Guidelines for creating impactful dashboards
Published in · 4 min read · Jul 13, 2020
--
Dashboards still form a crucial part of the data science process for any company. From data analysts to highest level executives, dashboards are still used to convey key information and develop insights from data.
There are many different areas one may consider while designing dashboards. The data, its availability, the need to visualize it, the way to visualize it, consideration of perceptual and cognitive science, the list goes on and on. And you will find hundreds of guides and rulebooks for dashboard design. With so much information, it is quite usual to get blogged down by the rules and not be able to deliver something useful.
This article aims to provide a checklist of 5 points you can consider, when you are asked to design a dashboard. Even if you are a novice or highly experienced, you can use the below pointers to create dashboards that convey information correctly and effectively.
So, let’s dive into it!!
Remember, the goal of the dashboard is important. Oftentimes the discussion about a dashboard gets focused on the appearance of the dashboard. People like to see dashboards designed as per the ones they are familiar with. This leads to dashboards that fail to solve the problem at hand. The key here is to focus on what problem a dashboard solves and not how it is designed.
Another important thing is that the dashboard matches the capabilities of the users. If we can understand the metal model of the users, we can better assess the requirements for the dashboard. The dashboard should neither be highly complex nor be too simple, but rather well matched to the abilities of the users.
Important questions to ask at this step:
- What are the 5 key questions the dashboard will answer? What actions will be taken in response to them?
- What is the audience of the dashboard?
- What information items need to be present on the dashboard and why are they important?
- Which items on the dashboard are answering your primary questions?
- At what level of detail(granularity) the information needs to be expressed?
- Which comparisons between items of information can help answer the key questions?
- How often does the dashboard need to be updated?
With a clear understanding of the objectives of the dashboard and information items, make a prototype of the dashboard. This helps you understand what components you have in hand and also find out something you might be missing.
The prototype can be a simple hand drawn sketch. No need to design a working prototype.
The aim here is to define the different sections of information that will be present on the dashboard, what their importance is and how they should be arranged.
Also, ask for feedback on the prototype. You may not get the design right the Svery first time. Try to bring in new perspectives on your design.
A picture is worth a thousand words. So choosing the right picture (chart) matters a lot! Charts are important to visualize patterns, trends and gain insights otherwise difficult to get just by looking at raw data.
- Trends over time: Line chart, Area chart, Bar chart
- Comparison and Ranking: Bar chart
- Correlation: Scatter Plot, Bar-Line combination chart
- Distribution: Box Plots, Histogram
- Part to whole: Stacked bar charts
- Geographical data: Maps paired with other charts
Additionally you can also check out data-to-viz.com for a detailed guide on chart types.
Beyond choosing the right chart type, we also need to ensure that the chart is effective.
Tableau has a white paper on visual analytics best practices, where they have outline some of the examples below on making charts easier to understand: https://www.tableau.com/sites/default/files/media/whitepaper_visual-analysis-guidebook_0.pdf
Don’t overload your charts with information. Instead create divisions to help understand the chart better.
Choose the correct orientation for your chart.
Avoid too many colors in a single chart.
Organize the chart so that it can answer a question easily.
Gather feedback from your users about the chart, so that you can evaluate it better.
Make sure your dashboards answer the key questions you designed it to answer. Make sure the chart types are right and charts are easy to understand. Find out if there is a requirement for filtering or drill downs in the chart. Based on the feedback improve the design.
Making dashboards is an art as well as a science. So in order to build great dashboards you need to put in effort and time, collaborate and reiterate!
Credits: Sumit Sharma
