Has forecasting become a lost art?
— 10 experts shared their view
Robert Cross, the de facto Godfather of Revenue Management, stated "…everything starts with a day by day, segment by segment forecast", yet many hoteliers no longer produce one as part of their strategic process. Most of today's advanced revenue management systems do not generate one, instead of leaving the user to rely on a total hotel forecast and to trust the "black box" for the details.
Should hotels be producing a detailed long-term forecast, and if so, what are the benefits of doing so? How far out should they forecast and how often should they do it? The automated systems (only 15% of hotels have one!) can be very accurate at projecting the optimized results of an existing strategy, but are they effective when contemplating a change in strategy, as an increase in group rooms, eliminating an airline crew, or even a renovation? How do hotels that don't do it satisfy other stakeholders like senior leadership and owners, who may have an interest in understanding the detail behind the hotel's forecast, especially when recent performance has not met expectations?
Corporate Director Revenue Management Palladium Hotel Group
There is a very popular sentence in the marketing world... "Cash is King and Brand is Queen", in our little world call Revenue we could say "Forecast is King and YOY is Queen", with that I would like to point the importance of the Forecast in our daily task. From my point of view, the use of technology is going to change the way we make a Forecast ... we will have a faster and sometimes better accurate data, the use of Big data and artificial intelligent is going to kill the "ancient art" of forecasting and probably we are are going to be replaced by machines with no soul and no art, but better data.
On the other hand, sometimes I have the feeling in this VUCA world is becoming really difficult to make a decent forecast, the new art of this Revenue world woulb be improvising new strategies and have the capability of changing that in minutes.
CRME, CHIA, Chief Commercial Officer at Remington Hotels
In my opinion, day by day forecasting is even more important now than ever. At Crescent Hotels and Resorts, we believe this is an integral part of our strategy, as it conveys to our stakeholders our strategic intent by day by market segment. The forecast is produced at a minimum once a month for 90-120 days out and full-year for major group updates. Revenue management systems have become very accurate at projecting occupancies, but not at the segment level. This makes the system generated ADR less accurate, which is why our Directors of Revenue Strategy drive this process. Once we have compiled our internal projections, we compare our numbers to the system generated occupancies to ensure a correlation between them and help us validate our numbers.
The forecast process satisfies our stakeholders for operational and financial reasons. Operational departments are reliant on occupied rooms as a metric, as opposed to total revenues. We also have a financial responsibility to our owners, especially when it comes to profit. We must focus on implementing the right segmentation mix and channel strategies, as they have an impact on the cost of acquisition. As a discipline, our main goal is to drive revenues through our ability to evaluate data and implement strategies based on our findings. We believe that the forecast process allows us to do that in a more effective way.
Vice President HSMAI Asia Pacific
Forecasting is the bread and butter of revenue management because all strategic decisions should be driven from it. Having done a variety of forecasts over the years, the most effective ones are still a day-by-day forecast which drills down to the market segments because each day of week can behave quite differently. For hotels with limited resources and time, it is really tying the forecast strategy to the rate-change strategy. At the minimum, I find that you need to do at least a 1-month out perspective so that you know what's happening.
What's more important is how you use the forecast to drive marketing efforts, pricing decisions and actual actions. If the forecast is scheduled into your decision-making rhythm and meetings it won't feel so irrelevant.
Principal | Revenue Generation, LLC.
It is very important for hotels to maintain a detailed forecast for both the short term (more frequent) and longer-term (outside 3 months). If your property has an automated system that provides demand information, it will likely only be reliable for the short term. It is just as important to understand longer-term trends for your particular market as well as segment trends that could significantly impact your overall profits.
This becomes even more critical in today's environment as we are in very uncertain times and trends change on an hourly basis. This is an opportunity for revenue strategy folks and analysts to instill confidence by paying attention to the changing trends and providing this information on a regular basis to key stakeholders.
Program Director, Master’s Degree in Hospitality Strategy and Digital Transformation at Les Roches Global Hospitality Education
When you take a step back and consider the purpose for the forecast, the need to employ both technology and human brainpower becomes apparent. Is it tactical or strategic? Are you forecasting to measure and adjust yield management controls or are you forecasting to validate and optimize your channel and market segment mix?
Just like I wouldn't debate a self-driving car's ability to consistently stay within a lane and follow at a given distance, yet I would not empower it to decide whether I should go surfing or play golf, automation is better at the tactical while human intuition is better at the strategic. This is determined by the level of detail, repetition, and intuition required to perform each of the tasks.
The reason automation is better at the tactical is that a person can define the parameters once, for example the expected booking pace of the important market segments and the overall available capacity, and a system can monitor the progress, identify the small adjustments and send the various communications needed to make sure the decision is reflected everywhere the hotel is sold. Like “keeping the car within the lane.”
There is no substitute for human imagination (thankfully not yet, at least), for considering the eventual outcomes of what-if scenarios that have no historical precedent. At this time, it would be nearly impossible to rely entirely on automation to decide if and how a hotel should reinvest and reposition. I have yet to see an automated system that could get anywhere near an accurate projection for unconstrained demand without a decent amount of intelligent human intervention.
We are currently in the midst of such an event with no historical precedent. The meteoric decline that the industry has experienced is going to drive some conversations unlike any that hotel walls have heard in at least a decade. Hotels and revenue leaders that have allowed a long run of good results to excuse themselves from maintaining a disciplined long term, day by day, segment by segment strategic forecast process may find themselves without one of the most valuable tools in tough times like these to justify their own existence.
Multi-Property Revenue Manager
The forecast used to be a great and realist KPI. However, its use has been more and more distorted to serve communication purposes and often does not reflect reality anymore. For Revenue Managers it is of course still key to have one, but not a formalized one anymore. Before taking any actions you would try to calculate what would be the revenue impact (the forecast improvement). This, however, does not mean that you need to have a daily, per segment, per channel forecast written down in an Excel spreadsheet and forwarded to the entire organization.
So an RM forecast is still needed to conduct a strategy and if it is not distorted by any other stakeholders, yes a formal one can be useful. Otherwise, time and energy will only be wasted.
Forecasting hasn't become a lost art. The revenue managers that my colleagues and I speak to every day are constantly refining their methods and eager for new technologies and data to help with the task, and they're well versed in the principles. But - and it's a big BUT - through lack of better systems, there's a sense that they've become over-reliant on data that only forms a part of the picture, such as on-the-books (OTB) and same-week-last-year (SWLY).
In normal times, this is inadequate; in these unprecedented times, some of this data is no longer at all relevant for some markets that have been hit hardest by travel bans and lockdowns. This is where we need to tap into better predictive analytics technology, which draws external data from disparate sources to infer demand in a more forward-looking way. Hoteliers will instinctively know how to use this when such technologies emerge. So the art hasn't been lost; the technology for volatile times needs to catch up.
Without a forecast it is difficult to understand all of the revenue opportunities for a hotel and how to weave them into a single, successful revenue strategy. Creation of this strategy has two components. The continual, repetitive forecast update process that is needed to adjust tactics to make the strategy come true is clearly better accomplished by automation. Here software can both save time, and create better forecasts by consistently applying the will of the revenue manager.
But the human brain is far better at creation of strategy, especially in the face of change. The Coronavirus presents an example of a situation with many (unknown) unknowns and potential uncertain future shocks (and even elements of game theory relating to the surrounding hotels). It is difficult to accurately program an algorithm to understand these dynamics.
At RoomPriceGenie, we believe that software will, for the forseeable future, always be a tool of the revenue manager to automate their strategy, rather than a full replacement. We also believe that the more transparency about the calculations, the easier it is for the revenue manager to add their own strategy and react to unusual situations. Our software is built with this in mind.
TLDR: Strategy is for the humans and calculations are for the machines.
The forecast is a fundamental piece of the hotel puzzle because it helps hoteliers, owners and stakeholders better understand the future, challenge themselves, support operations from a staffing and cost perspective and build credibility for the business. On its own, however, a forecast is just a number, and a constantly shifting one at that. Like a puzzle, a forecast requires many other elements in order to be complete. Specifically, two: clean, relevant data and a clear strategy. The first is delivered by technology—something impossible to do manually because of the vast, ever-increasing amount of available data. The second can only be done by humans, revenue managers in this case. This is where the synergy—not dichotomy—between human and machine wins.
Advanced revenue management system (RMS) solutions have the power to adapt to any change in the market and compare similar exceptional behaviours in the past at a rate unachievable by humans. The machine algorithms continuously learn at a faster pace day after day, and accuracy will only improve going forward.
In disruptive times like these, a hotel's perseverance depends on precise forecasting through predictive analytics, enabling revenue managers to rethink and adapt current and future strategies, which can be then fed back into their respective RMS, completing the forecasting puzzle.
Founder/CEO at The Revenue Company
Forecasting is critical in the process of a running any highly optimized business. In fact, there are many types of forecasts: demand, financial, occupancy, operational, etc. that do a variety of functions to a variety of stakeholders over a variety of time periods. Forecasting should have 2 pillars to take hold in the culture of the optimized business.
1. Detailed. All forecasts should be as detailed as possible. Day by day and segment by segment is a good start, but this must be easily rolled-up for summary discussions that do not need to see the detail. In addition, making the forecast customizable by time period is also key to make it an effective tool.
2. Simple. Simple to create, change, and execute. All forecasts should come from the same platform and the information needed can just be extracted and imported. Forecasts need to be optimized frequently, sometimes multiple times per day. All forecasts should be completely automated with the ability to make manual adjustments to account to fluctuations in demand that the machine cannot know.
Any organization not investing in automated technology to forecast, I believe will probably spend too many man-hours in the forecasting process to do it correctly. The machine can it do it faster and more precise when set-up properly.
Here is a good rule of thumb on a forecasting schedule:
a. Hourly strategy forecasts focus on immediate booking windows and demand changes.
b. Daily forecasts are usually operational in nature and done within the payroll period and weekly for the upcoming payroll period. Usually a 4 day update to the weekly forecast.
c. Weekly forecasts to address the upcoming payroll period.
d. Weekly strategy forecast snapshots the next 12 months (R12) to focus on booking windows and changes in demand.
e. Monthly forecast snapshots are annual or R12 and can go out as far as 5 years usually for financial reporting.