Skip to content


Reimagining Demand Forecasting For The Hospitality Industry

The pandemic made it obvious (and painful) to hoteliers that in any situation that poses uncertainty it is crucial to quickly develop and deploy strategies to adapt to both short- and long-term shifts in demand. But here’s the thing. Travel demand has always been uncertain. And history never repeats itself. Recent events just highlighted this fact for all of us and pushed the Revenue Management discipline towards further innovation in forecasting and optimization.

At the start of the COVID-19 pandemic, I was having conversations with industry experts and hotel company executives who were sharing their frustrations about how the RMS tools they were using completely fell apart in a highly unstable environment. Many of those tools had rigid algorithms using the same approach and logic that was built in from the beginning, specifically: extrapolating STLY (same time last year) data onto the future to make forecasting and optimization decisions.

But, as it’s now obvious for everyone, last year’s data quickly became irrelevant for forecasting.

In addition to that, it’s also clear that as we go into recovery, any hotel looking at last year’s COVID-drained numbers for guidance in predicting demand isn’t going to get any reasonable results. 

Because demand has always been uncertain, and history never repeats itself, in the new world, we as an industry need to find a better way to forecast and optimize.

Finding A New Way To Forecast

In early 2020 it became clear that forecasting and optimization algorithms had to be reimagined. Responding to a constantly evolving demand landscape requires an intelligent and flexible revenue strategy platform that quickly learns and adapts to myriad factors that impact the decisions you make. A platform that is agile and adaptable.

Here at Duetto we strongly believe in agility. Throughout the initial design, which came on the heels of the recession of 2008-10, the Duetto platform was originally built with the necessary flexibility to allow it to easily adapt to a changing environment, be it recession, pandemic, or other.

Because of this Duetto helped thousands of hotels to navigate through the pandemic and will continue helping them thrive as we come out of the crisis and the industry finds its new balance. 

How did we do this?

Due to the flexibility of the algorithm, Duetto was able to solve the forecasting problem for our customers immediately after we saw the first signs of the chaos that continued throughout 2020 and further into 2021. We instantly reacted by adapting the algorithm to the new uncertain environment. We changed the logic to put heavier weight on short-term history (2, 3, or 4 weeks of the immediate actualized hotel data) and forward-looking market demand signals to provide more accurate recommendations. 

Our new ‘short-term forecasting’ updates were pushed live to all hotels within a few days, which allowed Duetto to immediately pivot and help our customers navigate through the crisis.

MAD revenue

The graph above shows measurements of the accuracy of our forecasting algorithm over time for our entire customer base. 

  • Y-axis is the accuracy metric (we used Mean Absolute Deviation for the purpose of this exercise), for revenue forecasts that are 14 days before arrival.
  • X-axis reflects stay dates. You’re looking at the full 16 months from July 2019 through December 2020.

What we see here is that until March 2020 the algorithm was producing stable results based on the logic that was widely adopted by the industry before the pandemic. Then (in the middle of the graph, March 2020) - we started noticing the chaos in the data caused by the pandemic: demand instantly dropping, due to various government regulations, hotel closures, and guest concerns.

In just a couple of weeks, we adjusted the algorithm to be better adapted to the new environment and immediately pushed the updates to all customers.

What we see in the right portion of the graph is that after these changes were implemented, the predictive algorithm normalized and actually started performing slightly better than it was performing before the crisis.

It’s pretty awesome. Scientific magic in action.

Hoteliers Adopt Short-Term Forecasting

Our hotel clients today are achieving better performance because of this switch to a short-term forecast. Here’s what some of them have to say:

“We’ve been using the Duetto short-term forecast looking at two weeks of data. Once we re-opened we pushed out to three weeks. As it improves we will extend this to have as much data flowing as possible. And we will go on like that. Because Duetto looks at future data much more than at history, I think that Duetto is best placed to push a hotel in that recovery.” - Nicolas Stockman, Revenue Manager, Hotel Arena, in our blog Getting Ready For The Tipping Point.

“We adapted to the short-term forecast so that Duetto puts more weight on short-term pick up than on same time last year comparisons.” - Kris Vanaerschot, Chief Commercial Officer at Helios, in our blog Tech Investment & Social Media Boost Helios Hotels.

“It is hard for any RMS to forecast this [a pandemic] but as the forecast is based on the last few weeks it meant it ignored totally last year and only looked at very recent historical data. Given how dynamic the situation was, we didn't spend too much time looking at forecast accuracy and only used the Duetto short-term one.” - Emmanuel Lacour, Group Director of Revenue, edyn, in our blog Re-Strategizing Revenue Keeps edyn Ahead.

And we’re not stopping there. 

We continue heavily investing in improvements to our forecasting and optimization algorithms and other analytics-related initiatives to support our position as the innovative industry leader. We have a team of industry experts, scientists, data analysts, and engineers working around the clock to continue perfecting our product offering in an effort to help the hospitality industry become more efficient, more successful, and more resilient.

Want to find out more about how Duetto’s short-term forecasting could help your revenue strategy? Contact Us today to learn more.

Back to Blog

Speak to an expert

Contact us