United StatesTechnology and telecom2018 – ongoingUber
Customer Obsession Ticket Assistant for support agents
Uber reported over 10% shorter English-language ticket resolution times with its first COTA assistant
Problem
Uber's support agents had thousands of issue categories and reply templates to choose from. Its expanding services made ticket classification increasingly complicated.
Approach
Uber integrated COTA into its customer support platform to recommend issue categories and reply templates. The first version used machine learning and natural language processing. Uber subsequently deployed deep-learning models through Michelangelo and added automatic model retraining.
Results
Takeaway
The quantified resolution-time improvement belonged to COTA v1. Uber reported better online results for v2 without giving a comparable handling-time percentage.
Categories
- Size
- Large enterprise
- Industry
- Technology and telecom
- Business function
- Customer service
- Innovation type
- Process innovation
- Adoption
- Built in-house
- Evidence
- Company's own claim
Sources
Source summaries
Uber described COTA's recommendations to support agents and its move to deep learning. It attributed over 10% shorter English-language resolution times to v1 and reported further online improvements for v2.
Full details are in the original sources.