Bugzilla

I extracted some telemetry numbers from the 1st of December and version >= 69.0: 0.04% of result selections were through tab, 0.54% of users selected with tab at least once. In Nightly I see a 0.08% of tab interactions, and 1.75% of users doing it at least once. Take the numbers with a grain of salt, I had to use BigQuery because the telemetry dashboard is broken, but they seem consistent with what I saw on the dashboard earlier.

Verdi and a11y suggest we should disable tab results navigation when the interaction starts from the keyboard shortcut, so we preserve the behavior for mouse users.
But Dao has valid points about avoiding code complication just to serve what seems to be a barely used case, and that we don't know if tab users are more likely to start interaction with mouse or keyboard.

Taking the partial disabling path, we could monitor telemetry; if it drops drastically it means most tab users started interactions from the keyboard, then we did broke most tab users. At that point we'd need either to reintroduce it somehow, or completely remove it because at that point the benefit is even smaller.
To be on the safe side I still suggest going the pref way, clearly this doesn't clear our threshold for users benefit, but it's an historical behavior and a non-trivial amount of users seem to use it at least once.

I extracted some telemetry numbers from the 1st of December and version >= 69.0: 0.04% of result selections were through tab, 0.54% of users selected with tab at least once. In Nightly I see a 0.08% of tab interactions, and 1.75% of users doing it at least once. Take the numbers with a grain of salt, I had to use BigQuery because the telemetry dashboard is broken, but they seem consistent with what I saw on the dashboard earlier.

Verdi and a11y suggest we should disable tab results navigation when the interaction starts from the keyboard shortcut, so we preserve the behavior for mouse users.
But Dao has valid points about avoiding code complication just to serve what seems to be a barely used case, and that we don't know if tab users are more likely to start interaction with mouse or keyboard.

Taking the partial disabling path, we could monitor telemetry; if it drops drastically it means most tab users started interactions from the keyboard, then we broke most tab users. At that point we'd need either to reintroduce it somehow, or completely remove it because at that point the benefit is even smaller.
To be on the safe side I still suggest going the pref way, clearly this doesn't clear our threshold for users benefit, but it's an historical behavior and a non-trivial amount of users seem to use it at least once.