In order to introduce adaptive fetching rates in M55 for server-side content suggestions on New Tab Page, we need to collect data about user usage patterns already in M54.
Collect per user averages of number of hours between two successive times when:
- a NTP is opened,
- the user scrolls below the fold,
- the user opens a suggestion.
Later, we will use these averages (computed locally on the client) to classify the client and to tune his fetching rates accordingly.
Comment 1 by fi...@chromium.org
, Sep 8 2016