The main pillar of Revenue Management is demand forecasting. Anticipate the events certainty reduces uncertainty and allows us to establish strategies that help generate additional revenue. Without accurate demand prediction models supporting our strategies, we would never be able to identify real opportunities that optimize the GOP. Sometimes we consider that there is an opportunity without having a global vision, focusing on a single piece of information, something that generally generates risk. In principle, what we think that it is an opportunity can become a risk. Therefore, we can never talk about real opportunities if we do not consider all the variables. In fact, even with all the information we are not safe from making mistakes.
In the courses and Master’s degrees that I teach, I always begin my introduction by mentioning that Revenue Management is not an exact science and, unfortunately, it never will be. Considering all changes, our horizon advances faster than we can control, distorting our environment and moving us away from our comfort zone. And although it may seem like an uncomfortable scenario, it is actually the perfect moment to learn and evolve. In this constant learning, new tools appear that help us organize our business plan, such as Artificial Intelligence (AI). But how does AI collaborate with Revenue Management systems?
New parameters to consider.
In this sense, focusing on the importance of demand estimation as a fundamental pillar, algorithms play a transcendental role. They support our decisions based on the management of enormous amounts of data that we could never digest otherwise. In an initial stage in the RMS (Revenue Management Systems), the algorithms only considered certain internal parameters, such as historical occupancy data, competitors’ prices, seasons. In my opinion, they were trying to offer with this limited information an estimation of potential demand. Even today, there are some RMS that are still based on simplified internal data due to the difficulty of indexing external data outside of competitors’ pricing.
In the development of the RMS, in which I participated due to my experience in Resorts a few years ago, we included parameters never seen before and nowadays infrequently used. We added to the equation the competition’s prices by room type, affinity percentages for each competitor (…since we are never proportionally equal to each competitor), and more complex algorithms to determine the sensitivity of demand to price, among others. It therefore considered the historical impact of the pickup after price variations based on similar demand parameters, granting each future day a level of risk according to price variations, which was shown at various levels, low price sensitivity, medium or high. Important concept and in my opinion essential when establishing a pricing strategy. It is not more important how much we raise the price (simplified data), but the impact that this price increase generates on demand, since it could have the opposite effect to that desired. Even today, having this data, it is difficult to see algorithms that consider the price change to fluctuate the demand forecast. In any case, it cannot be static or based on historical data where perhaps there were no price changes.
With the arrival of Artificial Intelligence to the business sector and obviously to the hotel industry, a new stage begins that transports us to a new dimension. I am not referring to virtual assistants, but to the incorporation of more complex algorithms powered by AI, which will undoubtedly take center stage over internal data.
This capacity for global vision that I reiterate so much is precisely what AI has, since it can consider external parameters that had never been included in demand prediction models. Let’s imagine for a moment if, in addition to the existing data in an RMS, we add certain external factors controlled by Artificial Intelligence, for example:
- Socioeconomic factors: Evolution of consumption and future employment predictions, CPI forecast, economic situation and evolution of product prices per country, being able to measure their detailed impact by the hotel’s mix of nationalities and anticipate possible changes in the elasticity of demand.
- Impact of demand due to climate change factors or policies carried out by the different emitting countries in this regard. Airlines strategies, associated costs, new technologies in order to reduce the carbon footprint and its impact on consumption and the future of airline companies.
- Changes in air routes worldwide, and especially due to international conflicts that modify the demand between destinations, anticipating possible route changes due to the resurgence of conflicts. Today we can observe variations in pickups for this reason, but we seek anticipation in a precise and global manner with data provided by AI.
- Health, employment and social welfare data. The social climate is not always stable and may have an impact on demand. There are many studies in this sense, however, they are not always considered or associated with hotel demand.
- Demand models based on generational changes that is currently affecting the hotel industry, where the use of social networks exposes our business. We are looking for a model that associates the RPI (Revenue Performance Index) with a competitive set based on the anticipation to this generational change, considering factors from online reputation and reviews.
These are some examples where Artificial Intelligence can feed and adjust the demand forecast calculation model, considering external data that today seem irrelevant because they cannot be considered, however having an impact on future demand.
In Revenue Management, the first technological New Era of Artificial Intelligence is beginning to be outlined. In the tourism sector, where we are only able to see nowadays the top of the iceberg, the new data from Artificial Intelligence will inevitably take on a greater role in today’s Revenue Management discipline. I am convinced that, if at this time I would participate again in the implementation of a new RMS, we would consider algorithms including data from Artificial Intelligence, which, together with the capacity of machine learning, would revolutionize the way in which we interact with systems today.
In my opinion, with AI we are facing a new technological revolution in the forecasting of knowledge and information applied to business. And although at this embryonic stage we have doubts about the implementation in the hotel industry, it will undoubtedly be an opportunity to explore new limits in the context of hotel Revenue Management.
