Meanwhile another corporate rate is contracted at a guaranteed 10% off BAR and it’s fully yieldable – meaning the hotel can restrict it as required and the price can be managed when the public pricing is changed. From a business perspective, these two rates will often end up in the same market segment in the property management and central reservations systems, simply because they are both contracted corporate rates. However from a forecasting and optimization perspective, we need to differentiate between price-sensitive (dynamically priced) products and price-insensitive (fixed price) products as their booking behaviours are distinctly different and the way we can control them is also different. In revenue management there are many similar examples where rates that appear similar from a marketing perspective, behave differently and are priced and yielded separately, when you are utilizing in an RMS to ensure optimal revenue management processes.
Unconstraining the Demand Correctly
In simple terms, unconstraining is the process of estimating the total demand that exists for a type of inventory regardless of any constraints, such as physical capacity. Unconstraining is essential: if we underestimate future demand we will lose revenue due to under-valuing our rooms. However, over-estimating demand will also lead to lost revenue from over-pricing our rooms. In several studies (e.g.1), robust unconstraining has been shown to improve revenue performance between 3-7% percent.
Statistical approaches for unconstraining have been shown to consistently outperform approaches which are based on regrets and denial data, which have been shown to be largely unreliable. The more advanced RMS’ available today should be able to address unconstraining analytically to manage data censoring, sparsity, and noise and deal with multiple levels of data hierarchy to account for seasonality, day of week etc. This provides the most effective basis for the profit-enhancing controls produced by the RMS.
Traditionally, RMS have used a limited number of proprietary forecasting models at a level that is defined manually by the users to forecast demand for market segments or room classes. High-performance forecasting, on the other hand, relies on hundreds of advanced forecasting models where the most appropriate model is selected by the system automatically, based on the analytical behaviour and optimization we can apply, as described above. Then the forecast model parameters are calibrated to effectively understand the impact of the specific price-sensitivities, no shows, cancellations, booking curves etc. within the forecasting group.
Analytics can be employed to solve a variety of challenges, including adapting the forecasts to demand shifts, accounting for the impact of linked demand (i.e. rates that have a price which is derived from the value of another rate), and understanding demand as a function of price (The impact of price changes on the demand that exists for the product). Why is this important? Let’s say we’re trying to price our rooms for certain length of stay patterns that show similar behaviour; it may be better to estimate them together, because we can combine their data and build a more robust basis for forecasting. On the other hand, if we’re trying to estimate price elasticity, it may not be a good idea to pool room types from distinct classes because a deluxe and a standard may show different behavioural patterns in terms of willingness to pay.
Integrating the Price-Sensitive and Fixed Price Demand
Powerful analytics also matter when optimizing your prices in the presence of a common hotel business mix – i.e. one that includes a mixture of dynamically priced public rates, rates which are linked to those (often with a fixed, contracted discount e.g. -10%) and other fixed priced offers. For example, many corporate clients will not accept flexible discounts or may only accept a fixed rare. If they do accept a flexible discount percentage – the percentage is generally fixed across the contract term. It is then important for the RMS to understand that, when defining the public price, that any rates derived from that rate are also impacted.
This complex business mix that many hotels have to manage – a mixture of fixed and dynamically priced rates, Last Room Availability and yieldable rates and a mixture of complex booking conditions (or fences); highlights why it is so it’s important to manage all of the demand – dynamically priced and fixed priced in the property. This is where an integrated approach to optimization comes into play: optimizing for both pricing and availability. All types of products – the price-sensitive and the price-insensitive products, your fixed and dynamic corporate negotiated products, your public price (retail) product and any other dynamically priced products all consume the same limited bedroom stock. Therefore, pricing optimization needs to understand the trade-offs between them and optimize them together.
Why is All of This Important?
As you can see, powerful analytics are the real performance driver for hotel revenue management. Hoteliers worldwide should ask the following questions when evaluating the analytical engine behind the revenue management systems they are considering:
- How does the system handle the segmentation of the hotel data? Is this constrained by coding in the reservations systems?
- What types of forecasting methodologies are in place?
- How does it handle unconstraining of the demand for all different demand types: public pricing, fixed, dynamic, corporate negotiated, and wholesale?
- How does the RMS manage the optimization of different types of product?
- How does the system address the need for managing all of your business types (both pricing dynamic and fixed rate businesses)? What controls are produced and at what levels?
- Are all of the controls driven by analytics or through user constraints?
The answers to these questions will provide you with a better understanding of the nature of the RMS you are considering.
1 Queenan, C., Ferguson, M., Higbie, J. and Kapoor, R. (2009). A Comparison of Unconstraining Methods to Improve Revenue Management Systems. Production and Operations Management, 16(6), pp.729-746.