As a user uses an app over a period of time, we learn about their habits based on what they do in the physical world (e.g. running 3x a week, driving every day) using any of our over 200 live contexts. These repeat behaviours give us an indication of which audiences the user might be categorised into (e.g. joggers, vehicle users). As all of this happens on the device itself, there is no need to send this data to the cloud, eliminating the need for any identifiers.