1251314 - Probes to measure activity and usage of bookmark folders

Status: RESOLVED DUPLICATE of bug 1329138

(Firefox for Android Graveyard :: Metrics, defect)

Description

[Barbara Bermes [:barbara] - NI please!]

Let's get some metrics to measure and keep track of how bookmark management improvements will effect users.

Can we please confirm if we need to add some of those probes.

- measure what actions are being applied to folders, i.e. long-press on folder, delete, move, rename etc.
- measure (avg.) folder count (so we can see if people just use the default folders or create new ones) 
- measure ratio between # of bookmarks vs. bookmark folders to understand organization method of our users
- measure if a URL is loaded from within a custom folder vs. default folder vs. non-folder (root), so we can compare how useful folders will be

Comment 2

[Mark Finkle (:mfinkle) (use needinfo?)]

(In reply to Barbara Bermes [:barbara] from comment #0)
> Let's get some metrics to measure and keep track of how bookmark management
> improvements will effect users.
> 
> Can we please confirm if we need to add some of those probes.
> 
> - measure what actions are being applied to folders, i.e. long-press on
> folder, delete, move, rename etc.

This is in line with the existing long-press context menu probes we already have for listitems. We should be able to implement the exact same probes. Note that we use (ACTION, CONTEXTMENU, extra) for those probes. The "extra" is the ID of the menuitems. In general, we use "home_*" as the prefix for those items. If you want to differentiate folder actions from listitem actions when the ID could be the same, be sure to use a different ID.

> - measure (avg.) folder count (so we can see if people just use the default
> folders or create new ones) 
> - measure ratio between # of bookmarks vs. bookmark folders to understand
> organization method of our users

Both of these a histogram-type measures. If you feel they are useful, they can be added. Personally, I would like to move away from these "counts" and focus more on the behavior:
  * How many folders were created a week?
  * How many folders were navigated a week?

> - measure if a URL is loaded from within a custom folder vs. default folder
> vs. non-folder (root), so we can compare how useful folders will be

This is a very tricky probe. Somewhat unique in it's implementation. We already track loading of URLs and do not have the space to add "folders" into the payload very easily.

This could be approached as a funnel-problem using the event data: Look at the URLs loaded after a folder is opened.

I assume some of these probes are for trying to determine if bookmark folders are used. We could just focus on:
1. Do people make folders?
2. Do people navigate through folders?
3. Do people assign bookmarks to folders?
4. Do people use bookmarks more in general (save and load) when folders are available? (A/B test)