According to DigitasLBi, 70% of customers will not use a bot again after a bad experience, and 60% of those customers also find bots to be rigid and artificial. This is not a coincidence. Poor conversation design leads to stilted, robotic bot dialogs that impede user trust and block adoption. This talk transforms foundational linguistic research into actionable design best practices that help customer support team leaders craft a chatbot dialog that follows natural patterns of conversational behavior. The result is a helpful interaction that builds trust with users and increases CSAT. Examples from chatbots built by Salesforce customers using these best practices in the Salesforce Einstein Bot Builder will be shown.
Best Practices for Designing Natural, Helpful Bot Dialogs for Customer Support
Greg is Lead User Researcher at Salesforce on Einstein for Service Cloud, a product offering that leverages AI and chatbots to help support centers manage and grow their operations at scale. Trained formally in linguistics, Greg focuses his chatbot research on conversational UX cross-cultural conversation design. His work includes practical best practices that empower businesses to design natural, helpful bot dialogs that build user trust and meet human expectations for conversational behavior. Greg’s most recent paper, “Conversational Style: Beyond the nuts and bolts of conversation,” was published in the conversational UX volume of the Springer Human Computer Interaction Series.