Data, data, data. It’s all you’ve been hearing about for the past couple years.
And now that we’re amid a travel pandemic forcing hoteliers to look for additional insights on any available demand, data mining has moved from “important” to “critical.”
But, as we love to ask these days: What does it all mean? What data is out there, where can hoteliers find it, and – once they have it – what can they do with it to run a more successful business?
Often, the idea of collecting and analyzing data falls on the lap of a hotel company’s revenue team. Revenue leaders are tasked with seeking out the most relevant data for them, analyzing it, and then actioning it.
They often lean on tools – revenue management systems and business intelligence platforms – to help.
These tools are crucial simply because there are so many data points available that determining which are most relevant to your property and then crunching the numbers to find the best results simply cannot be done by a human.
As a simple example, imagine you were considering raising your Best Available Rate by $5. What effect would that have on the demand for each of your segments? For each of your channels? How would your competitors react?
A computer can run millions of “what-if” models in one second and spit out these answers to help you make the right decisions.
This, in essence, is the basics of “data science.” It’s about using data to create as much impact as possible for your business, whether that’s optimizing the business more efficiently or building data products more intelligently.
Data science typically follows the following process:
- Collecting hundreds of thousands of data points
- Exploring and transforming the data
- “Cleaning” the data to detect anomalies and determine what matters most to your specific problem
- Analyzing the data
- Determining the best way to apply “deep learning” to that data, whether it be through neural networks or other models.
Taking it a step further, perhaps the most important piece to data science is that machines can find and ask questions of the data that you may not have even thought to ask.
Data science applied to revenue management
Revenue management systems run the gamut from basic pricing tools that raise/lower price based on occupancy to full-fledged analytics platforms that ingest and analyze all sorts of relevant data sets and help hoteliers make decisions across their entire business, including but not limited to pricing.
When it comes to pricing recommendations, algorithms are trained to analyze all of this data with one goal: to maximize revenue. But the data sets that might be relevant to one properties’ algorithm might not be relevant for the next.
Take weather data, for instance: A resort destination in a hurricane-prone area might use weather data to determine upcoming demand, whether that means you’ll see an uptick from people fleeing a storm or that it’s time to board up the windows. But for a hotel in Chicago, weather data may have little relevance on business.
This is where data science shines as it runs hundreds of calculations in seconds to ideally determine which data sets are most relevant. While same-time-last-year data may be largely irrelevant, data measuring seasonality is still important. Expected demand on a Monday is not the same as expected demand on a Saturday, for example.
Data science will help your system understand this and improve your pricing algorithm by “weighing” the data inputs based on their relevance to your individual property.
Therefore, while a BI tool will help you visualize the data in an easily digestible format and even share with owners and other interested parties, it’s most likely not applying data science and deep learning to determine which data is most important and make recommendations that will affect the outcome of your business decisions.
For this, you’ll need a full-fledged revenue system.
Data science and forecasting
There’s been a lot of talk throughout the industry of ways to improve forecast accuracy, particularly in these challenging times when historical data is of little importance.
If there’s anything hoteliers need, it’s improved visibility into future business. Again, data science is critical here as it ingests all of the available data and “cleans” the data that doesn’t matter.
Most revenue teams depend on two types of forecasts: A demand or occupancy forecast and revenue or financial forecast. The former helps hoteliers understand what their demand will look like in the future, from tomorrow to as far as 365 days out.
The latter applies dollar amounts to those calculations, bringing in additional data sets to layer on top of the demand forecast to estimate how much top-line revenue the hotel will bring in on that day.
Hoteliers today are increasingly looking at additional data sets to help improve their forecasts.
Typically, a forecast is based on reservations data from the PMS (historical and on-the-books), as well as demand and pricing data from a hotel’s pre-determined competitive set. A demand forecast can be constrained, meaning it’s calculated based on the maximum amount of supply (number of rooms) available, or it can be unconstrained, based on the assumption there is no limit on supply.
To form a better-unconstrained forecast, hoteliers are increasingly looking higher up the funnel and ingesting pre-booking data like search results and flight reservations.
How much your competitor’s data – mainly their demand and their rates – determine your individual hotel’s forecast and strategy has been up for debate. Before COVID, when all hotels had a steady stream of demand to analyze, one could argue that hotels should focus on their value drivers and not make decisions based on their competitors’ actions.
But, today the actions of your competitors should weigh in on your strategy. Again, this is where revenue teams can lean on data science to determine the relevance of those data sets and provide recommendations based on their value.
Taking the first step
It might be tempting for a hotelier to trust they can import all the necessary data sources and analyze them on their own. Revenue teams understand pace and pick-up pretty well.
But it’s not aggregating the data that provides the “magic,” it’s the coding built into the data modeling and algorithms that ensure the data sets are relevant and allows systems to run millions of calculations to determine what the best outcome for the hotel would be. This requires a specific infrastructure built to store and process the data.
And data science can and should be applied to other fields outside of revenue. While data science has given us dynamic pricing, the next evolution will be dynamic management – not only adjusting price to changes in the environment but also continuously adjusting management of the hotel, including marketing, sales and operations.
Revenue management will continue to be a mix of art and science. The science portion is maximized with more complex data processing and optimization models, which will allow revenue managers to dedicate more time to strategy.
Data science will never replace revenue management, rather it will allow the revenue team to evolve.
David Moneo is vice president of AI initiatives at SHR.