Founded in 1986, the client is a regional last mile delivery company that services the Eastern and Midwest United States. It is based in Vienna and Virginia and has sortation centers in New Jersey, Ohio, North Carolina, and Florida.
For Lasership customer care center employees, it became difficult to handle the number of calls from customers which had increased from 75,000 to 150,000 per month and the company did not have the resources to scale. It cost them around 75,000 missed tickets (closed out without follow-up). The client was looking for an authentic, credible, immediate and effective response to customer’s requests in the form of an efficient chatbot application.
1. As a chatbot never learns on its own, so the biggest challenge was to create a conversational architecture based completely on the client’s business needs. Chatbots are about a continuous conversation between the bot & the user where the conversation between them can go in any direction. This needs to be mapped in terms of dialog flows & storyboards. The whole conversation needs to be tracked because a message has no value in isolation - they are part of a conversation and their context needs to be maintained & determined. After an extensive RnD we came up with a mind map which represented all the conversation flows that the Chatbot should follow.
2. Once deployed, another challenge was to understand how well the chatbot is working. For this, we had to analyze the conversational flows to understand engagement, deflection, & misunderstandings and then improved it by updating its intents to deliver a high-quality or more personalized experience.
The SPS team makes sure to fashion quality assured solutions and for that the team has devised a procedure that yields a high-grade end product. The procedure passed through several iterations to finally produce a chat bot that met all the desired objectives. Below iterated is the lifecycle of a chat bot implementation procedure, entailing names of the steps involved in development of a chat bot.
After completing all the steps of the lifecycle, the chat bot is continuously refined to serve the customers better. This means once you build a chat bot, you just cannot forget it if you are expecting it to bring in more sales/customers or give your customers a better experience. The learnings from the “analyze” phase are cycled back into the bot development process to build an ever-improving bot.
The solution was a friendly bot - Codee - that would answer any user query within seconds and helped the client generate more sales and build a deeper rapport with customers. Codee is a keyword recognition-based bot which is an artificial intelligence (AI) software that can be deployed on any website helping the organizations and businesses to simulate a conversation (or a chat) with a user in natural language through messaging applications to optimize customer service operation.
The chatbot was successfully deployed on the Lasership website in 2017. Project enhancements are still going on. SPS development team is working on adding sentiment analysis to detect emotional and language tones in written text.