1627730 - Should we restore sessions by default for users who previously exited with open tabs?

Status: ASSIGNED

(User Research :: Project Request, task)

Description

[Chuck Harmston [:chuck]]

Request Description
We'd like to investigate a hypothesis that users who exited a session with n tabs would prefer to have that session restored when they start Firefox up again. Two specific questions:

1. Is this hypothesis valid?
2. If so, what should that value of n be?
3. If we change this behavior for existing users, how large is the potential embarrassment factor?
4. If this change only affected new profiles, how much of the embarrassment factor would it mitigate?

Deadline
No specific deadline

Priority Level
2 - Medium

Priority Level Description
This is an investigation we're doing by VP request

Supporting Information
N/A

Comment 1

[Rosanne Scholl]

Do you want to enact this behavior for some users and then ask them how they felt about it? Or do you want to just describe it to a general pool of users and ask the hypothetical?

Comment 2

[Chuck Harmston [:chuck]]

Since this is likely to be a relatively low-impact change on users, a lightweight study would be appropriate.

If we decide to make this change, we could do some followup retention observation and fine-tune our findings if we notice anything amiss.

Comment 3

[Vicky Chin [:vchin]]

Feel free to open a new item to track but along similar lines, should open with pinned tabs regardless of whether a user has session restored enabled?

Comment 4

[Romain Testard [:RT]]

We looked into this before (Previous experiment design available on https://docs.google.com/document/d/1X5NTJvsRnsyUFtiSkM_CPP2yqEcYccF0rrenrqzQgmY/edit) and since it ended-up being more complex than anticipated and priorities changed during the course of the explorations it was left in the backlog. The summary is:

• Original bug: https://bugzilla.mozilla.org/show_bug.cgi?id=735914
• There was a privacy risk with shared computers where people used to not having sessions being restored could be surprised by this behavior (husband shops for a wedding ring, closes Firefox and his wife opens Firefox next - adult sites are likely more frequent than the ring scenario though) that only an opt-in solution seems to mitigate. I had considered user satisfaction measurement though it's hard to make a call whether upsetting only 1% of users is fine.
• The probe seems disabled on release now (browser.session.restore.browser_startup_page) but at the time of previous exploration over 5% of release users had session restore enabled automatically (opt-in through about:preferences), which seems to indicate value for a set of users.

My fear with a study would be that you end-up with 99% of users being happy though it would be a hard call to impose this change to 1% of users who would be pretty upset - so the study may be inconclusive?

My recommendation for a test would be to use the same approach as we did with heartbeat where we promoted features and got users to opt into them, measured engagement and retention impact (we could also now measure satisfaction i suspect) . Based on results we could make investment decisions for CFR or onboarding implementations (https://metrics.mozilla.com/~bmiroglio/experiments/promoted-features/promoted-features.html)

Comment 5

[Chuck Harmston [:chuck]]

Thank you for the context, Romain!

Comment 6

[Raja Jacob]

Followed up offline. Currently under resourced for this. We will revisit this if priorities change.

Comment 7

[Raja Jacob]

We will resource this for 2nd week of May.

Comment 8

[Chuck Harmston [:chuck]]

There are some additional findings from data science that bear on this research. When someone is assigned, please see me for that background.

Comment 9

[Raja Jacob]

Kamyar has been assigned and he will work with Chuck to get this going.

Comment 10

[Chuck Harmston [:chuck]]

We’ll be starting research here with an in-product survey, targeting users with large numbers of tabs and seeking to understand their tab restore preferences and awareness of our session restore options. We expect that to be deployed next week, to run at least through the following weekend (13-14 June).

The findings from this study will be used to validate the direction, which we expect may include:

1. A pref-flip study, where we measure retention for users who have session restore enabled.
2. An unmoderated interview study to understand more nuance about tab usage and restore preferences.

Comment 11

[Patricia]

Heyu,

Is there any way to flag some session storage data as "not to be saved when the tab is restored" or should I try something else? For example, will a session cookie perform more safely in this instance? https://www.mygroundbiz.website/