You don't even need metrics to figure out that, for example:
> The old design was worse for many of the things we value and try to improve. Like how much people share and converse with their friends. [old design: http://img.svbtle.com/gpnggaky8d1gog.jpg]
The "only icons" on the left + "icons AND text labels" for "Share/Like/Comment" + much higher text density on the right constantly shift your thinking form "photo/visual" to "reading/writing", it mind-fucks you in a very subtle way, so your brain ends up focusing on the only thing that makes sense: (a) the overall visual structure (that was beautifully designed, I admit) and (b) the photos.
If you want user engagement beyond the "click like" level you need to focus people towards the "reading/writing/verbal" mode of interaction. Like, if you have to read a button's label to know what it does, this puts your mind into "reading mode" so the comment that immediately follows the link/button has more chances of actually being read and of the people actually writing a reply instead of just clicking a like and staring at a cute picture. They really got this right (either through someone's insight or metrics, I dunno) with the "Like Comment Share" links - getting rid of the icons pushed you more towards "text/words mode" thinking, which is exactly the mode you need to be in to actually post a comment, and a comment is more content so it will be a positive feedback loop for even more and so on.
(Also, another obvious bad idea was the left bar - while theoretically good for screen estate, it's essentially "mystery meat navigation" to unsophisticated desktop users. Also it puts less focus on the Apps. Also by putting the active contacts list in the bottom left corner pushes them out of your mental focus.)
EDIT+: ...now that it really got me thinking of it, I can't believe how incredibly bad the "old new" design was. How did they even chose to deliver that? It looked like textbook "design driven design", it didn't focus at all on how the users think and what they actually do on Facebook.
Totally agree with that last line. I think the failure was that they came up with a design that looked really good, but fell short on the interactions that mattered.
It's a classic UX design problem: insular critique led the design in a certain direction, everyone loved it because it was pretty and "content focused." Then someone pointed out the user testing results, and probably used data to support an argument about how it didn't actually work that well, which made a whole bunch of people go "oh yeah, huh" and finally it fell flat.
Happens all the time. This blog post was spot on, and I found nothing unusual about it.
The unusual part is that an organization as large as Facebook had the ability to stop the momentum of a very large project because it wasn't that great. That's a lesson to take to heart.
I think it's fairly easy to come up with reasons after the fact why a design is bad, but it certainly didn't look bad and many people in the more tech community felt it was better than the new redesign: in particular the guy who wrote the article to which this one is a response to.
Maybe I exaggerated a bit and I wouldn't have considered it "incredibly bad" before having the latest version to compare it to, I agree.
But the "photo/visual" vs "text/reading/verbal" switching it imposed on you was obvious in it at the beginning. I know, this is not something that pops up as a "big red flag" to most other UI/X people, but to me it really does. Take HN as an example: the text only interface keeps you in the zone and focused on actually reading peoples' comments and replying meaningfully to the. Take http://www.microsoft.com/ as another one: the pictures swallow your focus and they make it very hard for you to actually read the link texts and decide what to click on to get to what you wanted. Also http://www.bing.com/ - you use it to search for text, but it bombards you with pictures, so you may actually end up getting worse search results not (only) because the search engine is worse, but because you mind is not "words focused" anymore and you fail to come up with more appropriate search phrase. Also compare https://www.ruby-lang.org/en/ and https://www.python.org/ on one side vs. http://www.scala-lang.org/ on the other side (the picture and design totally "unfocus" and disorient your mind so it takes you longer to find the relevant content, like "top language features" even if they are right there on the landing page and very well explained too).
> Irrelevant. The tech community is not a representative sample of Facebook's user-base.
Indeed they aren't, but who do you think designers interact with on a day to day basis? I think that the tech community's thoughts are important as they influence other tech people's thoughts and, therefore, might shift them away from reality.
> And looking good and working well are two completely different things.
Again agreed, which is why using metrics is a sound approach. You just have to make sure you're using the right ones.
> The old design was worse for many of the things we value and try to improve. Like how much people share and converse with their friends. [old design: http://img.svbtle.com/gpnggaky8d1gog.jpg]
The "only icons" on the left + "icons AND text labels" for "Share/Like/Comment" + much higher text density on the right constantly shift your thinking form "photo/visual" to "reading/writing", it mind-fucks you in a very subtle way, so your brain ends up focusing on the only thing that makes sense: (a) the overall visual structure (that was beautifully designed, I admit) and (b) the photos.
If you want user engagement beyond the "click like" level you need to focus people towards the "reading/writing/verbal" mode of interaction. Like, if you have to read a button's label to know what it does, this puts your mind into "reading mode" so the comment that immediately follows the link/button has more chances of actually being read and of the people actually writing a reply instead of just clicking a like and staring at a cute picture. They really got this right (either through someone's insight or metrics, I dunno) with the "Like Comment Share" links - getting rid of the icons pushed you more towards "text/words mode" thinking, which is exactly the mode you need to be in to actually post a comment, and a comment is more content so it will be a positive feedback loop for even more and so on.
(Also, another obvious bad idea was the left bar - while theoretically good for screen estate, it's essentially "mystery meat navigation" to unsophisticated desktop users. Also it puts less focus on the Apps. Also by putting the active contacts list in the bottom left corner pushes them out of your mental focus.)
EDIT+: ...now that it really got me thinking of it, I can't believe how incredibly bad the "old new" design was. How did they even chose to deliver that? It looked like textbook "design driven design", it didn't focus at all on how the users think and what they actually do on Facebook.