Mosaic Wellness Chatbot: Fixing a Conversion Problem from Scratch

On her first day at Mosaic Wellness, Janhvi was given a choice: work on the chatbot or work on marketing. She chose the chatbot because she could see a real problem worth solving. A large number of users were engaging with the WhatsApp chatbot, but the conversion rate to actual orders was low. The gap between engagement and purchase was the brief she set for herself.

She started by analyzing the chatbot flow end to end. She looked at where conversations ended, whether they concluded with a satisfied customer or dropped off abruptly, what reviews users were leaving, and at what point the AI handed off to a human representative. She mapped the full flow in a spreadsheet, identifying the friction points that were causing users to disengage before completing a purchase.

Redesigning the Flow

Working with the product team, Janhvi redesigned the chatbot flow to reduce the friction points she had identified. She also examined the push notification behavior on the company website. The existing notification appeared two minutes after a user opened the site, but the data showed that users typically made purchase decisions around the 15th or 16th minute of their session. The notification was firing too early, before users had reached the point in their browsing where they were ready to act.

Janhvi recommended delaying the push notification to align with the actual purchase window. The product team implemented both the chatbot flow changes and the notification timing adjustment.

The result was a 25% increase in orders through the chatbot. Janhvi owned the analysis, the redesign, and the coordination with the product team throughout; the implementation on the backend was handled by engineering.