What Do Machine Learning and Donuts Have In Common?

28 Aug 2018 by ReviewPro in Case Studies

How do most restaurants choose what to serve their guests for breakfast, lunch and dinner? Probably not via a semantic analysis algorithm capable of machine learning, right?

One hotel located in the Canary Islands did just that with a little help from ReviewPro’s customer success coach, Janire Rodriguez.

Robot hand and a donut for Semantic Analysis

Cristina Las Palmas is a luxury hotel well located in the heart of popular tourist destination island Gran Canaria. After noticing that that staff weren’t as active on their online reputation management dashboard as they should be, ReviewPro’s Customer Success department contacted them to offer some coaching.

Upon performing semantic analysis on feedback from guests, Janire quickly realized that the property was underperforming in one very key area: food and drink.

Semantic analysis of guest reviews

Hotel representatives Macarena Marrero and Elisa Martin, Revenue Manager, worked in collaboration with Janire to identify the following improvements:

• Improve the signage for hot dishes
• Improve the breakfast service
• Add juice and donuts to breakfast

After two months, guest satisfaction ratings related to the Food & Drink department radically improved, shifting from ranking among the five worst categories to achieving the top five best categories. Who knew donuts were so important to travelers?

Semantic Analysis Categories

As you can see in the graph pictured here, the satisfaction index for the Food & Drink category improved significantly from 75.40% to 80.10%. The upgrades also had a knock-on positive effect on the perceptions of value guests had of the hotel.

graph showing improvement in food and drink rating

What is semantic analysis?

Semantic analysis is a machine learning solution which uses natural language processing to understand the meaning of guest comments. Textual feedback is broken into concepts and assigned a positive, negative or neutral sentiment. Concepts may be broad, like “breakfast” or “parking,” or specific, like “curtains” or “bill”. By analyzing concepts and looking for patterns, hotels gain insights that ratings do not provide.

ReviewPro owns an exclusive semantic analysis algorithm that analyses 300 million reviews per day in 45+ languages over 175+ OTAs and review sites, offering unparalleled insights into what guests are really saying about your property.

Would you like to see how semantic analysis works?