Microsoft faced significant blowback for that ribbon change, and billions of dollars were spend on re-training hundreds of millions of office workers to do it. Because that change was in desktop software - users had the option of just sticking with the old version (which millions of users did) but you can't really do that with web applications.
Once you have worked with a product doing these kind of things, you'll understand you'll need to double or even triple the efforts to be able to support two branches of the same feature. Of course this doesn't apply to simple applications.
Confluence removed quite a few features from the new UI.
I don't know Circle CI but they probably have 2 orders of magnitude fewer users. Also in my experience with CI tools, few users actually manage them for many people and those people are either experienced or quickly become experienced with the UIs. They are usually not the average user getting confused by UI changes.
> Also in my experience with CI tools, few users actually manage them for many people and those people are either experienced or quickly become experienced with the UIs. They are usually not the average user getting confused by UI changes.
Well regardless, they still pulled off the harder technique.
That Atlassian Confluence model you gave had some serious issues, they they are going to turn off the option to keep the old UI as well, and it is an enormous strain on Engineering resources to keep two different versions of an entire application UI up to date.
One company that had done this a bit better was Salesforce with the old UI / Lightening UI switch.
Doesn’t that double the support surface? Bugs are still going to pop-up in the old interface, and so do you still tackle those or move engineer focus to the new interface? It seems weird to continue spending engineer time on an interface that is being phased out.
