Dashboards can be put into two broad categories- Informative and Analytic.
- Informative- Tells a user information
- Analytic- Used as a platform for further analysis
This article will provide users with SiSense’s best practices for creating Informative dashboards.
Making clean, readable dashboards can help a user understand information quickly, and get more value from a dashboard. This is not a complete guide, and I invite people to add comments with their own methodologies for creating informative dashboards.
The Short Version
- Don't put more than 7 widgets in a dashboard
- Tell a Story with the Data
- Minimize Clutter
- Use a Consistent Color Palette
Don't Put More than 7 Widgets in a Dashboard
This principle comes from a psychological study on the limits of short term memory. Miller’s Law argues that the number of objects an average human can hold in working memory is 7 ± 2. This is taken from George Miller’s "The Magical Number Seven, Plus or Minus Two: Some Limits on Our Capacity for Processing Information”, one of the most cited papers in psychology.
Using this information, we can infer that when a user opens a new dashboard, they will only be able to understand approximately 7 widgets at a time. This is the basis for the design principle of only creating seven widgets in a single dashboard. The goal of using SiSense is to make sense of data, and overloading a user with too much information goes against this goal.
Additionally, making a dashboard with fewer widgets will make it load quicker, which improves the user experience.
Extending this Principle
We can extend this rule of seven to make a seamless user experience. Every decision a user makes should be as simple as possible, and there are 4 decisions a user makes
- What folder is the dashboard in? Limit the number of folders that you share with a user.
- Which dashboard has the right information? Limit the number of dashboards in a folder.
- What widget am I interested in? Limit the number of widgets in a dashboard.
- What does the information in the widget signify? If the widget is a pivot table, try to have less than 7 columns. Most other visualizations simplify the data that a user is looking at.
Experienced User vs New User
This principle is built around the idea of providing a new user with information, which would require the utilization of working memory (aka short term memory). If a user accesses one dashboard many times, they’ll be able to use long term memory to understand the information in the widgets, which enables them to process more than 7 pieces of information.
Tell a Story with the Data
A goal of a dashboard should be telling a factual story in an effective manner. Journalism is a field that specializes in this, so we can extend concepts from this. One in particular is the inverted pyramid. (http://en.wikipedia.org/wiki/Inverted_pyramid) Essentially, you want to start a story with a high level summary (the headline and lead), and include more details and context further in the article.
This can be utilized in a dashboard. In the English language, users read from left to right, top to bottom. In a dashboard, the title should be the “headline”, and the widget in the upper left section should provide high level summary data. As a user reads the dashboard, the widgets should provide more detailed information, or data that provides context for the other widgets.
The following report shows a very basic dashboard that is ordered from highest level and most important, to most granular.
The simpler the visualization, the easier it is to understand. One way to do this is to minimize the amount of “ink” used to produce the image. If any non-background pixel isn’t conveying data, re-think the purpose of the pixel.
The following gif shows a bar chart that is edited from the default style to make the chart as clear as possible
- Eliminated the grid lines, since the ordering of the bars made comparisons between the bars simple and intuitive
- Eliminated the legend at the bottom, because the widget title already communicated what was being measured, and all the bars are green, so the color identifier was useless.
- Eliminated labels on the Y-axis, and added labels to the bars. This eliminated 6 numbers and 6 tick marks on the y axis, while adding 9 numeric labels. This reduced the amount of ink on the chart, while making comparisons between bars easier with exact labels.
Use a Consistent Color Palette
Sisense recommends using a consistent color palette of up to 5 colors. A consistent palette will make the dashboards look cleaner, and will limit the amount of information (different colors) that a user needs to process.
If you don’t currently have a color palette for your company, the following tool can help you select one: http://paletton.com/
After selecting the palette, click the Color Tables tab in the bottom right and click the Color Swatches tab on the right. This will list the hexadecimal code for each color in the palette, which you can use as the input for the custom color. The #Color field in the bottom right of this menu is where you would input a hexadecimal color value.