sync upgrade offer bar on every window is a distraction

RESOLVED WONTFIX

Status

()

Firefox
Sync
RESOLVED WONTFIX
3 years ago
3 years ago

People

(Reporter: chris hofmann, Unassigned)

Tracking

unspecified
x86
Mac OS X
Points:
---

Firefox Tracking Flags

(firefox37- affected, firefox38 affected, firefox39 affected)

Details

Attachments

(1 attachment)

(Reporter)

Description

3 years ago
In latest beta builds I'm seeing the bright yellow sync upgrade offer across the bottom of all open firefox windows.  It's quite distracting and there seems no way to actually reduce the number of windows where the offer is shown, or to dismiss the offer and not have it return.

This along with the hello page offer seems like we are stretching the bounds of promoting mozilla software to near the point of annoyance.

Is there any way we can consider reducing the number of offers we make to users or timing them to not hit more that one or two at a time?
(Reporter)

Comment 1

3 years ago
Created attachment 8572197 [details]
Screen shot showing all the bar + annimation of the hello promotion increases the distraction
We want to be aggressive about encouraging people to upgrade to FxA based Sync, but I take your point.

Note that we only show this infobar once per session, but maybe a reasonable and simple improvement would be to remove the notification from all windows once when it is removed from any of them?  We'd probably need to do that for all Sync notifications, but I think that's OK too.
(In reply to Mark Hammond [:markh] from comment #2)
> Note that we only show this infobar once per session, but maybe a reasonable
> and simple improvement would be to remove the notification from all windows
> once when it is removed from any of them?

We should certainly do that - I thought we already were.
(Reporter)

Comment 4

3 years ago
I'm seeing the yellow info bar more than once per session.  I haven't figure out all the things that trigger it but it looks like if I dismiss the yellow bar it reappears when I open a new window.
(Reporter)

Comment 5

3 years ago
did we do any performance testing on the patches that introduced this change?

I've got about 10 windows open and notice that when I open a new window I see a pretty long delay, beach balling, and lack of any input response via keyboard or mouse until the yellow info bar finishing rendering.
(In reply to chris hofmann from comment #5)
> did we do any performance testing on the patches that introduced this change?

No - we were just reusing the existing Sync notification bar (ie, I expect this to have the same performance characteristics as any sync notification) - which is apparently bad.

> I've got about 10 windows open and notice that when I open a new window I
> see a pretty long delay, beach balling, and lack of any input response via
> keyboard or mouse until the yellow info bar finishing rendering.

Possibly related to bug 1137087.  In that bug there is a suggestion to use the global notification bar, but it's unclear to me if that will change the performance characteristics (ie, I wonder if that global notification bar will have the same problem)
(Reporter)

Comment 7

3 years ago
from vague memory I recall that we had perf problems that would surface with the old iteration of the status bar that would some times trigger relayout of pages when it appeared and cause content to need to be resized.  Maybe we are hitting those cases again her.

Why can we just put this ad to encourage sync upgrades on the home/search page, or as one of the content services tiles.  That would probably hit nearly the same number of users with the offer and reduce the annoyance factor.
1. Does this bar show for all users or only users who had been using the old sync service?

2. This seems like a pretty heavyweight way of notifying users. (The intention is to show them a notification every time they restart?) Were alternatives like choffman suggested in comment 7 or putting a message in the hamburger menu or showing a bubble notification considered instead?

I'm tracking this for 37 as I want us to be clear on our approach. Note that there are only 3 Beta builds left (gtb Mar 12, 16, 19) to make any change.
status-firefox37: --- → affected
status-firefox38: --- → affected
status-firefox39: --- → affected
tracking-firefox37: --- → +
(Reporter)

Comment 9

3 years ago
again, I'm not seeing this everytime I restart or begin a new session.

I'm seeing this everytime I open a new window.  All windows get the yellow bar.

I dismiss the yellow bar by hitting the [x] on one of them and all yellow bars go away on all windows.

Then open a new window (then see the delays mentioned in comments 5 and 6) and all the yellow bars are back.

Its particulary annoying that the yellow bar does not have a "no thanks" button to keep this form being a continuing distraction every time I open a new window.  That seems like better UX where ever and when ever we users to do new things or try new things.   data on why users said "no thanks" is critical for understanding the chances of success in the marketplace.
Fun fact: the notification bars belong to legacy sync and were never intended to remain in the FxA-backed sync design (though they were not explicitly flagged for removal).

The reason we have not requested that they be removed is that they respond to sync problems before the new UI (in the toolbar menu and in Sync prefs). Typically I find that the new Sync UI requires a new window before it will present errors.

If the new sync UI can be as responsive to errors as the legacy notification bars, I recommend we remove them.

I hope this isn't muddying the waters.
(Reporter)

Comment 11

3 years ago
I think that's good insight.

Seems like we are overloading use of the error notification system to place an advertisement/invitation to upgrade.  I think that also increases the annoyance factor.

The yellow 'emergency'/take notice for quick action bar doesn't match with the importance of the content of the message from the users perspective or immediate needs.

Updated

3 years ago
Blocks: 1142276
Mark/Gavin - It seems that this promotion could use some more work before it's ready to ship. What do you think about backing this out / turning this off for 37 and creating a better promotion for 38?
Flags: needinfo?(mhammond)
Flags: needinfo?(gavin.sharp)
> 1. Does this bar show for all users or only users who had been using the old sync service?

Only for users who have been using the old sync service.  In fact it's triggered by a server-sent message in the sync protocol, so we can killswitch this on the server even if the code still ships in 37.

I also want to point out that this is not just a "promotion" in the same vein as e.g. the Hello messaging.  We're not just trying to get people onto a new system, we're trying to get them off the old system with some modest urgency, because we want to shut it down.

I don't think slipping to 38 would affect server-side timelines too badly though.
(Reporter)

Comment 14

3 years ago
Thanks for the comments in 13.  The message is key to help reach the goal.

We've rebuilt Sync to make it easier for everyone. Please upgrade...  ==  sounds promotional

We will be shutting down the *Firefox* Sync Service you are using soon. Please Upgrade if you want to continue using Firefox Sync... == sounds instructive

"No Thanks" (+ tell us why) seems like it ought to be an option and will be instructive in understanding why we can't get some users to move if we watch the feedback closely.

http://www.instructionaldesign.org/bad_error_messages.html definitely seems in play here.  We've been featured on that page before.  We probably don't want this to be another addition to that bad error notification hall-of-shame.

We also need to sort out the performance problems that seem to be associated with the way these yellow warning messages are working so indecisive users (maybe from installation that occurred long ago) don't get frustrated by the noticeable slowdowns.  e.g. users that are wondering 

  "what is this sync thing, how did it get on my computer, and why is firefox so damn slow lately... I think I'll just try another browser."
(since we're discussing timelines etc, /cc Edwin who is project-managing this on the services side)
(In reply to chris hofmann from comment #14)
> We will be shutting down the *Firefox* Sync Service you are using soon.
> Please Upgrade if you want to continue using Firefox Sync... == sounds
> instructive

I think this is the message the we want to send with a target date. 

"We will be shutting down the *Firefox* Sync Service you are using as of Firefox 38. You must upgrade Firefox Sync before then if you want to continue using this service."

Can we agree to push the message to 38 and figure out a target shutdown date for the service? 

Also, how many users are still active on the old service?
Ryan, can you weigh-in here?
Flags: needinfo?(rfeeley)
Lawrence, that string is far too long. The "We’ve upgraded Sync to make it easier for everyone” copy was written in close collaboration with Madhava. It’s a very unfortunate kind of upgrade that is hard to communicate as, from the perspective of users at least, it's a back-end change they would prefer not to hear about.

How many users are affected by the multi-window issue? (i.e. are legacy Syncers likely to have multiple windows)

Are we collecting metrics on how many Unlink (delete) sync?

The unfortunate thing is that no one will be happy with this forced upgrade. I don't think it's worth delaying migration for another 6 weeks for the new window bug.
Flags: needinfo?(rfeeley)
Less than 1% of daily users will see the yellow bar; only if you currently use sync 1.1.  If you disconnect sync or upgrade to sync 1.5 you will no longer see the yellow bar. 

I contend that the number of users with multiple browser windows is low compared to single window with tabs. Coupling that with 1% users on sync 1.1 and a fairly easy way to disconnect or upgrade, this should not be blocked for release in Fx37. 

One of the drivers here is that sync 1.1 has an infrastructure cost per user as well as dev and ops to maintain that service.  We believe sync 1.5 is the preferred option with its improve ease of use.

I'll talk with Bill and Chad for final approval from product.
We don't need to track for 37.
tracking-firefox37: + → -
Flags: needinfo?(gavin.sharp)
Flags: needinfo?(mhammond)
In discussing with product, UX, and eng - both leads agree that the multiple windows is sub optimal but should not block release.  We can pull in sumo and qa if desired, though Karl has been testing throughout. Some may feel the language of the message is ideal, but this shouldn't be consider a blocker at this time.  Note that Fx37 browsers will display a stronger message about the decommissioning of sync 1.1.  Lawrence and Chris let me know if you want to talk and we can pull together a meeting with stakeholders.
(Reporter)

Comment 22

3 years ago
Yeah, I'm interested in talking to anyone that will listen.  I don't think this is going to help us achieve any goals that I can see.  We can say it affects only 1% of users but that translates into a million or so people.  A 100k here, a 100k there and that just adds to our declining user numbers.  We ought to figure out ways to track the number of people that are inspired by the current set of patches and actually migrate from one version to the other (that the goal here right?), and how many people we just loose (don't show up on either version of sync) as the result of this planned roll out.  Those would be the lost Firefox/sycn  users.
(Reporter)

Comment 23

3 years ago
When we want to try and encourage our users to upgrade or switch to new services lets respect their right to choose and give them reasonable defaults and options.  That's what our competitors are doing.

http://techfrag.com/2015/03/18/google-pleads-with-firefox-users-to-dump-yahoo-search-in-its-favor/

notice the "no thanks" option.

upgrade, or dismiss the offer and get nagged again are not reasonable options and should not be part of our product, now or any time in the future.  Let's make great software.
We are tracking migration rates with telemetry/FHR data. We'll invest in improvements to the flow depending on the response we see.

This particular notification is meant to be noticeable. That it is so persistent in this particular use scenario (you ignore it and then open other windows) is a potential source of annoyance, but relatively unlikely to occur much in practice, and there are easy ways to mitigate it (complete the flow or disconnect Sync).

Comment 25

3 years ago
No perfect answers for this problem, and I think a lot of the folks Edwin talks about above have come up with the least worst solution for the <relatively> small number of people who will end up seeing the bar in each open window.  My view is the high order bit we are going to shut off this service later this year, that decision is made (I broke the news to a security engineer on old Sync yesterday) and there's no choice about that, let's do the best for our users by being clear and direct with them and not "string" people along. I would want to get that message clearly, in more than one place if possible. Chris, if you want to talk about other options, please get a meeting together with those who want to be there.
(Reporter)

Comment 26

3 years ago
I've think I've said all that I can say here.  I'll summarize.

- not having a "no thanks" option on the info bar traps users and is terrible UX (affects all old sync users) 

- yellow info bar that contains promotional v. instructive text also bad UX (affects all old sync users)

- yellow info bar constantly returns on each new window action and users aren't pointed at reasonable steps with options to get a better outcome (affects all users).  Gavin, thanks for the tips where I can go to shut the old sycn down.  I disagree that this is an intuitive step that any user will know about and will take.

- yellow info bar is scribbles on all open windows (affects users that open multiple windows)

- yellow info bar tickles pretty severe performance and lock out problems across the entire browser (affects users that open multiple windows)

I think that each these problems have been individually rationalized away by everyone providing feedback, and the collection of all of the problems reflects a lack desire and care to produce great software.  Fixing any or all of the problems would be better than what we are planning, but I haven't been able to convince anyone of that.

Sorry for wasting everyone's time on this.  Someone should mark wontfix.
(In reply to chris hofmann from comment #26)
> and the collection of all of the problems
> reflects a lack desire and care to produce great software.

I understand where you are coming from, but I take great offense at this statement.
(In reply to chris hofmann from comment #26)
> I think that each these problems have been individually rationalized away by
> everyone providing feedback, and the collection of all of the problems
> reflects a lack desire and care to produce great software.

Uncalled for.

This isn't about lack of care or desire; it's about a disagreement about the relative impact, and the need to prioritize given limited resources. Fine for you to disagree, but not fine for you to impugn our motives.
(Reporter)

Comment 29

3 years ago
also sorry if I've offended people.
(Reporter)

Comment 30

3 years ago
gavin, your right we should continue the discussion about what defines the idea of "making great software" in another forum. that's something that cbeard mentions as the reason he returned to mozilla, and something we all need to thing about and act on.  

We definitely have differences of opinion about what that means or what it should mean and they surface again in this bug.

For me there is a lot in this bug that holds true about looking at what we done, evaluating it, and if something does look like it will achieve its goals, and it doesn't feel right, we keep digging deeper until we get it right.  nearly everyone that commented in this bug said "yeah, this doesn't feel quite right..."   There is also the idea of anything worth doing is worth doing well.  if we don't do these things we aren't on track to make great software.  That's my definition. You can disagree with that, but I think cbeard wants us all to work on our definition of what great software is, and how we can build it.

markh,  definitely didn't mean to offend you personally.  I hope you can believe that.  I was referring to "the collective we" as failing on this bug.  Me included.  Especially me.  Maybe if I'd started the bug at comment 26 things might have gone better for making a case to take action on these issues.  Thats the taking greater care part I was referring to.   I personally regret not getting past the "this doesn't feel right to me" part faster, and diving deeper and moving quicker to a comprehensive view of all the issues and their potential impact.
Chris, I welcome any feedback on the full user experience plan: http://is.gd/sync_migration_ux_pdf

Each state in the flow links to its accompanying design.

Some background: because few users were able to grasp the unconventional UI of the original sync, we created a Firefox Account-backed Sync that used a well-understood login and password sign up. It proved to be successful and even with our shrinking user base we attracted 6× the users in ⅙ the time.

Unfortunately, because of encryption, we could not automatically migrate users over. From their POV, they already have credentials so why are they creating a new account? They stand to realize no benefit from the new sync because it's already working for them and only the authentication has changed.

Our messaging was carefully written to let users know why old sync is shutting down (it's easier for everyone) without making false claims about new benefits. The messaging is urgent because once syncing is shut down, users who eventually sign up are at the mercy of merges.

I like delivering wow experiences as much as any designer, but it would be the inappropriate tone in this context.
we weren't able to fix this in time of shutting down sync 1.1 deadlines. closing.
Status: NEW → RESOLVED
Last Resolved: 3 years ago
Resolution: --- → WONTFIX
You need to log in before you can comment on or make changes to this bug.