Basic Rules in Creating Data Visualization

This is the continuation of the transcript of DM Radio’s program titled “What You See Is What You ‘Get’ – How Data Visualization Conveys Insight,” recorded in September, 2012.

Eric Kavanagh:  Wayne, I guess I’ll bring you back in. What are some other basic rules or tenets that you’ve seen espoused or used effectively with respect to data visualization? Anything that comes to just the top of your head?

Wayne Eckerson:  Yeah.  Well, in my last book, I had a whole chapter on it and emphasized a few key principles.

One, make every pixel count. Don't add decoration for decoration sake. Two, understand that you know the data has a story and your job is to bring that -- to tell that story and highlight that story.

So the way you do that is understand what it is trying to communicate. Highlight that using various graphical techniques of which there are dozens and dozens. Many of which are very subtle but very effective because of the way our eyes work and how we visualize things. And then de-emphasize the rest.

Another key is three clicks to any data. A big problem with dashboards is that people think they are just flat displays of information with maybe one level drilling. But the best dashboards are layered delivery systems that bring to the top the most important metrics and filter those by roles.

So usually just bringing up the dashboard allows the viewer to see at a glance exactly what's right, what's wrong, or what they need to work on in the next hour or day.  But if they need to get at any other information to assist in the network or other work, three clicks and they have got it.

That's not easy to do.  Flat dashboards are a lot easier to do.  You also have to balance density and sparseness.  Sparse graphics are easy to absorb. So just in a dashboard environment you can absorb the key information at a glance.
Read how InetSoft was rated #3 for implementation in G2 Crowd's user survey-based index.

But, also, sparse dashboards may not have a lot of information, which may force you to click around if you are hunting for details.  Dense dashboards have a lot of information but may be difficult for a newcomer to find a way around and understand what's important to view and what's not.

Also, you need to have set some standards for visualization techniques. This also makes it easier for users to glance at and understand what the data is trying to say.

So, for instance, you may associate a certain type of data -- say employee satisfaction data -- with a certain type of chart -- say spider chart.  So when a user sees a spider chart and a dashboard to report, they automatically know, “Oh, that's employee satisfaction data.”

This accelerates comprehension. So those are just a few techniques that we recommend when designing effective visualizations.

Eric Kavanagh:  Those are really good rules of the road. I guess maybe, Mark, I’ll turn it over to you. What are some good rules of the row that you’ve come across somewhere to what Wayne is talking about there?

Mark Flaherty:  Well, one of the challenges that I think a lot of people who are supposed to be making dashboard support in order to consume run across is that the look and feel of the visualization does actually matter.

The challenge, I think, is because it's not always the case that the maker personally knows the data, or knows the business of that as a good design sense. But I think that it's something that shouldn't be overlooked.

Probably the best device is for the person who’s building dashboards for others to receive feedback from the users. This way you know if it looks too cluttered. Take that feedback and actually try to make your information access to product look good as well.

view demo icon
View a 2-minute demonstration of InetSoft's easy, agile, and robust BI software.