Luna the course search assistant chatbot
Analysis showed that users of the Whatuni product were not using filters in their search. This was despite our earlier research, which indicated that personalizing search results was a key requirement to help users alleviate their pain points.
Alongside a team of Data Analysts, I determined that usage of filters on Whatuni were decreasing. I consulted various Product Management blogs to help me discover methods used by other product professional to encourage greater usage of filters and personalized searches. This led me to suggest the company should build an AI chatbot to nurture users through the search process.
As a Senior Contractor I worked as the Product Lead for this project and had direct responsibility for the project delivery, owning the project from ideation through to development and final release.
Within 6 months of launch, the chatbot had increased the avg. user session duration by over 30%. The number of users who were using the chatbot and completing a search with filters applied was over 80% indicating that the chatbot was effective at nurturing users through the search journey and helped them to create personalized search results.
What was learnt
- This project really highlighted to me the importance of getting the copy and terminology right. Our initial draft content that was written was heavily criticized by the users we spoke to during the usability testing sessions.
- This project helped me to improve my communication and collaboration with multiple engineering teams simultaneously.
“Luna” is an AI chatbot that helps users to navigate through the course and university search journey on the Whatuni website and mobile app. Luna collects and uses a wide set of data to make personalized recommendations to each user.
Project goal & Success measures
- To help users to complete more personalized searches with the use of filters and personalized recommendations from an AI chatbot.
- To increase number of searches completed by users.
- To Increase the number of users using filters for their individual searches.