1112344 - double opt-in not respected for FxA address signups?

Status: RESOLVED INVALID

(Websites :: Basket, defect, P3)

Description

[David Durst [:ddurst]]

While QAing newsletter signup integration for Marketplace, we noticed that the `optin` param we're passing is respected when the signing up email address is NOT associated with a FxA, but is effectively flipped when it IS associated with a FxA.

As we have no way of checking that (currently), the confirmation message we show to successful signups is consistent with our API call (which assumes optin), as prescribed by Jessilyn in https://bugzilla.mozilla.org/show_bug.cgi?id=1097152#c3. The problem is that the confirmation message is wrong, then, for FxA address signups, due to a change in behavior once the request goes to basket.

(Marking this as a P1 because I was told that we could not launch if the functionality is not behaving as expected.)

Comment 1

[David Durst [:ddurst]]

Apologies -- perhaps things got lost in IRC translation. I was told that this was related to FxA, but apparently it's about the address being mozilla or not.

This was confirmed with tests just now:
A. logged-out signup with gmail (FxA) address 
B. logged-out signup with mozilla (FxA) address
C. logged-in signup with gmail address
D. logged-in signup with mozilla address

A and C received the "confirm your subscription" email; B and D received the "Welcome!" email.

Comment 2

[Elizabeth Hunt[:ezoehunt]]

Based on durst's recent testing, we can deliver a conditional "thanks" message based on whether or not the user entered a mozilla address.

If user entered a non mozilla address, this message (as in UX docs): 
Thanks for signing up! Please check your inbox to confirm your subscription.
You’ll receive an email from [e-mail] to confirm your
subscription. If you don’t see it, check your spam filter. You must confirm your
subscription to receive our newsletter.

If user entered a mozilla address, this message:
Thanks for signing up!
Please be sure to add our sending address: [e-mail] to your address book to ensure we always reach your inbox.

Comment 3

[David Durst [:ddurst]]

We were right the first time! \o/   Our tests were wrong! ಠ.ಠ

Ignore my comment 1. The issue as described in the Description is confirmed:
https://www.dropbox.com/s/ky2hj31mxihh06h/newsletter-two-acct-mozilla.mov?dl=0
and
https://www.dropbox.com/s/2v57ujrtmeb8dvk/newsletter-two-acct-striemer.mov?dl=0

Comment 4

[Paul [:pmac] McLanahan]

Oh wow. Just doing some bug gardening and found this. How big an issue is this still? Should I alert people and get it on the schedule? I believe this is happening because there's no real distinction between a normal newsletter user and an FxA user in exact target, and we have a mechanism that avoids people having to opt in multiple times. So I think this is working as designed, but it could be that the design isn't actually what's desired. I'm NI our email program manager to see if she has opinions.

Apologies for the delay.

Comment 5

[Jessilyn Davis]

Okay - this is what I can tell you:

If your email address already exists in ExactTarget on the Master_Subscribers data extension the double opt-in process is not triggered and you should get a welcome email. If you're staff - chances are you are on Master_Subscribers already per our integration with mozillians.org.

To test this functionality, you need to try all 4 testing scenarios in #c1 with email addresses that have *never* been put inside of ExactTarget before.

[e-mail]  will count as a unique email address.

Does this make sense? Ddurst, can you retest?

Comment 6

[David Durst [:ddurst]]

Forwarding to QA, and reducing the priority, as this is long since live.

Comment 7

[krupa raj[:krupa]]

When I use [e-mail], I get an email to subscribe to the newsletter which I think is the expected behavior. durst, can you confirm?

Comment 8

[Jessilyn Davis]

(In reply to krupa raj[:krupa] from comment #7)
> When I use [e-mail], I get an email to subscribe to the
> newsletter which I think is the expected behavior. durst, can you confirm?

Krupa - thanks for testing! If you are *not* signed in as a Firefox Account user with that email address, then your test is the expected behavior.

If you *are* signed in as a Firefox Account user with that email address ([e-mail]), you should have received a "Welcome!" email instead of the confirm your subscription email.

Both these scenarios need to be tested to make sure that the opt-in is working as expected.


Reasoning:
If you are logged into a Firefox Account, you have told us that you are the owner of that email address (you've previously had to confirm your email address to use the Account).

If you are not logged into a Firefox Account, we want to make sure that you own/have access to the email address you used to sign up for the newsletter before we start emailing it the news.

Make sense?

Comment 9

[krupa raj[:krupa]]

Let me list the STR i actually followed to test comment 8.

1. Create a new FXA user on Sync. ([e-mail])
2. Confirm the account by clicking the link in your email
3. Go to marketplace.firefox.com 
4. Without signing in to marketplace, enter your new fxa account to sign up for the newsletter
5. Check your email


STR 2:
1. Create a new FXA user on Sync. ([e-mail])
2. Confirm the account by clicking the link in your email
3. Go to marketplace.firefox.com 
4. After signing in to marketplace using the same fxa account, enter your new fxa account to sign up for the newsletter
5. Check your email

observed behavior:
You are asked to confirm your subscription as seen in the screenshot below. 

screenshot: https://www.dropbox.com/s/kbhdt3ydmk3gnwn/Screenshot%202015-07-10%2016.30.33.png?dl=0 shows that I was prompted to confirm subscription

Comment 10

[Jessilyn Davis]

:ddurst - looks like we were both thinking about this bug!

I've created this testing doc to dive into this further:

https://docs.google.com/spreadsheets/d/1xtOKfA10-a0yeoxc9uwkDZ9iXY2zRwR9_MLDav14yFs/edit#gid=676619064


I'm hoping to make some time do do some thorough testing this week - but if anyone wants to dive in and start rocking it - I'd welcome any/all extra hands/eyes on this.


:krupa - thank you for your testing! Per your comment in #c9 - it looks like double opt-in is not handling as originally spec'ed. I want to double check the subscription settings for your tests as well - thanks for putting in your test email addresses. I have it on my todos (this week) to look up the test emails and confirm the different subscription settings (as part of that above doc).