1187961 - Does the number of aborted add-on installs increase with the duration of the add-on download?

Status: NEW

(Firefox :: General, defect)

Description

[Markus Jaritz [:markus] (UX)]

To see if optimizing the add-on install flow is worth the effort, we need data on whether the current flow leads to aborted add-on installs.
(And ideally if it discourages users from installing more add-ons. But is is probably difficult to measure.)

Comment 1

[Markus Jaritz [:markus] (UX)]

:gfritzsche do you know if we already have data on this? Or what is needed to get that data?

Comment 2

[Georg Fritzsche [:gfritzsche]]

I don't think we have any measurements yet that we could use for that - ni?mossop to confirm.

What exact abort scenario are we looking into?
* the user manually aborting the install
* aborts not triggered by the user (automatic cancelling, browser shut down, ...)

I assume once we are specific about the questions, we could record the times since download start for the relevant scenarios of addon installs to answer this.

Comment 3

[Markus Jaritz [:markus] (UX)]

(In reply to Georg Fritzsche [:gfritzsche] from comment #2)
> What exact abort scenario are we looking into?
> * the user manually aborting the install
> * aborts not triggered by the user (automatic cancelling, browser shut down,
> ...)
> 
> I assume once we are specific about the questions, we could record the times
> since download start for the relevant scenarios of addon installs to answer
> this.

Especially automatic aborts (of add-ons with valid certificates) as I would think we could count those as "Forgotten Installs" or "Lost Installs".

Currently users have to click to initiate (and allow) install, and again after the download has finished to install the downloaded Extension. Depending on size and bandwidth there can be considerable time between those clicks. So a lot of time for the user to forget about the install.

Comment 4

[Dave Townsend [:mossop]]

(In reply to Georg Fritzsche [:gfritzsche] from comment #2)
> I don't think we have any measurements yet that we could use for that -
> ni?mossop to confirm.

No, nothing. See bug 1201304 where I started trying to add some.