The last 5%, simple tricks to make your Dashboards POP!

After accuracy, the most important part of a Dashboard is its' ability to clearly answer the primary questions of the audience. Deciding which charts to use and how to organize them is a very difficult task. The Analyst has to weigh a multitude of variables, such as

The Audience: Their questions, Data literacy, Position, Responsibilities, etc
The Questions: What is the dashboard's purpose, what questions is it trying to answer, etc
The Data: What is the context of the data, what is it limitations, what is its' granularity
And a whole lot more....

And based on those variable, make an informed decision on which charts to use, how to organize them on a Dashboard, and how to share them in the most effective and efficient way possible. And this is just the Functional part of the job, this doesn't even include the Technical skills that are required to create the Datasources, or the charts themselves.

Simple put...Being a Data Analyst is hard.

I consider all these things above, the first 95%. This percentage may be different from analyst to analyst, but for me, the first 95% of my job is all about the Data, the requirements and the "Chart Choosing".

I won't even attempt to act like I can speak on how one can learn this first 95%, I haven't even come close to mastering it myself. The only advice I could give is practice till you can't see straight, study and learn to accept failure.

So with that said, the Aim of this post is to try and help others with the last 5%. (I'm no expert at this either, but hope these tips will help!).

What is the last 5%? Well of course it's how make the dashboard clean, beautiful and engaging!

Nothing I am going to describe below is "new". Many analysts and artists before me have mastered the skill of Design, and many of them have written about it extensively. I have been influenced by the work of so many, and will try to include links to all of them at the end of the post. All of the techniques I'm sharing have come from others, this post is simply some of the techniques I've learned to love.

So let's get started!

I'm going to start of with a pretty simple dashboard that uses some standard techniques. For purposes of this post, let's assume the design fulfills it's functional purpose, we are just now starting the last 5%.

So overall, it's a decent dashboard, we get good insight, there is some breakdowns, nice KPI indicators, etc, but this dashboard isn't complete. There is definitely cleanup needed.

First: Let's give the Dashboard some structure.

When I was designing the dashboard, I was making decisions on how I wanted the Dashboard to be structured. I placed charts in certain positions for a reason. But in my view, it is kind of hard to see that. My mind's eye has to try and decipher the organization based on the white space. That's more work my users are having to do. I don't my users to do any more than they have to.

We can begin by first giving the Dashboard a border. This will help encapsulate it, and make it feel like a single project. I can do this by going to my Layout Pane, and finding the out most container on the Dashboard (With the exception of the Tiled Object).

Next I will right click, and press Format Container, and add a Simple grey border.

Perfect, now my dashboard has a nice border around it, making it great on Server, and presentation mode.

Now, one might be thinking, ok so let's just add a border to all of our containers, that will make it organized, right. Well the problem with using Borders everywhere, is it has the tendency to make your Dashboard feel free "blocky". It can make the dashboard feel very rigid, and not modern. Example below.

See what I mean? Instead of creating borders, we instead want to use Lines to organize things. But aren't border lines? Yes, but we don't want to fully encapsulate every object, instead we want to use simple lines to subtly break them up. At the current moment, there is no "Line Object" in Tableau, so we have two options (Both which I picked up from Jeffrey Shaffer). One is to create a Line Image, and use the Image Dashboard Object, and the Other is to use a Text Object. I will use the Latter here.

To create our lines, we first need to drag a Text Object into the Position we want to separate our objects.  

At this point we will get a very strange result because Tableau is going to predefine the amount of space this object needs. So we are going to see a result like this.

The next thing we need to do is give the Text Object a Border. We can right click on it in the Layout Pane or using the Down Carrot. Adding a Border is the same as we did to the Dashboard above.

Now here is where the trick comes in. Right click on the Text Object, and choose "Edit [Height/Width]", and set the Object to 1px.

This will shrink the Object to either 1px Height/Width depending on the Direction of the Container it is in. Placing a Text Object in a Vertical Container will Alter its' Height; Horizontal Container will alter its' Width.

Once you have learned the Technique, repeat it in all the areas you want to Organize you Dashboard. I will skip ahead and show what the result of this is. Note* I like to add Lines in-between my Banner KPIs as it helps to give each one more individuality.

Hey, this is starting to come along! But I don't like the way the Title and Footer of the Dashboard bleed in, I want to make them stand away from the actual viz. So I'm just going to give them a Shade of Color, so they act like a natural divider, and give the dashboard a little more structure. While I'm at it, I'm going to increase the Size of the Title as well, just so it is very clear to read for the user.

Much better! Even at this point, we haven't done anything to the viz in terms of Charts, or position, we have simple just added a little bit of design to help clean things up, and make it a little more modern. That said, there are somethings I think we can do to make this PoP a little better. 

I really like those KPIs at the top, but I don't think they stand out as much as they should. They feel suppressed, and I really want my audience to see those immediately. So I am going to pump up the Size of the Font until I get it to a Size I feel works for the Dashboard.

Now that's a little bit clearer to read! For this dashboard I increased the Font size to 28, which I really think helped bring those numbers out. However, as I am looking at it, I can't help but say to myself..."Does my Audience really need to see this down to the Dollar?". That top banner is really meant to be a high level summary. I think I could just show this in terms of Thousands, and get rid of some of the unnecessary detail. So I am going to go to each Chart, and change the Formatting of the numbers to be in Thousands, and I change the percentage to just show a single Decimal. This will eliminate the excess information, while giving us a cleaner look.

I'm also going to add some Custom Number formatting on my MoM Diff, adding a Unicode up/Down Arrow for extra emphasis.

"$"#,##0,.0K▲;-"$"#,##0,.0K▼

I think that is much cleaner, and it makes the MoM Difference standout more. We are doing pretty good at this point, but the Ratio between the Text size of Chart Titles, and the Dashboard itself seem to be a little off. Those titles are very large, and a bit distracting, so I'm going to decrease them down to 12.

My OCD is feeling better after every iteration! But now I want to do some more work on the Charts. I personally don't like Axis Headers and Titles. I think the spacing of them isn't great, and can be excessive when we are using Labels. I generally like to remove the Header titles from all places they are necessary for the visualization.

Next, I want to clean up some of the Grid Lines. They can be useful to aid the Users eye, but when they are too harsh, they can end up being a distraction. For Horizontal Bar Charts, I tend to remove the the Vertical Grid lines, and on Vertical Bar Charts, I will make the Grid Lines dashed, and a light shade of grey, allowing them to be there for reference, but not heavily present.

Like I said, this is a lot of very simple techniques, but they can dramatically change the engagement factor of the Dashboard. Next thing I want to do is clean up the Sales Trend Chart. I like the use of the Bar, but don't this the Separation in the bars is working here. I'm going to change the Bar Size to fixed to give it a slightly cleaner look.

The last one is really a personal preference, but it is something I really like. Here is how it looks on the Dashboards.

I really like the way this is coming out, but I don't like the labels on the Bar Chart to the Right. They are slightly hard to read, and don't give the Bar Chart enough space. So I am going to move the Labels to be Left Justified, and increase the font on it. While I'm at it, I am going to lighten the Color of Blue used on the Spark Line and Sales Trend, while making the Current Month Bar Chart Grey.

Now this dashboard may not be perfect, but it certainly has a much cleaner and engaging look that when we started. Everything I did was purely aesthetic, I didn't change the chart types, or drastically more the layouts. I just cleaned things up a bit, I used the last 5% to try and make the dashboard Pop. This style may not be for everyone, and you might have techniques that you prefer, but I hope this gave some ideas. We should always aim for efficiency, accuracy and effectiveness, but that doesn't mean we have to give up on engagement. Adding subtle hints of a design can really change how your dashboard is perceived. Putting in the last bit of effort at the end can give a dashboard a much more professional and polished look.

Here is the Before and After one more time.

I hope this post helps, and gives you some ideas for your next dashboard. Not all of these techniques are for everyone. If you have some different ideas, or disagree with some of the decisions I made, please share them below. I would love to learn more from others experience, and hopefully build on this in the future.

Inspirations:
Adam Crahen: @AdamCrahen
http://thedataduo.com/
Jeffery Shaffer: @HighVizAbility
http://www.dataplusscience.com/
Bidget Cogley: @WindsCogley
http://www.tableaufit.com/
Steve Wexler: @VizBixWiz
http://www.datarevelations.com/
Kelly Martin: @VizCandy
http://www.vizcandy.ca/
Andy Kriebel: @VizWizBI
http://www.vizwiz.com/
Andy Cotgreave: @acotgreave
http://gravyanecdote.com/