About the customer

  • Our client here was a 5 Star Luxury hotel, With over 11 award-winning luxury hotels spread across key destinations in India.
AI/ML Case Study




Neural Network, Deep learning, Python, LSTM, Google Bert, NLP, PANDAS, NUMPY

Project Duration

3 Months and Currently in a maintenance phase


Project Details

  • BigOhTech developed a product to capture the sentiments of the Users and provide insights to the Business owner on how the different vertical and Business Units are performing.


  • Starting with the hospitality sector primarily it has been a daunting task to capture the feedback and understand the intent of the sentiments
  • Feedbacks were getting captured at many a places, such as online reviews, Social Media posts, and using the regular feedback forms. This leads to too much source but no centralization and no analysis on the reviews
  • Business owners specially C Level executives have found it quite difficult to capture how a particular business venture is performing
  • Another problem was the availability of the data, Even if the data was available it was scattered at too many places. Almost impossible to fetch the exact intent of the reviews and couldn’t get a clear picture
  • Another issue was “Gibberish reviews”, Over the time there were lot of gibberish reviews that got captured and were just eating up the space, No use for the business
  • No dedicated dashboard to measure the performance of a particular business vertical, As feedbacks and the reviews were cluttered at many a places
  • Business expansion was a problem, No dedicated area where a business could grow
  • More than 40 % of negative reviews were left unactionable

Our Approach

  • The hotel chain used sentiment analysis to analyze customer feedback data from various sources such as online reviews, social media posts, and customer feedback forms. APIs were designed to capture these all reviews and map it back to our own database
  • PANDAs was being used as a DB to preprocess and analyze large volumes of text data being received
  • It was also performing the pre process and cleansing of the data to remove the noise that was their in the data
  • Google Bert was also being used to pick up the exacts words from a review that was on similar lines to our use cases
  • Post that Bert was classifying the set of data as positive, negative or Neutral
  • CNN was used to be trained to learn features from different parts of a sentence, This was helpful in training the data set as well
  • Deep learning models were used to learn complex and non-linear relationships between input features and output labels, making them well-suited for sentiment analysis tasks.
  • Based on the results, the hotel chain identified areas that needed improvement such as slow service, outdated facilities, and unclean rooms.
  • They invested in staff training to improve service quality, renovated outdated facilities, and increased the frequency of room cleaning.
  • They also used the positive feedback to promote their strengths and unique selling points in their marketing campaigns.
  • By analyzing customer sentiment and taking corrective action, the hotel chain improved the overall customer experience and enhanced their brand reputation.
  • Single unified dashboard was used to capture all the data and view the overall sentiment of a particular business within just 2 clicks


  • Responding to reviews can see a 12% increase in revenue per available room (RevPAR) compared to hotels that do not respond
  • 12 % increase in a month revenue was found due to taking action on the feedbacks that were being received
  • Guest Loyalty rate saw a surge of more than 10 % due to responding the feedbacks