The Pittsburgh Water and Sewer Authority (PWSA) is using machine learning technology to automate its customer service emails.
PWSA’s customers submitting an email to [email protected] are receiving quick and efficient responses to common inquiries with the aid of Y Meadows’ customer service automation solution.
In 2022, PWSA’s Customer Service department received 25,120 emails to [email protected]. To improve responsiveness and reduce customer wait time, the authority implemented a natural language processing software solution to respond to common message categories received by customers.
PWSA says that the software model, first introduced in July 2022, is now handling nearly 50% of incoming inquiries —freeing up customer service personnel to proactively reach out to customers on a variety of topics.
Previously, each email received by Customer Service staff was manually opened and routed to responsive personnel. While many messages require a customized response or additional research, there are an equal amount that can easily be answered with a scripted response.
“Technology improvements are an essential part of PWSA’s transformation to becoming a best-in-class water utility,” said Will Pickering, PWSA’s CEO. “Solutions like the machine learning model and the Customer Advantage portal demonstrate our investment in the customer experience and use of technology to enhance customer service.”
How the software model works
The model reads and identifies the nature of the customer’s request, and, if it falls into pre-selected message categories, it will respond to the email and forward it to the correct party for further review and investigation.
Y Meadows has programmed the system to handle requests such as tenant applications for service, change of address requests, email attachments from customers, and refund applications. It also routes emails directly to Dispatch in response to questions or reports about service interruptions. This assures customers receive a faster response, and, if they need to contact Dispatch directly, they receive an auto-generated email with the correct telephone number.
The software is specifically tailored to customer service email handling and is powered by machine learning and natural language processing. The model can intelligently interpret written messages from PWSA customers and perform useful automated actions to get inquiries resolved. Unlike chatbots, this system does not attempt a scripted conversation but rather works to resolve customer inquiries by pulling relevant data from the inquiry to generate a response.
“Precision and timely communication are expected by our customers, and these are essential functions of PWSA’s Customer Service department,” said Julie Mechling, PWSA’s Customer Service Director. “This advanced software model reduces customer wait time, ensures consistency in responses to routine questions, and improves internal efficiencies. With careful implementation, technology is a tremendous asset to more effectively communicate with customers.”
