No. All raw data is only processed on the device and is never sent out.
Yes. Such personalisation is based on users’ behaviour and environmental context.
We store our data on Google’s servers located in London, UK.
Our mobile SDK collects mobile sensor data on the device, and processes this data into live context predictions and ID-less audiences (still, on the device).
Audience data is currently refreshed every 24 hours.
We use a combination of proprietary datasets (collected in collaboration with device manufacturers and academic institutions), public datasets, simulated mobile sensor data, data collected from test devices, and live mobile sensor data from smartphones.
Whilst there is no single accuracy number, as we predict over 200 live contexts across various categories, all of our predictions are probabilistic with high degrees of accuracy.