All Revenue Management Systems (RMS) must satisfy a list of functional requirements, such as producing an accurate demand forecast that controls for seasonality, trend, and special events. In addition, it’s essential that the system recommend optimal prices to maximize revenue while accounting for business constraints, as well as continuing to provide a user experience that clearly communicates these recommendations by explaining the key drivers, and the level of confidence that the system has in its own recommendations.
However, what is often glossed over (or even overlooked entirely) is that non-functional requirements can be just as important. These can include producing daily optimal recommendations before 9 a.m. to each office, maintaining an audit of demand forecasts overrides and price changes to support post mortem analyses, and establishing a flexible and scalable system that can support growth and changes in the business over time. Of these, the last can be the hardest to achieve, but it’s perhaps the most important to ensure the system does not become obsolete, given that fresh scientific models are being researched all the time.
It’s important to expect the unexpected and adapt your business to changes in industry which may impact your analytics, and to keep in mind that external interfaces surrounding the RMS could be undergoing changes of their own. How can you design your RMS to handle these changes when it’s unclear when they will arise, and how do you prioritize what to address first?
The key is designing with modularity in mind. At the core of the system lies the analytical modules, and surrounding the core are layers that will insulate and cushion the system from changes around it. To achieve this, establish a contract of assumptions for each transition from one layer to the next, and from module to module. Then, wrap each module with the data transformations necessary for it to function, and perform data validation checks to raise alerts when basic assumptions for the interface are broken, which will proactively alert application administrators.
From a technical perspective, the most common external changes from the simplest to complex are:
- Migration to a New Data Warehouse. Typically, with this migration, you might see a change in the quality of the data and a change in the format, but the meaning of the data usually does not change. What should never be overlooked is the layer wrapping the Revenue Management System that performs a set of validation checks to proactively alert you to unexpected changes in the data inputs. This will ensure that any assumptions made by the system are not compromised.
- Migration to New Reservation Systems. This migration can bring with it inherent changes in the way a business operates. For example, in the hotel industry, one reservation system may only support selling limits while another supports hurdle rates, which is a minimum yield required for a booking to be accepted as determined by the yield management system. Any bookings where the price is above the hurdle would be accepted, and the exact value of the hurdle rate can change from day to day. One reservation system may support a business rules engine for both prices and ancillaries, while another may only support basic parameters for prices such as location, date and length of stay. In this case, the core analytical models are likely to change while the data interfaces remain about the same.
- Migration from an On-Premise Solution to the Cloud. This type of migration is often the most complex; by partnering with an experienced expert that has both technological and analytical expertise, they will take the time and move capabilities incrementally. The myriad of unknowns a company may ask itself are: Are the underlying technologies still supported in the cloud? Can interfaces to external systems still function and perform, given the volume of data moving back and forth? Are there security concerns that need to be addressed? Do new scientific models exist that could not just replace, but improve your existing system? These unknowns can be addressed by conducting a methodical review of the existing system. Your experienced expert will examine which components are higher value/lower risk, and which ones require further research. This approach prioritizes the return on investment, and will enable you to make the best decisions as each new unknown arises.
Each of these scenarios has a different scope of change, and change can be a difficult thing to manage. A robust system will ensure that when an external change is taking place, there are appropriate controls in place to limit any possible damage, and reduce the risk of becoming more widespread.
By establishing a mindset for protocols between layers, and focusing on showing benefits at each step along the way, your Revenue Management System’s adaptability will improve and its longevity will increase.