An industry survey reports, “91% of consumers are more likely to shop with brands who recognize, remember, and provide them with relevant offers and recommendations”. In a digital ecosystem, personalization has become a key business driver, as it offers contextual customized experiences. From advertising to marketing results, branding to improved website metrics, digital sales conversion to revenue growth, and improving customer satisfaction - personalization plays a major role around various aspects of the customer journey.

Our client is a leading omnichannel travel company in EMEA that offers a wide variety of booking services spanning hotels, flights, holiday packages, etc. With the advent of modern technologies, they have expanded their services across various touchpoints, leveraging digital customer experience for an enriched customer journey.

Their personalized recommendation was inaccurate leading to poor conversion rates, so they were looking for ways to improve personalized experiences.


To offer a relevant personalized customer experience and increase revenue, we built an AI/ML-based personalization system by:

  • Creating a data lake to bring the omnichannel data and capture user interactions in near real-time to quickly adapt and act to changes in user behavior.

  • Providing personalized recommendations and insights based on user demographics, past transactions, usage behaviors, and past searches, etc.


The Requirement

Our customer wanted to increase the user engagement and conversion rates for the booking of hotels, flights, and their holiday packages. To make this happen, they wanted a custom end-to-end platform and solution that enables personalized user experiences leveraging machine learning and artificial intelligence.

The Challenges 

The client faced low conversion rates and less user satisfaction due to these challenges:

  • Inefficient flight recommendations leading to low click rates
  • Lack of recommendations for hotels and holiday packages
  • Outdated technologies and disparate systems that did not help in transforming usage and click-thru’ analytics to personalized experiences
  • Absence of a systematic data pipeline to pass on the required data to the personalization system on time and get the relevant recommendations

The Solution

Srijan created an AI/ML-based personalization solution with the below key features:

  • Personalization and backend integration that built more relevant and personalized customer experiences for booking services.
  • Personalization enabled for these user personas - New unknown, Returning unknown, and Registered returning users
  • Provides personalized recommendations and Insights based on usage, the user (Demographics), and transactional data - ‘Searches, Views, Similar users, Most popular, Similar content’ etc. leveraging machine learning, NLP, and deep neural networks at its core.
    • Personalized content and recommendations are shown on Home page or other pages (thru’ Widgets) 
    • API-based Integration with CMS to serve the right recommendations with their content
    • Configurable parameters by content editors to edit the recommendation parameters easily
  • Trigger associated campaigns to drive more bookings (at offer price points) and push notifications for cart abandonments

Here is an overview of our implementation. We built:

  • An API-based solution to provide personalization (both Containerized and AWS Managed solutions) 
  • Data Lake to capture JS-based user activity data (light-weight JS) as well as CRM and Transactional systems integration thru’ an API approach
  • Backend layer with Caching as appropriate for specific needs;  Personalization “Gateway” to route this to Caching layer for near Real-time recommendations
  • Automated and ongoing ML Model training and refinements based on newer data 
  • Case study -1

    Fig: Personalization System interaction with CMS

    Case Study-2

    Fig: Event-Driven Model Training

Our Approach

Our solution revolved around improving the customer experience by better understanding their needs through data-driven methodologies. We leveraged the cutting edge deep learning techniques and natural language processing to:

  • build profile vectors for known users based on their implicit and explicit feedback 
  • the item vectors/embeddings of all resources including hotels, flights, packages, etc. 

Our ML algorithm also was equipped with various custom techniques to handle the cold start situations for the new users as well as unknown/anonymous users with no previous history/usage patterns. To capture the near real-time events and adapt the algorithm quickly over the new behavioral patterns, we built our own custom activity tracker leveraging vanilla javascript. It collects the user activity and pushes it into the data lake so that the personalization system can update the results of the recommendations based on accommodating those near real-time patterns. Our personalization system helped the client to effectively map the right resource to the right consumer which further helped them achieve good conversion rates and much better engagement.

Tech Stack

  • Python
  • Natural language Processing (NLP) and Deep Learning
  • Vanilla Javascript
  • AWS Cloud - Managed Services, Serverless, and Elastic Compute instances

Business Benefit

  • AI/ML-based personalized recommendations for immersive exploration and booking experience
  • Increase in customer delight and return rates to the site, driven thru’ associated campaigns 
  • Ability to leverage near real-time feedback to enhance user experience
  • Improvement in brand awareness and customer loyalty
  • Overall increased conversion rates by ~14%
  • Enable content editors and business managers to easily configure personalization parameters


Subscribe to our newsletter