ZEUX was tasked with building a best-in-class, AI powered, conversational sales bot for online shopping. The central UX brief was that the sales bot needed to be positioned as an intelligent sales expert that helped users substitute hours of research with a few minutes of interaction.
Based on our initial stakeholder visioning and reviewing the best practices in AI assistant design, we identified some key UX challenges:
Our approach for this project allowed us to explore the best practices for building a sales bot using an in-person, physical store sales experiences as a reference. One key insight was to position the sales bot as an expert, rather than an assistant, to generate trust in the bot and to ensure that the bot complements the digital interface rather than compete with it.
Three key experience pillars…
Setting Expectations: Well-designed bots are upfront about the fact that they are bots. Giving this context ensures that the users interact with the AI knowing that there might be situations where clarifications are needed.
Building Rapport: The best interactions that people have with each other is when they are on the same wavelength, especially in a sales environment, where trust and ease need to be built in for effective transactions. For a bot, the rapport needs to be built through asking the users questions, making quick inferences on the user’s personality and mirroring their cues and tone of voice. We further allowed users to choose from one of three bot personalities, making it easier to build the rapport needed for a comfortable transaction.
Speaking like the User: Speaking the user’s language proved to be extremely useful for a sales bot. That includes speaking conversational Hinglish rather than formal English or Hindi or sending memes or gifs in response to user inputs.
Our final design implementation included a robust design system and guidelines for all the core conversation flows.
Buzzo.ai partnered with an online shopping platform for their AI shopping assistant. The business impact within 3 months of launch was:
2.4x in cart addition rate.
6 was the average number of user queries answered per conversation.
94% AI response automation.