Einstein Analytics: Design & Interaction Principles by Rikke Hovgaard

Creating a dashboard is easy, but making it meaningful is much harder. This applies to any dashboard not just Einstein Analytics. Think about it, how many times have you looked at a dashboard you didn’t create yourself and ended up spending several minutes decoding what you were looking at? Is the pipeline for this month or the next? And are the cases closed or opened? Maybe you were successful in decoding the dashboard or maybe you closed down the dashboard without having decoded anything. And if the latter was the case you probably didn’t bother to have a second look another day. I can only say that in my line of work I have seen plenty of examples of poorly designed dashboards that is not straightforward to read. Yes, I am probably not the intended audience, nonetheless, I believe a dashboard should be readable by everybody even though the data story is not understood by all.

To avoid dashboards that are difficult to read I follow 5 simple guidelines when I create dashboards. I have picked these up while learning Einstein Analytics, but some of them are also based on knowledge and theory on communication. Though I will not make this into an academic paper, I guess 3 years communication study can come in handy in the most unlikely places. To illustrate my points I have created a dashboard using data on San Francisco Bike Share [find the data here].

1. Know your audience

In any form of communication, this is a key principle. If you know who your audience is you are better equipped to get your message across. But when it comes to dashboards, knowing your audience will also help you understand what information they are looking for and what questions they expect to get answered from the dashboard. With this information, you are one step closer to knowing what dashboards to create and what graphs to include.

Ask yourself who are you designing a dashboard for? Is it for top management? Account Executives? Support Reps? Or maybe it’s for your customers? When you know who the audience it becomes much more clear what story to tell with your data. Why would you include the individual account executive’s sales data for top management? They need an overview, not the sales and target per account executive – at least not at first glance.

2. Make your dashboard easy to understand

Make sure your dashboards are readable at one glance. Describe with text what the user is seeing and group your data in categories. In the sample dashboard, clear headings are letting the user know this is an overview dashboard with data on Bike Share Trips and Weather. Also, each illustration or widget has a headline so the user is guided to the key message of the graph. Make sure that the predefined criteria are included in the text, a typical could be planned or completed trips.

Even looking at the numbers, you want to explain if the duration is seconds, minutes or hours or is the temperature in Fahrenheit or Celsius. This can be done directly on the dashboard in form of text widgets or by modifying the measures in the XMD file. Either way, you want to make sure there is no doubt what is seen on the screen. The question I ask myself is “would my grandma understand what she is seeing?”.

3. Guide the data story

Understand what on your dashboard is most important to draw the attention to. Obviously, the graphs should and will draw attention as they are the key components in driving your story. But the text is equally important in telling the story. However, if everything on your dashboard “pop-up” it is hard for the eyes to know where to focus first, second, third etc.

You can use text size and colors to draw attention. For instance, use a bigger font size and darker font color for the main headings to instantly let your viewers know what they are looking at. Let your graphs or numbers stand out by themselves but use a lighter color for their describing text, that way you ensure not everything “pops” right away. Another good thing is to use lines and containers, they make a structure and help categorize your dashboard. In the example, you are not in doubt what is trip data and what is weather data as it is separated by two different containers. Also within each container, you know what headings and graphs belong together as you are guided by the lines.

4. Use relevant visualizations

In Einstein Analytics there are many different types of graphs to choose from and the last few releases have given us even more. Each graph has its own purpose and therefore you should consider which graph is best used to illustrate your data story. If you have 24 values in a grouping then the pie chart is probably not your best option and you are better off with a bar chart. Just as I have done with the “Popular Times of the Day graph”.  Also, this graph allows the viewer to better compare the difference from hour to hour.

5. Make your dashboards consistent

My final principle is: try to make your layout consistent. For people that are new to Einstein Analytics this is a hard one. It takes more than one try to come up with a design that works for you and your business. But try to nail down that layout and stick to it. Let’s face it we are all creatures of habit and if you keep changing the layout of the dashboard it will be hard to navigate as you constantly have to be aware. So if you always have your date filters in the top of the dashboard, don’t move them to the left. Or maybe you always want to have a “key metrics” section to the left. Try and remember when Linkedin or Facebook last changed their layout. Didn’t you get a little annoyed because you had the navigation nailed down and with the changes, you had to familiarize yourself with the new layout? Even if you didn’t think it was a big deal most users’ rely on their navigation habits, they get confused and complain about any major layout changes. So nail down your layout and stick to it.

Rikke Hovgaard is a makepositive certified Salesforce (CRM) consultant. Rikke is an expert in helping businesses get the most out of the application whether it being Sales Cloud, Service Cloud, Pardot, Analytic Cloud or a combination of those. Click here to read more of Rikke’s blogs.

Share this post: