E-commerce Customer care chatbot

Internal project aligned with a side-hustle venture, an e-commerce website specializing in bricolage components.

The goal was to reduce the workload of an employee solely responsible for handling customer support inquiries. The company utilized a Zendesk chat widget to collect requests from the website, but the employee was tasked with manually responding to even routine questions. I was assigned to develop a solution using a specific proprietary platform.

• Research: in order go gain a comprehensive understanding of the process and identify pain points, I conducted several interviews with my coworker. I analyzed historical customer support requests and the website’s content. In collaboration with my coworker, we compiled a list of the most frequently asked questions, which totaled approximately 20-30,
• Two distinct user personas emerged: hobbyists and professional carpenters. This insight influenced the chatbot’s name and personality,
• The platform I employed provided both a Natural Language Processing (NLP) engine and a prototyping tool for designing conversational flows.
• I crafted the conversational flows to capture questions pertaining to major product categories, so the chatbot could recommend products, provide information on availability, sizes, shipping details, and invoice documentation,
• All assistant responses were crafted with future development in mind, ensuring optimal pronunciation by text-to-speech systems in the event of future vocal interaction implementation,
• During the operator’s absence, I implemented an automated response mechanism that collected the user’s email address and phone number for subsequent follow-up.
• To monitor the chatbot’s performance, I utilized the platform’s console and established a Google Sheets & AppScript integration. This allowed us to capture customer data whenever a busy operator flow was triggered.
• As a daily morning routine, I was also responsible for training the chatbot and read through the users’ transcripts

The chatbot significantly enhanced the employee’s productivity, freeing up time for more critical tasks such as warehouse management and order fulfillment.

During the first quarter of 2022, the chatbot processed over 5,000 messages and independently resolved approximately 90% of inquiries. By the end of 2022, it had handled a total of 100 contacts.