The Firefox Health Report creates a daily data report to Mozilla of consisting carefully-selected data judged critical to gaining a greater understanding of our users.
Our goal is to provide real user value by allowing a user to compare their data against the aggregate data collected by the service, to highlight performance issues or common problems for a given configuration.
Technical Description of Data Submitted
For an exhaustive list of data submitted by the report, please see the current list
Summary of Key Data Points
The data submitted by the Firefox Health Report consists of a number of key components:
FHR reports a comprehensive set of data about the current build and active platform, specifically:
- Full build information (version, target OS, build ID, distribution and update channels)
- Current environment (current locale, OS information, CPU (cores and architecture) and RAM (megabytes)
- Profile age and current session length.
This data is probably only directly useful for class-of-hardware/memory count messaging. However, it can be used to filter aggregate data to provide more relevant comparisons elsewhere
This section provides data on previously-used versions, and is probably not high-value to a user-facing report, aside from perhaps detecting "this user lags updates" and recommending auto-updates.
This section provides information on all extensions and plugins, excluding those where the developer has requested that data not be provided. The key data that is most useful includes:
- Currently installed add-ons + enabled state (enabled/user-disabled/blocked/incompatible)
- Install/update dates, and third-party installed flag
The most obvious case is to warn users about problematic add-ons, and possibly highlighting recently installed add-ons.
FHR collects a number of per-day stats to create a comprehensive profile of usage over time. As we only retain the current report, each report contains up to 180 days' worth of this data.
- Searches by UI component. This data includes engine names where Mozilla has a search partnership agreement, otherwise it is part of "Other"
- Session lengths, active times, and shutdown stats (clean shutdown or otherwise)
- Add-on counts by category (Theme, lightweight theme (Personas), plugins, extensions)
- Startup performance (multiple measurement points)
- Places info (total bookmarks + total history items)
As we can process data over time, we can detect aberrations in performance by date, and cross-check against both app version and add-on install/update dates (with some fuzz) to suggest possible causes. (i.e. startup performance degrades by 40% when Bob's Awesome Add-on gets installed? suggest disabling and checking perf!)