Background.
Millions of people rely on Instacart to fulfill their grocery needs. While our demand is growing so are our support costs. Currently 1 in 11 orders result in an issue where human agents are needed to resolve. This process involves frustrating wait times and can be costly to maintain.
The rise of AI-powered chatbots has made it possible for us to build our own support chatbot to transform the customer experience by handling these requests faster and at scale.
Research
80% of customer issues can be traced back to 4 main topics
Missing items, delayed deliveries, refund questions, or membership inquiries are the top asked questions with the remaining 20% being mostly uncateogorized.
Agents are often wrong
In customer interviews we uncovered that support agents often misquote policies and the lack of consistent answers frustrate customers.
Help needs context
Standard static help center answers aren’t helpful as most users have questions about their specific scenario (ex: my order is missing)
