Harvard Business Review
Harvard Business Review
A conversational user interface: Building a Best Practices Slack bot for Harvard Business Review
Conversational User Interface Design · User Testing · Kano Model · UX Writing
Read my case study below
See press for HBR Bot for Slack: Digiday, Fipp, Media Makers Meet, VentureBeat
200+ Best Practice Articles
Vermonster partnered with Harvard Business Review to build a Slack bot to help consumers discover daily Best Practices content from the contributors at HBR. Our goal was to create a familiar experience within a platform—Slack—that consumers use every day. Daily or weekly, users receive practical advice on topics such as how to ask for a raise, how to give constructive feedback, or how to tell a great story.
By subscribing users to Best Practices, we created expectation-aligned, discoverable, and pleasant direct-message interactions where they can find and peruse more articles within Slack. We adopted a Node.js-based architecture, well-suited for bots due in part to its optimal concurrency model. The bot persists usage data and patterns to enable analysis of user interactions. The system automatically updates bot content whenever new articles are published to the HBR website.
User Research and Prototyping
Prototyping in the early stages is crucial to save development time and resources later. To test our ideas on what the Slack bot could be or should do, our design teams collaborated and scheduled remote usability tests with users around the globe. To give users an experience as close to the final product as possible, we ran tests on Slack in which we acted as the bot to simulate different user scenarios in a conversational interface. We ensured that the bot was smart enough to have a conversation while maintaining its bot-like properties so that users did not ask for more than the bot could provide. This strong collaboration between the Vermonster and HBR usability teams resulted in clear iterations. Our Kano Model questionnaire, part of our user research methodology, helped us rank the potential product features.