We built a chatbot for our direct-to-consumer eCommerce brand, YourLibaas, to improve customer service and reduce response times. The goal was to assist users during checkout and increase conversion rates. However, we initially failed miserably due to the scope being too narrow and not handling complex queries well. We learned that chatbots are best suited for repetitive queries and created a set of predefined queries and responses. After beta testing, we found that the initial results were disappointing, but we continued to improve the bot by monitoring failure causes and adding conversation flows on the go. We also made sure every time the bot failed, an available human representative was assigned. To increase chatbot usage, we created a tutorial video and introduced basic natural language processing abilities. The final result was that customers started rating bot interactions as good as human agents, with successful resolution for over 60% of queries. The journey is ongoing, but we've learned valuable lessons about the importance of ongoing improvement and the role of chatbots in supporting customer support teams.