Program Management/Programs/Telemetry/Improve Opt-in: Difference between revisions
(→Goal) |
|||
Line 1: | Line 1: | ||
=Goal= | =Goal= | ||
It is thought that the current Telemetry opt-in rates do not provide enough data to be useful. In general, Telemetry is only useful if users are providing feedback. The issue with Telemetry data lies in the sample population. Because Telemetry is opt-in, the sample population is not random and therefore needs to be very large (80-90%) to answer a broad range of questions. It may still be possible to answer specific questions with a smaller sample size (metrics can engage on an ongoing basis to assist with determination of whether a question can be answered). Alternatively, the existing sample population may contain enough variety to be representative. | |||
==Investigation== | |||
The metrics team (Chris Jung) will investigate to see if the current Telemetry data is representative based on the information that we have available from daily add-on pings. See {{bug|698848}}. | |||
Lawrence will investigate changing Telemetry to an opt-out option. | |||
===Release channel=== | ===Release channel=== | ||
Current: ~2%<br/> | Current: ~2%<br/> | ||
Goal: | Goal: | ||
===Nightly (dev) channel=== | ===Nightly (dev) channel=== | ||
Current: ~2% <br/> | Current: ~2% <br/> | ||
Goal: | Goal: | ||
=Tactics= | =Tactics= |
Revision as of 19:50, 1 November 2011
Goal
It is thought that the current Telemetry opt-in rates do not provide enough data to be useful. In general, Telemetry is only useful if users are providing feedback. The issue with Telemetry data lies in the sample population. Because Telemetry is opt-in, the sample population is not random and therefore needs to be very large (80-90%) to answer a broad range of questions. It may still be possible to answer specific questions with a smaller sample size (metrics can engage on an ongoing basis to assist with determination of whether a question can be answered). Alternatively, the existing sample population may contain enough variety to be representative.
Investigation
The metrics team (Chris Jung) will investigate to see if the current Telemetry data is representative based on the information that we have available from daily add-on pings. See bug 698848.
Lawrence will investigate changing Telemetry to an opt-out option.
Release channel
Current: ~2%
Goal:
Nightly (dev) channel
Current: ~2%
Goal:
Tactics
The following are some tactics that may be used to increase the opt-in rate for Telemetry.
Change to opt-out policy
metrics pursuing opt-out policy - follow up with them to see how they're progressing Daniel for follow-up next week will require security review
Change opt-in ui
Issues with the current opt-in UI:
- UI may disappear quickly
- UI may be hidden in another tab
These issues are related to the prompt mechanism (used for know your rights message) not being designed to be used heavily.
Need to discuss UX goal.
Bugs:
- Replace opt-in UI bug 692861
- Current UI gets dismissed too easily bug 691951
Want to pursue a more aggressive opt-in UI for nightly. Suggestion in bug 683340 (incomplete and needs UX design). Another suggestion is to put a checkbox on the download page (like Chrome) but this has two notable drawbacks: Firefox is multiseat and this only addresses new downloads not updates.
Follow up with:
- UX
- Brian Dils is engaged and looking at the current notification mechanism. Lawrence and Brian will sync up at the mobile work week (Oct. 17-21)
- frontend
- Spoke with Dietrich and Chris Lee. Chris is prioritizing the feature from a product management perspective to allow for fx resources to work on the issues.
Increase publicity
Using the Mozilla blog (others?) promote Telemetry and request user help with improving the performance of Firefox.