Present user study/survey to "frequent" users

[cassidyjames]

I kinda brought this up on Slack but wanted to write up some longer-form thoughts here for posterity. This may well also be a bad idea, but I want to hash it out a bit.

If possible, I'd like to ask "frequent" users of the Music app how they use the app (e.g. for #671). For privacy reasons, we want to ensure any interaction with a remote server would be opt-in. We'd also want to ensure we're not spamming users, and I think we'd only really want to ask people who actually use the app—whatever we want to determine that means.

Current thought mulling around in my head:

• Use GSettings to locally log when the app has been opened, e.g. as last-launch
• When the app is launched, if it's been less than a week(? but more than a day?) since last-launch, iterate a counter, e.g. frequent-launches; if it's been more than a week, reset the counter
• If the counter hits 3(?), consider this a frequent user of the app, and show an infobar with a request to fill out a user survey.

An advantage of this approach is that we're "targeting" however we determine frequent users completely client-side and anonymously, and not even doing anything with that data other than showing an opt-in survey. A disadvantage is that it's pretty narrowly limiting the scope of people who would participate.

Other thoughts, in no particular order:

• We probably only want to consider users with items in their library?
• Should we count "plays" instead of "launches"?
• If we really wanted to go nuts, we could do a native, client-side survey that just submits the results to a remote URL; this is more work but would probably be a better overall experience. In that case, we'd need to write/host an open source server-side component instead of using a third-party survey service.
• Regardless, do we want/need some sort of validation to ensure it's actual users who are filling out the survey instead of randos who somehow got the link?

[Suzie97]

• Should we count "plays" instead of "launches"?

I think, launches should be counted, as I know people, including me who sometimes just open the music app to manage their music library. Personally, sometimes I just open the app when I've downloaded a new album, just to make sure that everything is well organised and properly named.

• If we really wanted to go nuts, we could do a native, client-side survey that just submits the results to a remote URL; this is more work but would probably be a better overall experience. In that case, we'd need to write/host an open source server-side component instead of using a third-party survey service.

I love this idea, but of course this will require more work. Also, this is really great for privacy as only the results will be sent to the remote URL.

• Regardless, do we want/need some sort of validation to ensure it's actual users who are filling out the survey instead of randos who somehow got the link?

It should be ensured that real users are involved. We don't want a lot of meaningless responses piling up and affecting the meaningful data.

[Suzie97]

@danrabbit as Cassidy is no longer involved with elementary, is this user survey going to be continued? If you think so, I could help the team in conducting it. Let me know if you need any help.

[danirabbit]

Hey @Suzie97, yes if you'd like to work on a survey here I would happy to work with you! I think probably the place to coordinate would be in the community Slack in the design channel there. If you could ping me once you've joined there we can talk about how to move forward :)