There and back again: the product dilemma
Why we changed one of our most important features — then changed it back.
On June 1st, we removed one of Arc’s most unique features.
As much love as we felt toward Arc’s “one window to rule them all” philosophy, we simply got so much feedback about the confusion that could result from having the same Today tabs across different Arc windows. So we changed it.
Two and a half months later, on August 24th, we changed it back — a total 180 on something we thought we had figured out.
Why? Probably better to show you than tell you.
Below is an internal memo from designer Peter Vidani, presented in its entirety.
On Syncing Today Tabs
June 30th, 2023
I want to give an update to the Today tabs syncing project we launched in Okra.
After much consideration, we have decided to reverse this change. That is, we’ve decided to go back to the Arc where Today tabs sync across windows and machines.
How did we get here?
The decision to sync Today tabs was made fairly early in the company. And while it was made with at least as much consideration as today, the low membership count made it difficult to validate.
As time went on, this differentiation in Arc became one of our top membership complaints. Especially from new members who expected tabs in windows to behave like other browsers and were fairly confused as to why we did it this way.
Over time, this feedback became particularly hard to hear as it contributed to Day-1 churn. Personally I believed this was, “a dumb reason for someone to not use Arc,” and that we should no longer sync Today tabs this way. When we were about to make the change, I told Ellis we could frame it as a “finally” launch, as in, “we finally did what you all wanted.”
But at launch and in the weeks that passed, a new top membership complaint emerged — that we had dropped an important, even critical, feature of Arc. Members point out that this change reduced the value of other core Arc features like Spaces and Pins. In our own language, the change violated an important value of ours: Never lose a tab.
Issues with syncing tabs
The previous model of syncing tabs was not perfect. Membership heard from new members about its confusing differentiation, but it also had some of its own inherent flaws.
If you are used to the way other browsers work, you were likely to close a tab by accident now and then. You may have opened a new window, saw the tab was available in the other window, and felt safe to close it only to see both were now gone.
Two instances of the same tab could diverge. If both had a form, video, call, or any dynamic state, the progression of one could wind up completely different from the other. Without much effort, data loss was likely.
Lastly, it did contribute to churn. For new members, it’s one of the first things they notice. And if they feel that strain before understanding spaces, boosts, renaming, or copying a URL without trackers, we just end up looking odd.
Going back to this model will not be a slam dunk. There are real areas to improve — Arc-specific problems that we introduced and need to close out for the system to work as expected. In some ways that will be education, and in some ways new features.
Issues with not syncing tabs
But the current, un-synced model has its own issues. While it appears to match the model of other browsers, Spaces and similar features introduce new ways of getting lost in the app.
We use a “Restore” button to help with the times you accidentally close a window full of tabs, but we hear from members and internally that this created a new feeling of friction, with a new decision to make.
Members are likely to lose tabs in a new way: closing a window not knowing they also closed another space’s tabs. Or closing a window thinking they were in sync because Top Apps and Pinned make them look so alike.
Finding tabs across windows is hard enough, but with multiple spaces it gets way harder.
The value of spaces goes down. Should I make a space, or a new window? Another way to feel like you’re using Arc wrong.
Put simply: synced tabs was not perfect, but it was easy to understand. It made Arc feel smaller.
Lastly, there’s the higher-level idea of product differentiation.
While not as important as truly improving the browser, there is long-term risk in following Chrome’s lead. Surely some of our growth can be attributed to Arc being a cool, hyped product. To some members, we can do no wrong. But once that inevitability fades, we’re left only with the product we built. And if in three years that product is “Chrome but used to be cool,” we’ll have no foundation to stand on.
Arc isn’t for everyone — we know this. But I think a more accurate framing is, “Arc isn’t for everyone today.” We’ve seen some members take over a year to finally come around. We can get them sooner if it’s 100% familiar on Day 1, but familiarity isn’t that interesting if we’re not solving anything new.
The next steps will be to get an internal build of tabs syncing across windows, which we’ll be kicking off in Cycle 2 of Potato. During that time we will also consider improvements we can make to the experience — fix things that were broken about the last model. We’ll consider these changes, their scope, and whether or not it makes sense to revert the change with or without them.
Some possible improvements:
Solve the issue of Window 1’s instance of a tab trailing off from Window 2’s instance of a tab.
Commit to a more Arc-like ⌘N
E.g. Little Arc, or new temporary space
Have a strong point-of-view on why a member should need a blank slate and what form that takes
More restore options in the Archive: restore spaces, days, folders, and other smart groupings.
Thanks to everyone who supported this project going in either direction over the last few months. Frankly it feels weird to me to make this change given I was so recently a champion of the current version! But we’re early as a company, and changes like this will eventually be almost impossible, so I’m grateful for the advice and encouragement.
I want to specifically thank Sherry who put an incredible amount of work into building this top Membership ask. The project was challenging not just technically, but to wrap our heads around and test against all the different ways it could go wrong.