According to McKinsey, AI-based pricing can deliver between $259.1B to $500B in global market value. But the critical question remains: can Revenue Management be completely automated? The answer is: theoretically, yes, but, in practice, things are a little more nuanced. But let’s step back for a second and try to reword the original question, at least slightly: should Revenue Management be completely automated? The answer, in this case, is a big, resonating, capital letter YES. A recent study by MIT-BHI showed that companies that “undertook AI-driven pricing transformations achieved more than $100 million of revenue improvement 70% more often than companies that focused on another area”.
“Self-learning algorithms are evolving fast, becoming highly sophisticated, and they already have a high impact on operational efficiencies and increased yield. As a result, there is no doubt that the future of revenue management will be fully automated,” says Alexander Edström, CEO, Atomize.
The pieces of evidence are all around us, and not only in travel. Some examples? Amazon uses artificial intelligence to drive dynamic pricing; Starbucks adopts predictive analytics based on its data from over 90 million weekly transactions, and multinationals such as Coca-Cola or Johnson & Johnson have been using AI pricing for years. During the 2019 edition of the Revenue Management & Pricing in Services Conference, hosted by the prestigious Ecole hôtelière de Lausanne, Kevin Hof, Data Scientist at RoomPriceGenie, shared several case studies where hotels experienced an average of 22% increase in revenue by adopting RMSs, and similar results can be found on dozens of similar publications.
“The hospitality industry is very fragmented when it comes to tech adoption and AI implementation in revenue management. Many hoteliers are still very protective of their own pricing and strategy; they believe that their historical knowledge and gut feelings know better than any algorithm. The truth is: that they don’t trust what they don’t necessarily know, understand, and cannot control (like a Human Revenue Manager). That’s when tech Vs. human becomes a dilemma, and that’s when we need to go back to basics and work on the “tech it easy:” step by step education followed by measurable results. Revenue Management is a hybrid human+tech cooperation, and the future is already now,” says Silvia Cantarella, Commercial Strategy Expert & Founder Revenue Acrobats.
PMSs, GIGO, and Adoption Blockers
Based on these facts, we can all agree that autonomous intelligence (in the sense of automation of price decision-making processes with little or no human intervention) is likely the direction towards where revenue management is headed. However, this is not going to be particularly easy, and not (or, at least, not only) because modern RMSs are not up to the task, but because, oftentimes, PMSs’ are the real adoption-blockers. GIGO is a concept most revenue managers know way too well: the quality of the output is determined by the quality of the input. “Garbage” data “in” produces nonsensical (“garbage data”) outputs. And RMSs’ rely heavily on PMSs’ data. These systems can fail to provide the correct information for a number of reasons:
1. Human data entry errors. That’s the main issue when it comes to PMSs’ accuracy. Frighteningly enough, the error rate in clinical data repositories can be as high as 27% and, in our industry, a study by eHotelier pointed out that up 50% of hotel databases are either corrupted or incomplete;
“We need to rethink market segmentation or rate architecture with data quality in mind. Data need to be actionable and allow us to effortlessly and effectively uncover opportunities. But more than anything, the setup needs to prevent data entry errors,” says Christoph Hütter, non-traditional Revenue Manager.
2. Poor API design. Tyler Charboneau calls this the “Instant Gratification Trap”: “This potential short-term gain is attractive.” Charboneau says, “we can discuss the ideological war between engineering and marketing, (and) chasing instant gratification is understandable, but dangerous. It’s like developing a tentpole API when your core services really require thick rebar. Ideally, the design process would include thorough stress testing and optimization. Reliability is also crucial.” It’s not uncommon for PMS companies to heavily market how “open” their systems are, yet we should remember that quality always beats quantity when it comes to integrations. We cannot stress this enough, “One thing that is often forgotten, when the benefits of RMS systems are calculated, is the RM maturity of the organization. Installing any simplest algorithms will bring great returns when you start from zero revenue-mamagement-know-how. I have unfortunately witnessed a top RMS with poor human setup run property market share below its quality position due to poor know-how on RM strategy,” says Hanna Lak, Tourism Knowledge Management and Total Revenue Ambassador, Founder Nordic Revenue Forum
3. Lack of functionalities on the PMS. Well, this is quite obvious, yet, if your property management system has not been developed to collect a specific piece of data, the RMS (and even the human revenue manager, for that matter) will not be able to pull the proverbial rabbit out of their hats. “While we still largely discuss room revenue where the data granularity from a PMS is pivotal and where automation should already be the status quo, the industry should focus on total revenue management, investing on how to integrate data from different revenue sources (the so-called ancillary revenues, such as SPA, golf, food outlets, etc.) into one solution, a single source of truth,” says Damiano Zennaro, Hospitality Senior Advisor.
To Automate or Not to Automate?
Full automation of your revenue management strategy is tempting. Not only because it can (and usually does) increase your overall revenue, but because it also dramatically reduces costs. The average salary for a revenue manager is $81,399 per year in the United States, while an entry-level RMSs’ cost $50 per month. Well, you do the math.
“Given that a typical hotel will make roughly five million pricing decisions every year, it is not humanly possible for any revenue manager to get every decision right, every day, without the support of an automated system. Especially considering the sheer volume of data to be gathered and analyzed. A robust RMS not only generates prices that adapt to market changes but actually anticipates these variations in advance. In a changing hotel market, slight pricing changes can have a big impact on demand. Therefore, any hotelier operating without systems that can analytically decipher the impacts of a specific price change on occupancy and the resulting revenue benefit (or lack thereof) for their property is operating at a disadvantage,” says Klaus Kohlmayr, Chief Evangelist & Development Officer, IDeaS.
Moreover, revenue managers can rely less and less on historical data, especially after the pandemic. In this new paradigm, external data (such as weather reports or public events information) became crucial for a solid RM strategy. First, however, humans have to collect and aggregate them, bringing us to square to problem #1. But these data could automatically be retrieved on services such as OpenWeatherMap, Picatic, or Allevents.in. They’re, literally, one API away.
“Typically revenue management systems (RMS) and revenue managers use historical, comp set pricing and market data and combine this with forward-looking demand signals like pacing to recommend the optimum rates. Unfortunately, due to the pandemic, historical data has become irrelevant. I also believe comp set pricing data has also diminished in value – how sure are you that your competitors are competent in their revenue management practices and use the right RM tools and not just plagiarizing each other’s rates in a suicidal downward spiral?
The need to adjust the property’s rates dynamically, based on real-time travel demand, is the reason why the pandemic put an end to the most favorite revenue management tool: the Excel spreadsheet – and created the urgent need for an AI-powered cloud RMS to handle the complexities of the post-pandemic era,” says Max Starkov, Hospitality & Online Travel Tech Consultant & Strategist.
Conclusions: Where Do We Go Now?
A recent pricing maturity assessment conducted by BCG and the Professional Pricing Society, revealed that “more than 50% of all industrial goods companies still use Microsoft Excel to build their primary pricing tools.”
“It’s clear an AI-powered RMS with human supervision is an essential starting point for today’s hotelier. As revenue tech moves closer towards marketing functions we should expect to see more integrations between the RMS and marketing tools to optimize the guest journey on the hotel website, on the guest mobile device, and other important digital touchpoints that are currently ‘out of reach’ for today’s RMS solutions in isolation,” says Erik Muñoz, Chief Revenue Officer, Userguest.
It’s not surprising, because revenue managers often are forced to fill the blanks with the little information they have, and that’s mainly the vendors’ fault and responsibility. However, a new, easier, more effective, and more accurate way of doing revenue management is knocking at our door, and it’s time we all team up and, finally, open it.