Geo-fencing is a location-based service that allows marketers to send messages to smartphone users who enter a predefined geographical area. This helps them to target the foot traffic in the vicinity of a point-of-interest. Current geo-fencing technology creates static geo-fences and thus is oblivious to users’ mobile app behavior. This results in suboptimal targeting of potential customers with respect to when, where and what messages are triggered.
This project aims to assist marketers in creating smarter geo-fences. Our algorithm automatically segments users based on their geo-distributions of mobile app activity, identifies points-of-interest and then suggests creation of geo-fences customized to each user segment. The geo-fences can be specific to a product or service and is parameterized by a threshold to balance between effectiveness and unwanted targeting. Marketing messages can be personalized to each user segment and can be auto-triggered on entry/exit from a geo-fence. Our data-driven geo-fencing algorithm is less prone to human biases and can sometimes capture surprising aggregate mobile app behavior in the form of geo-fences around unexpected points-of-interest.
Here is a short demonstration video for the project.