1828172 - Launch experiment for ASRouter spotlight message to encourage users without FxA accounts to create one

Status: RESOLVED FIXED

(Firefox :: Messaging System, task, P2)

Description

[Mike Conley (:mconley) (:⚙️)]

The spec for the spotlight modal is here: https://www.figma.com/file/qe9yX8x51jjbt6DJYgayg2/Device-Migration?node-id=2261-19163&t=nTM9i8p1GgGtIwHS-4

The modal should:

• Primary CTA takes user to SUMO page that bug 1828114 links to.
• Secondary CTA - close the spotlight modal.
• “x” icon closes the spotlight modal.

Ultimately, we're hoping to message users with the following characteristics:

• Existing Windows Vista/7/8 users
• Do not have a Firefox Account

This message should not overlap with other experiments and spotlight modal regarding “default”/ “pin to taskbar”

Show spotlight modal 1 time per user

Comment 1

[Shane Hughes [:aminomancer]]

Hey Mike, do you know which trigger we want yet?

OMC agreed it's most conventional for these experiments to land just the strings in train. We can do either of the targeting points in message targeting or in experiment targeting, but it seems reasonable to target the experiment to existing users for windows version < 10, but target the message to !isFxASignedIn, since the latter is prone to change.

Comment 2

[Mike Conley (:mconley) (:⚙️)]

I've asked Tati in Figma for more information on the trigger.

Comment 3

[Mike Conley (:mconley) (:⚙️)]

According to Tati, the trigger for this message is on launch.

Comment 4

[Mike Conley (:mconley) (:⚙️)]

In Figma as of this writing, that's the default browser check trigger. We're currently chatting in Figma about options for deferring that if necessary.

Comment 5

[Mike Conley (:mconley) (:⚙️)]

A note that the Figma spec calls for the spotlight modal to be both "hard" and "soft" dismissable:

1. A "remind me later" link below the main CTA dismisses the spotlight, but allows it to re-open some days later (number of days TDB - whatever we've done historically, like 7 days, is probably fine)
2. An "X" button on the top right of the spotlight modal will dismiss it, and it will not resurface.

Comment 6

[Shane Hughes [:aminomancer]]

Gonna file a separate bug to land the strings, since I have time to handle the strings. Someone else can take the experiment, or else I'll pick it up after closing up my outstanding experiments

Comment 7

[Punam Dahiya [:pdahiya]]

Offtrain and recipe should be created in Fx115 beta for QA to test

Comment 8

[Shane Hughes [:aminomancer]]

We've decided to launch the "control" version of the spotlight (the one whose strings are on train) as a rollout, and concurrently launch this experiment targeting 4 locales with 3 branches:

1. Control branch, single message, you can copy it from the rollout version that I'll implement
2. Treatment A, "multi-message" configuration, 4 messages - one for each of the 4 targeted locales - each of those messages will have message-level targeting for its locale. This means someone enrolled in Treatment A will download all 4 of those messages, but which one is shown will depend on their locale. So each of the messages are specific to 1 locale and therefore can't overlap.
3. Treatment B, same approach as Treatment A, just different copy.

Sizing info is forthcoming, will update when I have more info.

The rollout work is in bug 1832696, and I'll take the lead on that one. Hanna, you'll be able to just copy your control branch from the rollout, so I'll let you know when that's finished.

As for Treatments A and B, they will need to use the multi-message format, which if you haven't heard is a way to squeeze multiple messages into a single branch. The format looks like this (with no other properties):

{
  "template": "multi",
  "messages": [
    { "id": "MSG_1", ... },
    { "id": "MSG_2", ...}
  ]
}

So within each of those branches, you'd have a messages array with 4 messages, basically 4 localized versions of the message with raw strings for that language.

But to accomplish that, we first need to get the strings translated. Someone will need to file an issue (would appreciate if you have time for that Hanna?) to translate the strings for Item 2's experiment versions #1 and #2 from this copy doc. Under Item 2, you can see 3 screenshots for the 3 different branches. But we only need translations for the 2nd and 3rd versions, the ones labeled experiment version #1 and experiment version #2, because we already landed the strings for the "control" version (the first one under Item 2) in bug 1828753.

The issue that needs to be filed can just be modeled after this one. It's the same process so you can treat that issue as a template. In this case, we have 4 locales instead of 3. Those are:

English (en), German (de), French (fr), Italian (it)

If you need any additional information, check the experiment brief or feel free to ping me on slack.

Comment 9

[Hanna Alemu [:hanna_a]]

Thanks for all the info! I've filed this issue for the translation https://github.com/mozilla-l10n/nimbus-l10n/issues/9