Why Machine Learning Works for the Hotel Industry

Amazon frequently receives credit for successfully employing machine learning to engage consumers and drive sales with its well-known recommendation engine, which generates 35% of the company's revenue, according to McKinsey. However, competitor Walmart has a surprising amount of machine learning activity going on behind the scenes. For instance, Walmart created a facial recognition system that allowed the company to pinpoint customers who were unhappy about waiting in line. The system alerted sales associates that new lanes needed to be opened, which increased customer satisfaction and helped the retailer to manage employee workflow more efficiently. While hotels are, in some ways, worlds away from retailers in terms of the scope of operations and product, the hospitality industry can learn from the experience of retailers when it comes to machine learning, positive customer service, and merchandising.

Machine learning automates the analysis and modeling of large volumes of disparate data, on a scale and scope that is impossible to do manually in a timely and cost-effective way. For those in customer service, machine learning can provide customized recommendations and promotions and, as Shopify notes, optimize pricing strategies in real-time with models that "can take key pricing variables into account, including supply, seasonality, and demand and offer insights on how to adjust... prices accordingly." For revenue managers in hotels or hotel groups, machine learning can provide real-time pricing, offer assortment, and recommendation decisions for bookings and upselling that maximize profit without requiring teams of analysts, and can do this at any point in the guest journey, from booking to pre-arrival to check-in to in-stay.

Historically a front desk supervisor provided a rate table to agents with static room rates for upgradable rooms. Take, for example, a King to a King Suite upgrade offer - the static rate table showed the upgrade could be offered at $100. But with machine learning, an intelligent system would present an offer to the agent based on a high probability of conversion; that $100 per night upgrade should actually be priced at $130, because the model has predicted it's the appropriate price for that guest for that stay. Consider that extra $30 across even just ten two-night reservations each week and a property has brought in an additional $30K+ that never would have been realized using a static rate table. Then consider that the same system can be used to calculate pricing for everything from breakfast add-ons to late check-outs. This heretofore unrealized incremental revenue directly impacts the bottom line.

While the financial benefit of pursuing this unrealized revenue is obvious, there are more significant ramifications for the guest. Netflix uses machine learning to solve a wide range of business and technical problems. Their recommendation system helps viewers discover more content that they’ll like so they’ll keep binging on Netflix. According to Netflix, the machine learning system saves the company $1 billion per year. Why the savings? Because the recommendations increase viewership and keep churn rates down. They lose fewer viewers to competitors like Hulu and Amazon because they make it easy for viewers to get more of what they like without having to search for it. Martech notes that Netflix can spend more money on new content because they know that the viewers will keep coming back. By investing in keeping the viewer happy and engaged, they can improve their product.

The same holds true for hotels. An upgrade isn't just a revenue line item; upgrades create a better overall experience for the guest. Who doesn't have a better experience in a more spacious room where room service can be enjoyed not on the bed but on a table, perhaps with a view? Each of these things—the spacious room, the room service, the view—comes at a price to the guest, but when it comes at the right price (as predicted by the model, which is generated by feeding data into an algorithm) to the right guest, the guest has a better experience, may provide a better online review, and will probably return, perhaps via a lower-cost channel.

Let's go back to the front desk. Most front desk agents receive a commission for upselling generally between 5% to 15%, but many incentive programs give agents discretion to discount the upgrade within a maximum and minimum rate range (a King Suite would typically sell for $600 per night but can be offered for no less than $450, for instance). This widely accepted model is rife with problems, the first of which is that it puts an undue burden on the front desk agent to determine the right offer at the right price for the right guest. Further, it encourages front desk agents to offer the lowest possible rate in order to achieve the upgrade, which reduces revenue potential. Though one would think agents would be inclined to go high, they will mostly go low to get a little commission rather than get no commission at all. Or they will abandon upselling on average rooms, looking instead to make the offer to just a few guests who have the potential to upgrade to the best possible suite. With machine learning predicting the right offer for the right guest, hotels can generate the most upgrade revenue while agents are assured of the maximum possible commission. Their only task is to offer the upgrade at the rate provided and focus on delivering excellent service to the guest standing in front of them.

The impact of machine learning on just this one aspect of hotel operations (upselling) has a ripple effect throughout the organization - increased revenue, happier and better compensated front desk agents, and more satisfied guests. Each of these, in turn, has effects that cascade outward including more operational funds for physical improvements, increased employee retention, and more reliable future revenue from loyalty, all from applying machine learning to upsell pricing.

In times of tight capital budgets, large-scale spending on new systems isn't easy to come by. A more achievable solution is the application of artificial intelligence like machine learning to existing systems to reveal hidden revenue and benefits, the ones have a ripple effect across the organization, that will keep some hotels profitable and healthy in the days to come when economic swings are inevitable. The hotels that join the Amazon's and Netflix's of the world in finding innovative ways to generate revenue while reducing churn rates will survive and thrive.

About Nor1, Inc.

On November 18, 2020, Oracle announced that it has entered into an agreement to acquire Nor1. The acquisition extends Oracle Hospitality's OPERA Cloud Suite by adding Nor1's Merchandising platform that enables hotels to provide personalized offers throughout the guest journey using AI & machine learning, thereby improving guest engagement, and driving incremental revenue and improved loyalty for hotels.

Nor1 is the leader in hospitality upgrade, up-sell, and merchandising technology. Headquartered in Silicon Valley with offices across the world, Nor1 provides data-driven pricing and merchandising products that maximize incremental revenues for Hilton, IHG, Radisson Hotel Group, Accor, Wyndham, and other global hotels and resorts. Nor1′s real-time pricing and merchandising intelligence engine, PRiME®, powers eStandby Upgrade®, eXpress Upgrade™,  and CheckIn Merchandising™,to recommend the most relevant upgrade to the right guest at the right time for the most optimal price. For more information, please visit www.nor1.com, or contact us at [email protected].

Alan Young
Puzzle Partner, Agency of Record
Nor1