By Tim Deluca-Smith, vice president of Marketing, WDS, A Xerox Company

Do you read product manuals when you need customer support? More often than not, the answer to this question is no.

But why? Certainly for consumer electronics, manuals can be laden with technical jargon and complex language that demotes this resource below alternative forms of support. When things go wrong, 65 percent of customers will pick up the phone or head back to the store to consult with a customer service or sales representative.*

These reps are able to engage customers in two-way conversations where they can ask clarifying questions, and query the challenging issues that lie beyond the remit of product manuals. They also have the most up-to-date information on ever-changing policies, promotions, and product and service performance through their access to wider pools of support knowledge.

Nobody Reads the Manual

That’s why, when our product team designed WDS Virtual Agent, they knew they too had to leave the manual in the box. Instead, they focused their attention on how customer problems get resolved in the contact center. Only then could they build a virtual agent that could replicate a successful customer care interaction.

Creating a Virtual Agent That Can Learn Like a Human

Mirroring the best customer care interactions with both customer convenience and operational cost efficiency in mind is no mean feat. We know that in order to cater to the next generation of customers, our customer care systems need to be more intelligent. But just how intelligent do they have to be?

Some virtual agents take the keyword-search route, but, using Xerox research, we took a smarter approach.

A form of artificial intelligence, machine learning enables computers to learn without being explicitly programmed. It focuses on the development of computer programs that can teach themselves to grow and change with the introduction of new data.

In that way, machine learning is similar to data mining, as both systems search through data to look for patterns. However, while data mining extracts data to support human comprehension; machine learning uses that data to improve the program’s own understanding. With this, machine learning programs identify data patterns and adjust their actions accordingly.

Thanks to machine learning advancements derived from the Xerox Research Center Europe and PARC, A Xerox Company, WDS Virtual Agent is able to learn in the same way that humans do, through the experience they gain on the job.

But in order to learn effectively, develop the necessary experience and ultimately provide the right answers to customers, the system must first be able to understand what the customer is saying.

WDS Virtual Agent

Learn more about WDS Virtual Agent, which is designed from state-of-the-art machine learning and natural language processing innovations.

Understanding Customers

In addition to machine learning technology, virtual agents that deal with customer care queries can develop more accurate associations based on how customers describe their issues. This is hugely beneficial because customers do not speak in the language of “issue drivers” or technical jargon from product manuals. Using natural language processing, the virtual agent can replicate the innate understanding humans have during their interactions with each other; including the ability to develop and build on existing knowledge to make bigger and better associations.

Because the virtual agent’s learning is continuous, its accuracy is always improving and new problems are identified and understood quickly, as a matter of course. What’s more, is that this intelligence can be used across the broader care mix, improve the proficiency and the adoption of digital channels, while reducing the fallback traffic in the contact center. That’s why this approach is smarter: It can deliver value far beyond that of virtual assistants posing as a gateway to static knowledge bases.

*Source: WDS, A Xerox Company, Customer Care Transformation

Subscribe to Simplify Work and receive email updates when we publish a new article.

This article was originally published on the WDS blog.