How to make your data projects stand out?
Let's get knees deep into dashboard design!
DATA VISUALIZATION
7/4/20236 min read
Just yesterday I got asked (by one of my brightest students) a very important question - how to make data projects stand out. I really love questions like this one because they're challenging and there's no one size fits it all answer to them.
I've put together a guide with steps that will guide you through the entire process of putting together the data project presentation. When we're starting out with that plan, we first need to figure out the medium we'll be using. I personally think that nowadays a dashboard is a must. I've got nothing against a good old Power Point presentation, but the truth is, dashboards are the present and the feature. I do realise many companies still rely on PP presentations and there's nothing wrong with that, but if we want to create an eye-catching representation of our data project we should definitely go for a dashboard.
You can choose any tool you're comfortable with - Tableau, Python, Power BI. It doesn't matter as the principles we'll discuss are applicable to all tools.
In order to make a dashboard stand out we have to define what standing out really means. It can have various dimensions - it could stand out with its sleekness, its original design or the unusual format you've chosen to represent the data. The one below is a good example of creativity and sleekness:
Unfortunately, the original dashboard has been deleted, I saw a screen shot here.
So, in order to arrive at the end design we need to look at things at a rudimentary level first. Your dashboard won't be eye catching (in a good way) if the basics aren't covered. In the sections below I'm going to try to break down the necessary steps of building an impressive data dashboard in a chronological way - from planning to execution.
1. Get out there with a clear story
One of the most important success factors in creating an appealing dashboard is how quickly it allows the viewer to grasp the essence of the story presented. If I have to stare and explore the dashboard for good 5 minutes until I get what it's showing me, that's a straight NO from me.
Here's a great example of what a self-explanatory dashboard looks like:
This dashboard doesn't leave one second of a doubt what exactly it will present to the viewers. It's about different characteristics of the 10 highest grossing actors of all times. There's a clear title and the ribbon with their animated caricatures (which act like dynamic buttons) supports the message that we'll be seeing different bits of information about certain people.
What to do to improve your own dashboards:
Think how you can summarize your project objective in one line
Now turn this one line into a catchy or at least a very unambiguous title
The title needs to be descriptive enough but not too long
2. Choose your elements wisely
No one likes a crammed dashboard. A crammed dashboard will do exactly the opposite of what it's supposed to do - instead of enlighten you, it will overwhelm you. How to achieve this? With a very careful selection of the elements that will go onto the dashboard.
Do you remember that title you created in the previous point? This was an important step for you to clarify to yourself what the essence of your analysis really is. Now it's time to ponder how the story you're presenting should be told. What are the key metrics that tell this story? If you're creating a research question dashboard, what are the metrics that answer this question? Don't be tempted to include a certain type of visualization just because it "looks cool". For example, maps always look cool and people tend to love them. But if a map isn't conveying a vital piece of the story you're telling, then using one is pointless.
How to apply this in your own work:
I highly suggest doing this old-school style -->
Write down the most important findings of your analysis in a list. Regardless of the type of dashboard you need to create you should be able to pin point the crucial takeaways, the backbone of your work.
For each takeaway you wrote, think of the most suitable way to present it visually. Is it a scatter plot? A bar chart? A map? Have a strategy before making the visualizations and stick with it.
Now think how you can make some of these elements interactive. Some nice interactive features are based on filters with categories. Which categories in your data could you use to add a level of interactivity?
3. Visual design - putting it all together
I personally find this part to be the hardest. Here we're not in our element anymore - analytical skills don't help when it comes to design. I'll probably never find designing the actual dashboard easy, but here are a few tips that help me keep my head above the water.
What to look for when designing the content of the dashboard:
Think alignment before anything else - how would you separate the area so that the charts are evenly distributed? Imagine a gridline and try to order the charts around the lines. If you can align the charts to each other, that's great. Have some space on both side of the dashboard. Try to make it look orderly and spacious but without too much space.
Next thing is colour: you know the rule of not more than 5 colours on a dashboard. Try to be consistent with it.
Ink to data ratio - a fancy term coined by Edward Tufte. The data-ink ratio is the ratio of elements in a visual representation conveying information to the total elements in the image. What this means for you is that you should spend your "ink" only on the elements that are really important to your story. Drop chart gridlines, unnecessary text and other fun elements.
I created an example Tableau Dashboard to illustrate some of these principles. It's far from perfect but in the video below I'll explain my thoughts when creating it and what I paid attention on, also what needs further improvement.
A more finalised version of the dashboard is here. It's still a very basic dashboard, but it's a start.
4. Getting inspired
Unless you're some kind of a design genius driven by natural talent, in which case I doubt you'll be reading this post, you'll have to see a lot of dashboards in order to develop the design mindset needed for the task. You can go the Viz of the Day Section in the Tableau site and check out different visualizations. Try to make notes of what you like - whether it's the colours, the order of the elements, the charts used. Take mental (or physical) notes and try to build this into a vision for yourself. Next time you're about to build a dashboard, go back to your notes and try to follow the principles you outlined. This will make your life a lot easier! Remember, creativity is often simply a collection of matured impressions and experience, so don't expect of yourself to be able to blast out a fantastic dashboard on your second project. It took the masters years to get there, and you'll be no exception of this rule.
Here's my personal collection of great Tableau Dashboards to serve as an inspiration:
One final thought - pick one tool and really master it. Whether it's going to be Tableau, Power BI, Python or even Excel, don't get scattered around different softwares. Creating some of the more fun and complex elements requires a solid knowledge of the tool you're using. So if you want to focus on your data visualisation skills, I'd highly recommend picking one program and getting proficient until you move onto the next one.
That's it, folks, I hope this was helpful! If you feel I've skimmed on some of the points or you'd like me to elaborate on something, drop me a line using the box on the right below!
