Comment by luke14free

I work on models like those of Sensor Tower and App Annie use to estimate downloads. They are extremely precise, the cross-validated error is << 10% for apps that are this small, and in the US (it increases with much bigger apps in countries where little data is available).