How Predictive Analytics Can Help Set Perfect Hotel Room Prices

Somewhere between the occupancy rate, future booking trends, the weather forecast and other factors lies the perfect price for any particular hotel room. But few hotels have mastered the art of setting it.

One of the primary performance metrics that hotels consider is RevPAR, or revenue per available room. Hotels divide their total guest room revenue by the room count and the number of days in the period being measured and that provides an indicator of how well the hotel is doing.

Hotel revenue managers often compare their hotel’s RevPAR with that of other hotels in the same market, of the same type and with the same target customers. Some data companies even use voluntary surveys to collect RevPAR across markets, before compiling it, sanitizing it and selling it back to hotel managers. Then hotels use that data and their own reservation information to set their prices.

Generally hotels set a rate for their base rooms and then increase the price by a set percentage for types of rooms with more features. When those hotels need to move the needle on RevPAR, there’s really only one tool left to them: Fill more rooms.

Big Data and Predictive Analytics Boost the Bottomline

Craig Weissman, CTO and co-founder of revenue strategy solutions startup Duetto, and former CTO of Salesforce.com, believes big data and predictive analytics can revolutionize this part of the hotel business.

“If we can help them optimize price, it blows almost entirely to the bottom line,” Weissman says. “That could be hundreds or thousands or even millions of dollars a year for that hotel.”

[Related: How UPS Uses Analytics to Drive Down Costs (and No, it Doesn’t Call it Big Data)]

Because the margins in the hotel business are so narrow, Weissman notes that even a 6 percent movement in RevPAR could double a hotel’s profitability.

“The industry has become a little too focused on filling rooms at any cost,” he says. “They’ve gotten obsessed with filling rooms but not with raising price when it’s appropriate. The profit in the industry is not so good. They sell out too early on too many discount channels. It takes a certain amount of trust in the system to withhold rooms and trust that you’re going to fill out at a higher price point. Those types of decisions can make a massive difference to your bottom line.”

To help hotels optimize their pricing decisions, Duetto gathers data from a multitude of sources: web shopping regrets and denials, the most popular days search for on booking sites, social review, air traffic into a hotel’s city, weather forecasts (even forecasts for surrounding cities) and more.

Hotel Data ‘Nicely Structured’

“We need to pull three years of history out of the hotel systems,” Weismann says. “The good thing is that reservations are pretty important multi-hundred dollar transactions. They tend to be very clean, nicely structured data.”

[Related: 3 CIOs Reveal How They Got Started With Predictive Analytics]

Duetto pulls all that data and forward reservations into its high-performance, multitenant enterprise data warehouse and analytic engine built on MongoDB (and Amazon EC2). MongoDB is used for both the metadata store and Duetto’s Star Schema Analytic pipeline, which includes a bulk ETL engine and analytic server. Its cloud application, Duetto Edge, draws on this data and uses a combination of algorithms and business rules to present visualizations that revenue managers can use to make decisions.

“We are not yet in a world where it is totally on autopilot for hotels,” Weissman says. “You need to combine some human intelligence with an algorithm. Directionally, when our algorithm detects an upward trend, we want to increase the price. At the same time, we have guardrails so we don’t precipitously drop the price when demand is low.”

Follow Thor on Google+