Chatbot for Travel Booking Site
Cozmo Travel is a leading travel booking website in the middle east. They were looking for a solution that could help reduce the work load on the customer service team and help customers with their queries. We built a chatbot solution for them that allows their customers on the platform to ask general queries, helps reduce the workload on customer service teams resulting in cost savings without affecting customer service experience.
The company’s customer service teams with getting overwhelmed with the amount of calls for general enquiries and questions. This was increasing the workload for customer service reps and was affecting the customer service experience for other users who were calling regarding their travel and bookings queries. This ultimately affected the customer service experience resulting in loss of revenue and higher customer service costs. The company was looking for a chatbot solution to install on their website which could help users with their general and bookings related queries, increase efficiency and make cost savings whilst ensuring the customer service experience does not suffer.
Chatbots come in a range of different forms. From hard coded bots that have a fixed logical flow to advanced deep learning bots that have been trained from real user conversations and/or simulated user conversations. Different chatbots require different tools and technologies and this particular solution we designed a chatbot system will serve as a conversational bot whose objective will be to help users with their queries around services the company provided.
We built a chatbot solution using technologies like Google Dialog Flow. The chat bot agent was trained with custom dataset that we created to serve as knowledge base. Below were some of the main features of the chatbot
1. Achieve high accuracy in answering user queries with the right information
2. Help users find the right content around bookings and travelling
3. Have different contexts for different intents which can allows the chatbot to gather and
ask user for follow up questions
4. Collect and store conversational data for model retraining
5. Collect different metrics and statistics for monitoring the performance of the chatbot
6. Analytics dashboard to show the quality and performance of the chatbot
7. Deployment as a separate micro-service that can integrate with the website using a chat
Our cloud powered Chatbot System was integrated with the customers website. A realtime dashboard was also provided which the client could access to track different KPIs, predict future trend changes and access other actionable intelligence and insights from user conversations.
After the exploratory analysis and requirements gathering phase, we collaborated with the company managers to gather all the data from various sources to create a knowledge base which was used for training the chatbot for custom user queries and answers. This process allows the chatbot to learn from past user conversations to answer user queries and have a conversation with a user.
The diagram below highlights the architecture of the chatbot.
All the components were built and deployed in the cloud. We used Google Dialog flow for Natural Language Understanding and Dialog Manager and for Natural Language Generation. These components were integrated with the rest of the system which collected and stored different data points for analytics. The whole system was integrated with a chat widget and installed on the company website.
With our AI powered solution, the company saw a decrease in calls for general queries and questions and our solution helped them achieve
- Cost saving resulting from decrease in customer service workloads
- Increase in overall customer satisfaction from an efficient customer service experience
- Access to analytics which helped decision makers provide better service to their customers
Increase in customer engagement