The Clean Zone feature enables users to designate a small area on their map to do more targeted, ad-hoc cleans. But the feature adoption rate was lower than expectation. Utilizing Roomba's camera, we built a feature that suggests potential Clean Zone based on robot-identified objects to minimize user effort and better clean their homes.
ROLE
Lead Designer
TIMELINE
2020
The Clean Zone feature met the user’s need for targeting small areas. But the metrics and qualitative research showed that the adoption rate was lower than expectation and many people haven’t heard about the feature or were too lazy to add the Clean Zone to their map. And at the same time, the competitors were catching up with the robot features including Clean Zones.
The overall experience of adding a Clean Zone and sending the robot to clean those areas needs to be effortless. Users should enjoy targeted cleaning, not be stressed about it.
Gain a higher understanding of the user’s home by suggesting user input if needed. Gathered data will improve object recognition and localization performance for more accurate recommendation.
When robot identifies an object in your home multiple times, it will send a recommendation message for users to review. It includes map and room name to clarify where the zone is for. The size of the recommended zone varies by actual object and will be previewed on the Smart Map with dimensions.
Before saving the recommended zone, user can edit the size and placement to fine tune the Clean Zone. After placing it and naming the zone, the Clean Zone is added to your Smart Map and ready to be cleaned with new mission.
If the user doesn’t want to accept the recommendation, we want to know why to gather data and improve. Once they reject the recommendation, the recommended zone never resurfaces for the same object in the home.
We wanted the cleaning experience to be as natural as possible. We added spatial prepositions to our voice commends. If users want to vacuum around the couch, simply say it using voice assistance as if they were talking to each other.
This project was very data heavy and I worked with data scientists to figure out the sweet spot of recommended Clean Zone sizes. We started with recommendations for three objects, Dining Table, Couch, and the Counter, as they were the most popular areas with manual Clean Zones.
+10%
Initial acceptance rate vs KPI
+9%
Increase in acceptance rate over 3 months