The world’s largest high-grade nickel and palladium producer was suffering from poor master data management (MDM). The analysis had identified the weak performance of their existing MDM system chatbot among root causes, as well as an opportunity for a quick win.
- Analyzed two years of user questions from the chatbot history, grouping those according to answer types to cover every one of the questions asked
- Generated the 68 most popular answers based on the chat history, user manuals, and process regulations, covering all user scenarios related to catalogue and records management
- Formulated over 50 alternative questions per answer-group to improve chatbot accuracy by providing the ability to give the correct answer regardless of the wording used
- Chatbot accuracy increased to 95%
- Chatbot now answers any question related to record creation, editing, duplicates search, filling rules, and other use cases that ultimately improve master data quality in the system
- Immediate improvement in user satisfaction