Zendesk Adds Answer Bot to Zendesk Guide

Zendesk has enhanced its Zendesk Guide, a knowledge base and self-service product, with the general availability of Answer Bot. This new feature, available as a monthly subscription, uses Zendesk Guide’s knowledge base to suggest not only relevant articles, but the content within them to best respond to customer queries. Designed to provide answers to simple support inquiries, such as those concerning product specs, order statuses, and subscription changes, it can resolve a ticket before it reaches an agent.

Answer Bot uses deep learning techniques to gain information on over 12 million successful agent responses that use knowledge base content, and continues to learn from the feedback it receives. The interactions Answer Bot has with customers are presented to the support team, so they can see where it works or falls short, and provide pointers to improve future performances.

Because Zendesk has already trained Answer Bot in a vast amount of customer service interactions, it has reduced the complexity required to get it up and running. As such, it can be used without needing minimum data volumes or lengthy training, while continuing to grow as usage grows.

Answer Bot was designed to be used as a “bot-like” experience, and typically resolves tickets in around 12 minutes. Zendesk clients who have been using Answer Bot have already seen thousands of tickets resolved, with a successful resolution rate of around 6% percent, with some customers much higher. More importantly, it has managed to reduce the average resolution time for some clients by half, allowing agents to more efficiently resolve the tickets that automated suggestions alone could not.

Only a fraction of Zendesk's 107,000 customers are using Answer Bot so far, but their experiences have been providing important data. One company that provides personal products on a subscription basis, for instance, has been using Answer Bot to help customers with questions regarding subscriptions. So far, Answer Bot has solved around 13,000 tickets, or about 10 percent of the company’s ticket volume.

A financial software company focused on helping people improve their spending habits was another Early Access Program customer. Its support team was often overworked, with five people serving hundreds of thousands of customers, but found that most questions from customers would relate to one of three base categories. They then used Answer Bot to efficiently respond to their most common questions, freeing up the agents to better assist with more pressing customer concerns.

Currently, Answer Bot supports email, and Zendesk is working on adding web form, web widget, chat, and SDK channels. There are also plans to add a feature that stops it from suggesting articles customers have already read, to avoid frustrating them by recommending articles that were of little help the first time.

According to Blair Pleasant, co-founder of UCStrategies, “More and more organizations are turning to bots and machine learning systems to help augment live staff, enabling staff to focus on more complex questions and issues. I expect to see a huge increase in the demand for these types of solutions, as they help organizations provide more consistent and fast service, while saving time and money.”

For more information, visit www.zendesk.com.