The crux of the hospitality and guest experience industry lies in welcoming guests from all over the world, no matter what language they speak or which culture they imbibe. Every establishment in the industry - hotel chains, motels, independent hotels, B&Bs, hostels, restaurants, destination getaways - all of them need to understand their customers better so they can improve their hospitality services and ensure that they are always ahead of their competitors.
For this they need to analyze how guests compare them to their competitors, which aspects of their services they are happy or dissatisfied with, and if their experience will result in making them into return-customers. To get truly accurate insights, a sentiment analysis platform for guest experience needs to analyze the voice of the customer data natively in the languages the customers have left the reviews in. Translations can mean diluting the essence of the content and as a result give false positives or negatives.
The client is a Europe-based, award-winning guest experience and reputation management solutions provider to hotels and restaurants the world over. The company offers cloud-based offerings so that its clients can get a better understanding of their individual brand reputation performances as well as their strengths and weaknesses. The client was looking for a genuine sentiment analysis platform that could analyze non-English languages accurately, at speed and scale. Since other platforms in the market used translations to analyze multilingual guest feedback data, the company was unhappy with the intelligence results.
Repustate provided the client with a customized, multilingual, aspect-based sentiment analysis solution for hotel reputation and guest experience management. The machine learning platform could analyze data natively and semantically in all the 23 languages that Repustate supports, including Arabic, Spanish, Italian, Korean, Japanese, and German. When it came to social media data, the model could read and analyze colloquial words and phrases, code switches, industry jargon, hashtags as well as non-text content like images and emojis. This improved the overall accuracy since no details were left out.