Algorithms are more present than considered, influencing strongly our lives. Nowadays, there are algorithms controlling and tracking everything that we do online, used later for many purposes, including commercial actions. Thanks to the algorithms and their sequential instructions, it is possible to handle a very large volume of information (inputs), in order to obtain a result (output). We can therefore say that algorithms are behind each decision-making process.
Focusing on the world of hotel Revenue Management, the interaction with Revenue Management Systems (RMS) is crucial. Thanks to them and their algorithms, we can establish an appropriate strategy by analyzing the demand sensitivity, anticipating that demand in order to find the best optimization. The different distribution channels also use algorithms, analyzing the conversion in the booking funnel. We cannot imagine the current distribution without algorithms. However, what has happened to them during the hotel closure and how do they act after the current stronger demand? Let us analyze it.
In 1972, a British Airways analyst, Kenneth Littlewood, came up with a mathematical model that calculated forecast demand and how it affected airline ticket sales. His theory “Forecasting and control of passenger bookings” was ahead of its time. This stated that when there are two prices for the same product, it should be sold at the lower price until the probability of selling it at the higher price is greater than the ratio between the low price and the high price.
Indeed, it was the first demand optimization calculation called “Littlewood’s Rule”, which was the fundamental basis for the development of algorithms in the travel industry. In the following years, there were progress in this direction. In 1985, American Airlines, led by Robert Crandall, applied an automated system that incorporated the Littlewood formula, offering different price options based on the advance booking, being at the end of the 80s when these algorithms reached the hotel world.
These advances in recent history revolutionized the hotel industry, although in the beginning it was difficult to imagine that the same product could have a different price depending on its demand. In 2014 I was selected to participate in Boston to create a new Revenue Management system for Starwood Hotels & Resorts called `ROS´ (Revenue Optimization System), which was later applied in more than 1,100 hotels around the world. It was at that moment, when I realized the great impact that algorithms have on our hotel industry. My contribution focused on the unique Resort peculiarities compared to city business hotels, where there is no doubt that the behavior and parameters are very different. Therefore, we had to change the algorithms and add new instructions to the “input” more typical of a Resort property in order for the “output” to get optimal results. We created different seasons to group different demands and adapt the algorithms to similar patents. Additionally, we adapted the pricing to the real number of guests to associate the price strategy with the competitor Resorts; we modified parameters by seasons in advance of the reservation and cancellation and finally we applied this intelligence to the different demands to obtain more precise results by segment. After six months of constant changes and verifications, `ROS´ was born. The solution was to add extra filters to an initial configuration to adapt the calculation of the algorithms. This allowed the same “brain” to act differently, taking into account the peculiarities of each hotel. Therefore, there is no doubt that concepts such as artificial intelligence or machine learning, business intelligence and Big data have completely transformed the way of managing current hotel Revenue Management.
However, these systems, operating correctly thanks to a stable demand, were threatened by a sudden change in demand when COVID-19 arrived. We went to “zero tourist” in just one week, completely messing up these algorithms. After this initial break, the hotel reopened, going through a period of very low demand until the current strong demand for the holiday Resort properties. These have produced an unprecedented variation in inputs, never experienced before, and hardly bearable by RMS.
How RMS do reacted after such a drastic change in demand?
There is no doubt that the algorithms used by RMS systems are not prepared for such abrupt and sudden changes in demand. Historical data and trends were no longer valid to predict zero demand after hotel closures. It is true that RMS incorporate algorithms that exclude unusual data and learn from the changing situation, modifying their outputs appropriately. However, even in this case they act slowly. At the beginning, despite zero demand, the algorithms showed positive estimates due to the influence of historical data. If we add to this the imminent cancellations that our RMS did not consider, the RMS were ineffective at that time. Only question was…where is my algorithm?
The RMS and its algorithms left their leadership to human action, being essential to consider other variables that had not yet happened. Subsequently, after the soft hotel opening, not only manual predictions were in place to adjust the forecast and be able to estimate RevPar and GOP, but also the algorithms were adjusted to adapt the outputs to the new trends, not looking in the same way than before the historical figures.
What have we learned and what does the future hold?
Although the algorithm adjustments, various systems currently incorporate new algorithms combining historical data such as Booking window, average stay, day of arrival, pick up evolution, web activity and even competitor price variations. RMS systems with their algorithms are indispensable for the decision-making process, analyzing all the information in a very precise way. They do the work that we could never do on our own on a day-to-day basis. They are therefore necessary!
However, in my opinion, they should never completely replace human labor, especially because within Hotel Revenue Management there are many other variables that close the circle of optimization managed as a whole. Such as direct channel strategies, different distribution channels, online marketing, group strategy, smart contracting, etc. A RMS by itself cannot consider all the external variables and strategic changes for mid or short term.
In conclusion, there is no doubt that we are constantly developing and with unprecedented technological advances. In the future, we will enter the world of quantum computers, which will be able to solve easily and faster more complex algorithms. The day that RMS incorporate this technology, we will enter a New Era, adding new challenges to the exciting world of Hotel Revenue Management.
