Graphs that Distract
August 23, 2010 | 4 Comments
Edward Tufte coined the term chartjunk to describe aspects of graphs that distract from the information being conveyed. As chartjunk has gradually disappeared, something else has taken its place – a deluge of graphs.
Death by Graphs
I’m going to pick on Google Analytics. Take this screenshot:
There are nine different graphs here, all showing something different. That’s way too much information. If I was asked to summarise what I’d learnt from this page, I wouldn’t be able to think of a single thing.
But things get exponentially worse when you consider all of these graphs are interactive. Clicking on sections of the screen above shows yet more information or, worse still, takes me to a different page with different graphs.
Death by Distractions
Whenever I look at Google Analytics, I find myself clicking around aimlessly, drifting on a sea of distractions. I end up seeing everything and understanding nothing.
I can understand how you could argue that’s not really the tool’s fault. After all, how can it help me when I don’t know what I want?
But I think there’s something else going on here. Google Analytics is trying to be all things to all people. And when you’ve not got a single clear use case for your tool, you need to provide for all possibilities.
In Contrast…
Check out the project status screen from the excellent time tracking web application Harvest:
No interactivity. Only one chart. I don’t go away from this screen confused. Nor do I end up clicking around for hours. That’s because the screen is well designed and focussed on a specific task.
The Bottom Line
If graphs are to inform rather than distract, they should target very specific tasks and should be built into a focussed tool. A bit of good old fashioned restraint couldn’t hurt either.
Comments
4 Responses to “Graphs that Distract”
Leave a Comment
I would like to disagree with you on this one.
The graph shown up there visualises very well the information one would need to check and adjust when optimising a website. So right off the bat, there really is nothing I’d want left out in terms of data. (Bounce rate for instance tells you a lot about how many people either land on your site by mistake or don’t find what they’re looking for; both need to be tackled in order to build a useful website – the example of 54% is pretty bad, for instance and in dire need of fixing)
Plus, chartjunk as coined by Tufte actually refers to the amount of design elements in a particular chart that do not display relevant data. And while I think Tufte is overdoing it in some situations, I think we would agree that Google’s graphs per se are very, very clean and uncluttered.
I’ve obviously not communicated my point well enough.
I wasn’t giving Google Analytics as an example of chartjunk. I agree – the graphs are very clean with no extraneous visual elements.
I’m not talking about examples of chartjunk, either, but something I think is analogous to chartjunk. I’ll reedit the article, since I say in the intro
Looking back, that’s not really my message here. It’s more that the number of graphs on one page is distracting, not the appearance of each graph.
The more features they add to GA, the more harder it is to use. Everything can be customized, which is overwhelming. You can drag panes around in the dashboard, enable beta features, reports, filters, an API… it starts to feel like an operating system unto itself. I agree with your assessment that it wants to be all things to all people.
As for more focused – I think you’re talking about people that just want to know how things are going. “You’re doing good” or “you can improve your landing page”. But then the tool needs to understand your market and what the competition doing. Google has all this data and could tell you amazing things, but they wont.
The just make the tool do more and leave it in your hands. You can hire an expert, read a bunch of stuff and figure it out. Or, find another tool that tries to help the user understand the data in a context you can understand. I think that’s the problem with GA at the end of the day. I’ve noticed quite a few alternative tools that have more focus are popping up now.
And who can actually claim to know if a bounce rate is good or bad if they don’t even know what the site is about? If it’s a page that says “go away” then 54% is pretty good.
Great comment, Jeremy. You make some compelling arguments for something more intelligent. This is the way I’ve been thinking for a while now.
Despite all our advances in technology that people keep rabbiting on about, we’re still at the stage of manually having to tell the computer every last detail of the task you’re trying to achieve. The thing I’m most annoyed about (I think this is going to have to be a blog post actually) is that almost no-one is working on this stuff!
It seems that social networking is *still* the new shiny thing that startups want to pursue. It’s really infuriating when there are all these hard but rewarding problems that are waiting to be solved and the vast majority of the incredible talent out there is trying to get more followers for some social networking site for demographic Y. There must be so many excellent coders who are working in companies who are pushing out unimaginative but well executed social networking tools that have got X million of funding and will never make any money, and if they’re lucky, be bought by Google or Nokia then die a sudden death.
I want a world where I *guide* my computer towards what I want with a light hand, not perform 13 step processes. A computer should be able to guess common task pathways. OK, there will always be exceptions, but it should assume the road most travelled or at least allow you to choose the situation you’re in – as you say, choose a scenario, then adjust the interface to suit that.
As you say, GA is the perfect example of the tool saying “Here’s the info. Figure it out yourself.” to which my response is always “Er, no. You’re the computer. You should be helping me here. You figure it out!”
Hopefully what I’m doing will, in a tiny tiny way, bring some educated computer guesswork to bear on a problem that plagues thousands of freelancers and make their day just a little nicer.