RPA CASE STUDY

AI solution trained to answer monthly deluge of customer questions

Industry : Cross-Industry, Retail and Consumer Services

Function : Contact Center

Results

85% response rate emails

85% response rate emails

Faster responses to customer questions

Faster responses to customer questions

Monthly savings of $136,000

Monthly savings of $136,000

Company Profile

A global leader in food and beverage services, this company sells, markets, and distributes food products to restaurants, healthcare, educational facilities, lodging establishments, and other customers who prepare meals outside the home. Product offerings also include equipment and supplies for the food service and hospitality industries. The company operates more than 326 distribution facilities worldwide, serves more than 600,000 customer locations, and has more than 57,000 associates.

Scope Highlights

  • Krista Cognitive Issue Resolution
  • Krista Automation Platform
  • 20,000+ inbound emails
  • Continuous machine learning
  • Service operations

Are you ready to add WonderBotz to your organization?

Get a Quote

Challenge

Each month, wholesale food product distributor received 20,000 emails sent to 75 different email addresses. Emails contained questions requiring handling from different departments. Employees had to read, route, get the answer and, ultimately, respond to the email manually. This was a time-consuming, error-prone process that was expensive to run with each email costing about $8 to fully handle on average. Management sought a faster, more cost-effective approach that would also improve service to customers.

Solution

The solution uses natural language understanding (NLU) to read each email and extract information to fulfill each request. It logs each request and response into ServiceNow and a custom CRM system for tracking and reporting. The model learned how to handle 85% of the questions and routed the remaining 15% to the appropriate service representative. As service representatives respond to the remaining messages, their actions are captured and used to train the solution to handle similar questions in the future.