Confluence User-Interface
Some time ago I got a mail from someone asking what I meant when I wrote "The out-of-the-box interface sucks big time" (see section on Confluence). I wrote back a long mail, but there was no follow-up reply, and because I've got several mail from people who also struggles with that very default interface, I have decided to post that mail here so that a) the information isn't lost, and b) you now know what I think about the Confluence default interface (at least before version 2.3). Also, we may package up our enhanced interface as an open-source thingy so that others can use it, but I'll have to sort some stuff out first. Here we go ;
> Could you elaborate some more on exactly what
> sucks about Confluence's UI so much to you?
Not so much to me, because I'm a geek, and with a bit of clicking and a bit of huffing I learn the paradigm of the interface and start using it. My users, on the other hand, hated it. Every test we did left them confused and uncertain about where to click next. My users are not geeks, nor incredibly computer litterate. And I'm an usability guy. Sparks surely would fly.
Maybe a few specifics are in order. I'll talk about some of the biggies ;
Where am I? If we are true to the Wiki form, every page is a page, but in Confluence this is not so; there are pages, label-pages, news, configuration, reports, and so forth. How does the interface reflect where you are and state you're in? The breadcrumb is really the only way that this is reflected, apart from the content (or sometimes hints within it) itself. but the difference between the content parts are cognitive hard to tell apart; the information shape is too similar across them all, and so we become relient on analysis of content instead of cognitive recognition. (basically the tabs aren't clear enough, nor consitently "content" based as sometimes you use them for functionality, and the same real-estate and concept is swapped depending on what you do and wherer you are)
Further on to this is the confusion people have when they're on a page and they can't edit it. (Difference between a page and a news item, or even a list of labels when things get a bit unclear) using tabs as functionality is also confusing here. A bigger problem occurs after some use of the system; why does a page have four function based tabs when "pages" is one of many content tabs? Where did my page go? Aren't pages part of the space? What is the difference between a page and a news item? And I could go on and on about using titles of pages as persistant links and the tree-structure imposed. :)
Why can't I label stuff as "fish fingers, bollocks" as with Flickr? Why can't we do controlled vocabularies? (This stuff isn't that hard to do; we've hacked Confluence to support it!) How can we do facetted navigation which better suits complex Wikis? Why aren't there better ways of dealing with lots of spaces? Why is added metadata to a page so hard (and with macros, so ugly)? (In fact, why aren't there a really neat user interface to attach metadata to pages, like properties, using bandana? Macro properties only works with extreme geeks, not human beings!) I can go on and on, and I'm pretty sure that there's *good* answers for why things are the way they are, and I certainly understand that a lot of these things are hard to change as they become legacy. In fact, I suspect most of the default user interface is built up over time and no-one dares to change it because it is very complex and rigid. (For example, why are there no Velocity templates for dealing with labels? I suspect a rushed job?)
Having said that, though, Confluence is fantastically flexible, and can do pretty much everything we want it to do, but at some point we need to add users to it, and that's when things start to break down a bit. We can't embed metadata in content, as normal people would balk at the macro scripts!
As a technologist I understand every decision that has been made, why things work the way they work, and I can figure out how to do pretty much anything I like; I grasp the paradigms and I can get around the interface to make it do what I do. I understand that conceptually a page belongs under the "pages" tab, but the interface doesn't reflect this and adds confusion. I understand how I can use the system to create hidden pages to do special stuff, and then include that content into a column of special info, and so forth. I can figure out what files involve the labels (and just that they're called labels instead of tagging makes things further tricky; another word to learn that means the same as something else) and implement a scheme of local and global controlled vocabulary. I can figure out how to have access control outside the realms of user administration, or to do news aggregation across spaces and repurpose these in external applications, and more. But I'm telling you, it hasn't been an easy ride! :)
I brought Confluence and JIRA into this organistation, and we use them both, even in synch, and technically it fits us like a glove (although with a few added features it would make a better kill, of course), but the user interface failed again and again. People will not battle with a user-interface; they'd rather forget about the tool alltogether. In our scenario, these are normal plain people trying to do very plain stuff. They're not geeks. Even the very concept of a Wiki is scary to them, and as such the interface must be as gentle as possible.
...
Anyways, I could ramble on about the user interface, the cognitive challenges it imposes, the confusing paradigms, Confluence as part of a greater set of "web 2.0" tools (sorry for using that expression), information architecture in a Wiki world, persistant identification (and how Confluence fails especially through repurposing of content), facetted navigation, content semantics and news aggregation (which is a neat business model in itself!) and so forth. I don't mean to give the Confluence guys pepper (hmm, a Norwegian expression, I fear), but I've spent over a year in the complex belly of the beast in trying to make the interface reasonably user friendly for our very boring normal users. So, um, forgive me. :)
> Could you elaborate some more on exactly what
> sucks about Confluence's UI so much to you?
Not so much to me, because I'm a geek, and with a bit of clicking and a bit of huffing I learn the paradigm of the interface and start using it. My users, on the other hand, hated it. Every test we did left them confused and uncertain about where to click next. My users are not geeks, nor incredibly computer litterate. And I'm an usability guy. Sparks surely would fly.
Maybe a few specifics are in order. I'll talk about some of the biggies ;
Where am I? If we are true to the Wiki form, every page is a page, but in Confluence this is not so; there are pages, label-pages, news, configuration, reports, and so forth. How does the interface reflect where you are and state you're in? The breadcrumb is really the only way that this is reflected, apart from the content (or sometimes hints within it) itself. but the difference between the content parts are cognitive hard to tell apart; the information shape is too similar across them all, and so we become relient on analysis of content instead of cognitive recognition. (basically the tabs aren't clear enough, nor consitently "content" based as sometimes you use them for functionality, and the same real-estate and concept is swapped depending on what you do and wherer you are)
Further on to this is the confusion people have when they're on a page and they can't edit it. (Difference between a page and a news item, or even a list of labels when things get a bit unclear) using tabs as functionality is also confusing here. A bigger problem occurs after some use of the system; why does a page have four function based tabs when "pages" is one of many content tabs? Where did my page go? Aren't pages part of the space? What is the difference between a page and a news item? And I could go on and on about using titles of pages as persistant links and the tree-structure imposed. :)
Why can't I label stuff as "fish fingers, bollocks" as with Flickr? Why can't we do controlled vocabularies? (This stuff isn't that hard to do; we've hacked Confluence to support it!) How can we do facetted navigation which better suits complex Wikis? Why aren't there better ways of dealing with lots of spaces? Why is added metadata to a page so hard (and with macros, so ugly)? (In fact, why aren't there a really neat user interface to attach metadata to pages, like properties, using bandana? Macro properties only works with extreme geeks, not human beings!) I can go on and on, and I'm pretty sure that there's *good* answers for why things are the way they are, and I certainly understand that a lot of these things are hard to change as they become legacy. In fact, I suspect most of the default user interface is built up over time and no-one dares to change it because it is very complex and rigid. (For example, why are there no Velocity templates for dealing with labels? I suspect a rushed job?)
Having said that, though, Confluence is fantastically flexible, and can do pretty much everything we want it to do, but at some point we need to add users to it, and that's when things start to break down a bit. We can't embed metadata in content, as normal people would balk at the macro scripts!
As a technologist I understand every decision that has been made, why things work the way they work, and I can figure out how to do pretty much anything I like; I grasp the paradigms and I can get around the interface to make it do what I do. I understand that conceptually a page belongs under the "pages" tab, but the interface doesn't reflect this and adds confusion. I understand how I can use the system to create hidden pages to do special stuff, and then include that content into a column of special info, and so forth. I can figure out what files involve the labels (and just that they're called labels instead of tagging makes things further tricky; another word to learn that means the same as something else) and implement a scheme of local and global controlled vocabulary. I can figure out how to have access control outside the realms of user administration, or to do news aggregation across spaces and repurpose these in external applications, and more. But I'm telling you, it hasn't been an easy ride! :)
I brought Confluence and JIRA into this organistation, and we use them both, even in synch, and technically it fits us like a glove (although with a few added features it would make a better kill, of course), but the user interface failed again and again. People will not battle with a user-interface; they'd rather forget about the tool alltogether. In our scenario, these are normal plain people trying to do very plain stuff. They're not geeks. Even the very concept of a Wiki is scary to them, and as such the interface must be as gentle as possible.
...
Anyways, I could ramble on about the user interface, the cognitive challenges it imposes, the confusing paradigms, Confluence as part of a greater set of "web 2.0" tools (sorry for using that expression), information architecture in a Wiki world, persistant identification (and how Confluence fails especially through repurposing of content), facetted navigation, content semantics and news aggregation (which is a neat business model in itself!) and so forth. I don't mean to give the Confluence guys pepper (hmm, a Norwegian expression, I fear), but I've spent over a year in the complex belly of the beast in trying to make the interface reasonably user friendly for our very boring normal users. So, um, forgive me. :)
6 Comments:
Alex, I would love to see what you guys have come up with (as well as some tips on customisation!).
Hi Alex,
Thanks so much for posting this. Your criticism is smart and well thought out. It's just the kind of thing we need to hear.
We're planning to revisit the UI in the upcoming Confluence 3.0, and I hope that we can make some strides improving it soon. The non-geek experience is something we're very concerned with, although it's harder for an office full of programmers to figure out. Your experience with your usability testing would be really valueable.
We would really love to see what you've come up with, even if you can't share it with the world. Of course, if you could contribute to the library that would be even better. Either way, feel free to contact me.
Thanks again for your thoughts. We're definitely listening.
Cheers,
P.S. in a possibly related note, we’re looking for a UI developer/designer to join the Sydney team, so if you (or anyone you know) is interested, give me a shout. Job description here: http://www.atlassian.com/about/jobs/ui-designer.jsp.
Have you tried Socialtext's UI?
http://www.socialtext.com/node/105
I wonder if you've ever complained about Word having too many features.
Anyway, we're about to roll out Confluence at our University in the UK - the biggest changes we've made to the UI are simplification all over the place. Things like taking Page Info away from the main tabs; make browse space mostly about import/export and permissions (and make the 'browse space' link less prominent), etc.
We've had wikis running internally (not Confluence) for about a year and it seems that so long as you stick to a simple navigation (we use a left nav menu), and 'view' page and 'edit' page - the users get it. Mostly it just means concealing some of the up-front complexity, but we're pretty confident that power users will find it anyway.
The worst thing we've found in our usability sessions so far is no-one really knows which links are which colours and why. There seem to about ten of the swine.
Anyway, I'll follow this up in my office and might post something about it on my own blog too. Interesting stuff.
Pat: Yeah, I think I'll blog about it soon, and possibly wrap it up in a package you can download yourself and play with.
Jonathan: No problem.
Ross: Yes, and in what we've done is actually very similar to what they're doing. Some things are really hard, though, due to what objects Confluence dumps in what context and on which pages (for example, all labels don't have Velocity templates at all), but we've tried hard to find the mostly undocumented objects we need and reshuffle the interface. Some times we've just reverted to some Confluence defaults, but with quite a bit of clean-up.
Pip: Of course I've complained about Word, but not at the features but how the present them. this is not about functionality (of which Confluence is absolutely amazing), but about usability. Taking info tab out also takes out functionality that isn't always easy to wrap in the origin page (like PDF/Word conversion), but yeah, pretty much all we've done is to simplify and create clearer areas in which you'll find what information. Some times it's easy, other times not. Links we've too have tried to clearify (we've now got three types of links). Anyways, I'd be interested in what you've done and how you did it. I'll post a follow-up on this soon, as I've received quite a lot of private mail about this as well.
Ross is the founder and CEO of SocialText, so it's "what you're doing" :)
"like PDF/Word conversion" - I realise there's more and you're generalising, but the default PDF export is in the top right of all pages, we've added the Word icon up here too (for the time being).
We're only really in the early stages, but I'll see if I can get some screenshots out. We're mostly curious about what other people have done to or about the Dashboard.
Links to this post:
Create a Link
<< Home