Revenue Management in a Crisis—Rise of the Humans
We have elevated machines to perform much of our work in Revenue Management. It’s time for humans to step forward.
Revenue Management systems have become technological marvels. They have consistently won or been finalists for INFORMS’s global competitions for applied analytics.
We’ve almost become used to Revenue Management systems performing herculean tasks for us. We’re no longer in awe of what they can do. They manage billions of rows of data with ease. They perform tens of millions of forecasts per night across thousands of properties. They accurately set millions of rates per day, deftly balancing rate and occupancy throughout the year. Advances in artificial intelligence and machine learning have made them reliable and self-correcting. They have been our noble, autonomous agents working tirelessly for us 24 hours a day, 7 days a week.
We’ve relied on them to instantly access enormous databases to answer virtually any question which could come to mind:
- “Which hotels in Singapore will need more groups in September?”
- “What properties in Eastern Europe did not meet RevPAR expectations last quarter?”
- “Will we need to offer deeper discounts in South Florida this coming winter?”
Now, the underlying source of the Revenue Management systems’ intelligence—their databases—have become victims of COVID-19.
Even the most advanced artificial intelligence and deep-learning systems must rely on predictable data or rules. Self-driving cars must have real-time sensors, as well as know the “rules of the road” and the laws of physics. IBM’s Watson is dumb as a brick without consistent, reliable data. The creation of a machine with human-level intelligence that can be applied in any unknown environment is the Holy Grail for many AI researchers, but it's not something we’ll see anytime soon, if ever.
GIGO is the fundamental precept of computer science. Garbage in, garbage out. A global shutdown which began in Wuhan, China, in January of this year has trashed Revenue Management databases for at least a year—maybe more.
In an interview with HSMAI, I was asked if I had ever seen any disruption to the economy like this. “No” I replied. “The current crisis has the immediate impact of 9/11 and the long tail effects of the 2008-2009 financial crisis.” I projected that if you combine the impact of both those events and multiply by pi, then you’d get some idea of what we are facing with COVID-19.
“Old World” Data and the Coming Murky New World
Cameron Davies, SVP Decision Sciences of NBCUniversal, has experience with many situations in which historical data was non-existent or has become irrelevant. He cautions that we should not simply throw the “old world” data out as garbage. Although the new world will be completely different, he suggests that the old data can be a “foggy window” into the new world.
I like his sentiment that the old data still retains some value. However, it may be more like a carnival fun house distortion mirror than a foggy window. The data that’s currently in Revenue Management databases will have some probative value—but at this point we have no idea how, when and where we will be able to use it. Data that we are currently collecting is also completely distorted, but some of it may foretell future customer activity.
It appears that the COVID-19 infection rate peaked in mid-April in the United States and that we may be nearing the global peak of the pandemic—assuming there is no second wave.
As this is being written, it is unclear and confusing how and when the U.S. economy will reopen. Debates are ongoing. The CDC & FEMA have drafted a plan, State governors are forming coalitions to develop re-opening steps and the White House has issued its own guidelines. Any re-opening of the U.S. economy should be tempered by testing and tracking. We can expect starts and stops if infection rates hit an upward inflection point. Global re-opening is even more uncertain.
Perhaps more important than what governments will do post-COVID is how customer behavior will change. After the initial shock of social distancing guidelines and stay-at-home orders, we’ve become accustomed to teleworking, e-health, distance learning and shopping on-line. Many of these habits may remain. However, the greatest influence on customer behavior will be driven by the completely unknowable impacts which the virus and economic shutdown has had on economic, financial, psychological, and societal trajectories.
The human brain is an amazing organic computer. According to some estimates, our minds can store up to 2.5 petabytes—about the same as all of Yahoo’s computers combined. The clock-speed of the neurons in our brain, at only one kilohertz, is substantially slower than the one gigahertz processor in our cell phones. However, our brains have 86 billion interconnected neurons. A neural network computer with that many connections could conceivably be built in a few decades, but it would require a gigawatt of power—about the amount of energy consumed by Washington, DC. Our brains run on 12 watts of power—about the amount of energy consumed by a single high-efficiency light bulb.
Far more important than our prodigious data storage and processing power is the interconnectivity of all those neurons. The connections the neurons create give us the ability to perform tasks, make decisions, and solve problems based on what we call our human instincts, our common sense, and our life experiences. That interconnectivity of neurons enables the human qualities we refer to as creativity, ingenuity, imagination, and inspiration.
Because of these uniquely human traits, we can form mental representations, make inferences, draw conclusions, and make decisions in circumstances we’ve never seen before. This problem-solving ability when facing complete unknowns in a new environment separates us from machines.
Our brain is not merely a computer. We can do things that computers will never be able to do. We can be empathetic, and we have emotions. While facial recognition software can determine a person’s mood, it can’t form an empathetic bond with a patient. Amazingly, there is statistical proof that empathetic doctors have better patient outcomes than non-empathetic doctors with even better technical skills. This empathy and understanding is essential to us as individuals and for businesses as well. As consumers, we want to engage with brands that, we believe, understand us and appreciate us.
What Humans Must Do Now
It’s still too early for us to go back to practicing Revenue Management as we know it. There is little that can be done to create and manage demand. More harm than good can be done by trying to force recovery before it is possible. First, do no harm.
Nonetheless, there are many things that Revenue Managers should be doing in order to prepare for the essential role of humans after the re-opening of the economy.
Acquire Knowledge to Prepare for A Problematic Recovery
You must arm yourself with additional knowledge in order to face the unknowns in front of you. Fortunately, there is a great wealth of material available specifically related to the COVID-19 crisis and its impact on Revenue Management. (Perhaps we authors have more time on our hands!) Unfortunately, there is so much available that you might become overwhelmed. Here are some of my favorites.
- Boston Consulting Group has published and excellent high-level overview of what to expect in Revenue Management for all travel & transport industries. The U.S. Chamber of Commerce wrote about the expected impact of the coronavirus on a broad range of industries including hospitality. Spoiler alert—they see fewer people traveling across-the-board.
- Great articles abound. The New York Times reported on an excellent survey of what to expect from changes in travel behavior among various geographic and demographic markets.
- Closer to home, there are a number of articles which give very specific guidance for Revenue Managers. Hospitality Net has published a series of articles with suggestions from industry experts. One of the best overall collections of perspectives, insights and recommendations for hospitality Revenue Managers has been assembled by HSMAI.
Acquiring new knowledge may also include acquiring new skills. Jack Easdale, SVP Revenue Management & Enterprise Analytics at Las Vegas Sands Corp., recently told me that he has used this time to encourage his team to ramp up analytical abilities. He licensed an online learning tool called DataCamp for all the members of his team and challenged them complete courses pertinent to their responsibilities. He recognized that humans will need to take a more assertive role in managing the systems and mathematical models they use for the Venetian, Palazzo and Sands hotels.
Finally, look beyond articles on Revenue Management to comprehend the economic, financial and societal forces that will change the behavior of your customers. My favorites include The Financial Times and The Economist. Both of these have ongoing, comprehensive coverage of the coronavirus outbreak as well as its effect on commerce. In addition, they continue to publish insightful essays on the economic consequences and political and social impacts of the pandemic.
Hopefully, since your databases have been trashed, you’ve already questioned the automated rate and inventory adjustments made by your Revenue Management systems. You will have significantly narrowed the system’s floors and ceilings, if you allow automatic adjustments at all.
The data will eventually begin to heal itself, but you will need to take an active role in assuring that your data and systems can recover as quickly as possible. With appropriate intervention, the dependability of your Revenue Management systems can recover faster than the economy. You need to start working now towards their revival so that they can help you manage the post-COVID environment. Here are a few of the questions you need to ask your systems providers or database managers to speed up the process:
- How will you remove “corrupted” historical data while retaining the data that still has probative value? How can you assure those fixes will work in the future for a second wave of infections?
- How can we save the invaluable historical seasonality components in our data? How do we assure that future seasonality elements are not corrupted by current events?
- What external data (e.g. STR data) can we use as a proxy while waiting for our data to heal? Are there any macro-economic indicators that we can use as leading indicators as the economic recovery rolls in?
- How will you account for geographical and sub-market differences in the adjustments you make?
- How will you systematically assure continuity and reliability of the data going forward?
In addition to database adjustments, we must assure that we have enlightened human guidance modifying forecasting and optimization algorithms. More questions:
- How will you adjust your forecasting models to account for this unprecedented disruption? Which variables are most impacted, and how will you account for that impact?
- How will your forecasting models “learn” faster in this chaotic environment? What changes will be necessary for the model’s error-checking logic?
- Measuring customer price sensitivity will be more important than ever. How will you change segmentation models to account for changes in price sensitivity? How will we be alerted to anomalies which may require human override?
- Have you considered the impact potential “social distancing” policies could have on capacity? How will variable market capacity affect price & inventory recommendations? This is huge! Many Hotel Revenue Management systems have not generally had to deal with variable capacity, unlike the airlines. Please make sure you address this. Having the supply side (capacity) in play may wreak havoc on many pricing models!
- If competitors’ prices are an input to price optimization, how do we assure that “panicked” prices by our competition do not adversely impact our post-COVID pricing strategy? In other words, how do we prevent an automated downward price spiral?
A number of articles advising Revenue Managers about post-COVID response exclaim: “Stay the course!”
Don’t even think about it. There is no course to stay. It’s been obliterated.
Many people in Revenue Management will be scrambling to “get back to normal.” The shrewder ones will be actively seeking “the New Normal.”
While others are desperately seeking “normality,” the most astute Revenue Managers will stay nimble and flexible to gain share. They will be the ones who are able to, in the words of Steve Jobs, “Think Different.”
The brilliant and insightful people who are “thinking different” realize that there will be no consistent recovery. They will have the greatest success. They won’t be looking for any sort of “normality.” Unprecedented disruption has occurred. As devastating as the near-term impact has been, the residual effect on the economy and, in particular, travel is completely unknown and unknowable. Successful people will move quickly and adapt to volatility.
I’d like to leave you with a few ideas that I will flesh out with more detail in an article next week:
- Think Revenue Creation—As Revenue Managers, our primary job has been to manage demand; we now need to be in demand creation.
- Rethink Customer Segments—Different customer segments will recover at different velocities. If your primary segments aren’t going to recover early, you’d better pivot fast.
- Look Upstream—Don’t wait for bookings to show up in your pace reports. Look upstream for signs of life in your various market segments.
- Change your Channel Strategy—You’ve been working so hard to wrest control of your customers back from OTAs and to reduce costs. Sorry, but they will need to be your best friends in the short term.
- Leapfrogging Tech Stacks—Huge, complex Revenue Management systems with rich functionality had been great. Advances in technology have enabled smaller, nimbler systems which may be better for this new environment.
Rise up, Humans!
This unprecedented situation calls on people to do things that a computer could never do. We need to call upon our uniquely human traits. Use our wits. Try new things. Assess risks and make judgement calls when there is absolutely no data to support our decisions.
We can do this. We’re trained & ready. Let’s roll!