Streamlining Tech Support with Natural Language Processing

Any company that makes and sells consumer or business products faces technical support challenges. With even the best-made products, some customers will inevitably have questions or encounter issues that cause them to turn to a help desk for support.

For the tech support organization, the goal is always to streamline and accelerate the support experience, so customers get their questions answered and their issues resolved quickly and painlessly. To achieve this goal, the support organization needs tools that enable its agents to find  the right information in an efficient manner.

At Dell Technologies, these tools now include a homegrown machine-learning system that leverages natural language processing capabilities.

What is NLP?

In simple terms, natural language processing, or NLP, is a form of artificial intelligence (AI) that allows a computer application to understand human language, either spoken or written. The concept of NLP encompasses coding, understanding, interpreting and manipulating human language.

NLP applications use computers to translate languages, convert voice to text and text to voice, and create human-like conversational agents to help an organization’s employees, customers and partners deal with issues, questions and concerns. This can be an extremely valuable tool on a tech support desk, where agents require fast routes to insights and answers.

At Dell, we’re proving this point every day with our machine learning tool, which we developed in a project we call “Digital Resolution.”

An intelligent support tool

The genesis for our intelligent support tool was a desire to streamline the customer support experience. Prior to the development of the tool, our agents worked from what was basically a big page of web links leading to decision trees and troubleshooting guides. Some of these links sounded almost synonymous, so it could be hard to know which link might lead to the right answer.

And then the agent would go through a process to diagnose the problem and come up with a potential solution. This was another place where we saw the opportunity to use machine learning to build more efficiency into our processes, so our agents wouldn’t have to go through so many steps to get to the right solution.

Our tool streamlines this entire process. It gives our agents predictions for the best troubleshooting steps to suggest to customers who call in to report issues with products. This easy-to-use tool, which incorporates multiple machine learning models, helps agents diagnose and solve problems quickly and accurately without having to navigate through a maze of web links, troubleshooting guide and decision trees.  Now, 25 percent of the time, we can skip diagnosis and go straight to a solution, greatly improving the customer experience.

Today, we have more than 3,000 agents using the tool, servicing more than 10,000 customers per day. And we’re seeing some great results. The tool has helped us achieve a 10 percent reduction in call times, which means the customer gets off the phone faster, and each of our agents can field more calls. And it is helping us improve service accuracy, which reduces the number of customers who have to call back — because they got the right answers initially.

What’s next?

An initiative like our Digital Resolution project is never a one-and-done undertaking. Through this project, we continue to enhance our algorithms and scale up our system to accommodate more users.

In fact, we have included even more AI capabilities in the last month by translating the entire tool for non-English users. Through a major project, we were able to include functionality to support Spanish, Portuguese, Mandarin, and Japanese. This utilizes Microsoft AI translation services in addition to custom Dell-specific translation support. We are in the process of onboarding users from these non-English teams and expect to triple our potential user base!

One of the takeaways here is that you don’t have to try to boil the ocean — or to do it all at once — when you develop a tool like this. We started small with a limited support target and a proof of concept, we got executive buy-in, and then we grew from there. Machine learning is an iterative process – first make it exist, then keep working to make it better.

The idea is to start small — but always keep your eyes on the big vision.

To learn more

• For a look at the how-to side of the story, see my presentation Getting Models from Development to Production.
• To learn about howNatural Language Processing Could Be Key to Your Company’s Digital Transformation read my colleague Luke Wilson’s blog.

For a broader look at how we at Dell are incorporating data science into our operations, see the podcast Data Science Driving Next Generation Customer Service at Dell