Small design decisions determine whether users trust and return to a site-specific AI chatbot. Nielsen Norman Group studied real users across multiple sites and identified 10 concrete guidelines that separate useful chatbots from frustrating ones.
The research isolates three specific pressure points: how the chatbot introduces itself, whether it follows users across pages, and how it surfaces product recommendations. Each of these, handled poorly, drove participants away. Handled well, they drove repeat use.
The full article is worth reading for the specific mechanics behind each guideline, not just the list itself. If you are making decisions about chatbot UX right now, the detail on prompt suggestions and contextual awareness alone justifies the click.
[READ ORIGINAL →]