Zendesk has announced Satisfaction Prediction, a new machine learning and predictive analytics feature for its helpdesk software. The predictive capability is designed to spot problem areas for customer service teams before problems happen, providing businesses with a data-driven early warning system to let them proactively avoid subpar customer interaction scenarios.
Continue Reading Below
Satisfaction Prediction uses a machine learning algorithm to process results of Zendesk's historical satisfacion-oriented online surveys, learning from signals such as the total time to resolve a ticket, latency between customer service representative and user responses, and the specific wording of tickets cross-referenced with customer satisfaction ratings. Zendesk then uses predictive modeling to calculate the probability that a particular customer conversation—be it via phone, email, or a social media interaction—will result in a high (or low) customer satisfaction rating. That data can be acted upon immediately or entered into an organization's customer relationship management (CRM) software for use in later interactions.
Zendesk's predictive feature plays a similar role to self-service business intelligence tools in improving a business's ability to predict based on analyzed data. Based on a projected customer satisfaction score on a scale of 0-100, help desk managers and staff can adjust procedures, alter workflows, or pursue integrations or software improvements to fix problem spots in their customer service. The machine learning algorithm processes both live and historical customer data, and can break down individual satisfaction scores over time to create predictive profiles for specific customers.
Adrian McDermott, Senior Vice President of Product Development at Zendesk, said the continually learning technology is aimed at reducing customer effort when interacting with help desks and lets companies use predictive data to improve their organizations' long-term relationships with their customer bases. Satisfaction Prediction is currently available in beta to Zendesk Enterprise customers and will be generally available in early 2016.