[Editor’s note: This content was written to highlight a particular event or moment in time. Although that moment has passed, we’re republishing it here so you can see what our thinking and experience was like at the time.]
Bonnie Cowan, director of finance at OneFinance, a global Microsoft partner for financial transactions, had a problem: 1 million support tickets flooded her team’s resources each year. Handling simple queries such as cancel an order and find the status of an order left operators little time to provide individualized attention to more challenging customer requests. The OneFinance team needed a new tool to break this bottleneck.
OneFinance already used Microsoft Dynamics – a line of enterprise resource planning and customer relationship management software applications – as its customer service support tool, and Cowan isn’t shy with her praise for Dynamics: “Everything about it has been awesome.” Managing Microsoft accounts payable and buy center processes is no small task, and Dynamics is essential in orchestrating workflow at OneFinance.
To address the overwhelming volume of routine support tickets, Jovalene Teo, senior global program manager at OneFinance, collaborated with the Dynamics team to find a solution. Teo and the Dynamics team decided to leverage a Microsoft machine-learning technology called Language Understanding, or LUIS. The goal was to automate the management of 30 percent of support tickets—those that were deemed not to require human staff intervention. If the initiative succeeded, 300,000 tickets per year could be handled and closed by chat bots.
[Learn more about how Finance is using AI and chatbots to simplify finance tools at Microsoft and how Microsoft is creating efficiencies in finance with Dynamics 365 and machine learning.]
LUIS is a machine learning-based service that enables users to build natural language into apps, IoT devices, and bots – such as OneFinance’s chat bots. “LUIS builds apps around customers’ natural way of getting answers,” said Azharuddin Mohammed, a Microsoft senior software engineer. “Customers don’t need to learn something new. They talk to a ‘person’—a chatbot—to get their answers.” In other words, when a user tells an app to book a flight, order a pizza, or remember to call Dad, the LUIS technology translates these commands to the app so that the action can be executed. “LUIS is the bridge that converts natural language into something a machine can understand,” Mohammed explained.
At OneFinance, the LUIS service integrated seamlessly with Dynamics tools already in place, enabling staff to divert and resolve routine support tickets while funneling more complicated tickets to human operators. The suite of services even allowed the team to work in a variety of languages.
As a global organization, the OneFinance team responds to support tickets generated in dozens of languages, so LUIS had to work with the integrated Bing Translator. The Dynamics team anticipated and accommodated this need. Mohammed pointed out that now when tickets come into OneFinance in Japanese, Bing Translator automatically rewrites the ticket in English. Most translated tickets are resolved quickly by the chat bot or operators, saving the company from paying premiums for translation services. In fact, LUIS supports 37 languages.
Recently, Microsoft’s voice-controlled virtual assistant Cortana and LUIS joined forces at OneFinance. Customers seeking support can speak their requests or call in, just as they would with a live operator. “Now users can go ahead and talk to our chatbot and get those same answers, instead of waiting for up to two days to talk to an agent,” said Mohammed.
LUIS has quickly streamlined OneFinance’s customer service. About 20 percent of users now direct their simple queries to chatbots and operators, and 97 percent of those tickets can be closed at “first touch.” “People can come in today and interact with the bot,” Cowan said. “If their question is answered, their query is closed, and we don’t have to go any further. For more complicated queries, they can choose to chat with an operator.” At this time, bots can answer 200 standard user questions, and to date, users have interacted with the chatbots 28,000 times.
LUIS’ positive impact on OneFinance operations is clear. “It is producing cost savings because we can drive efficiencies, but it is also about improving the customer experience,” said Cowan. “Because we have so many simple queries and we are trying to get to all queries, we weren’t able to provide the high touch we are looking for.” When the resolution of simple queries is automated, Cowan points out, complex queries can receive a “white glove service without an increase in costs.”
Mohammed agrees, describing the LUIS integration solution as a win-win situation for customers and businesses. “The benefits are huge. Customers are getting information faster, and they’re getting information accurately,” he said. “When machines take care of things an agent used to cover, you save money.”
Looking ahead, OneFinance plans to push LUIS to its full potential. Next in line: utilize LUIS’ active learning capabilities. “We are collecting data from chat logs with our bots and using that data to see what has been useful to our users and what hasn’t,” Teo said. “We are working on how to translate that feedback into improvement through LUIS.”
Using LUIS’ active learning capabilities, businesses can constantly update and improve the service. For example, when a user asks LUIS a question that it cannot process, the business can review the chat logs to adjust right away. “The best part is that this is easy to do. You can see in real time what users are doing and saying,” said Mohammed, “and then change the model to incorporate that input.”
Based on the quick success OneFinance has experienced by incorporating LUIS, Cowan now believes her team’s goal is in sight. With technical support from Dynamics, OneFinance should be able to close 300,000 simple queries per year—approximately 30 percent of all support tickets—using chatbots by next year. “The technology is new, and our volumes are low but our quality is high,” Cowan said. “We are just now starting the process of spreading the word across Microsoft. We are right on track.”
