People often think of data analysis as a left-brain activity – data living as numbers in orderly black and white columns. But that is totally wrong! The concept of data visualization, or creating visual representations of data for easy consumption, is a constantly growing, billion-dollar industry. With the help of increasing interest in business intelligence software, the visual appeal of data presentation plays a larger role than ever before. So, how can data become designed instead of simply presented?
Why consider visual aspects of your data?
Data visualizations revolve around preattentive attributes. These are things that humans understand intuitively about any visual object, such as which bar in a bar chart is longest or shortest in length. These attributes also affect our understanding of color – green is usually regarded as good while red indicates something negative or unwanted. Look at the chart below showing number of orders categorized by weekday:
Does any insight immediately become clear? Probably not. The use of color in this example does not enhance understanding of the data’s meaning. Now, compare the first chart to the second one below:
Instantly it becomes obvious that the number of orders on weekends is noticeably lower than on weekdays. This is the effect of color on our human minds using preattentive attributes.
Mastering the design of data visualizations requires understanding these preattentive attributes and using them to your advantage. Organizations such as Storytelling with Data offer valuable resources for learning more about this topic.
Key Principles to Consider in Design
By considering the principles listed in this post while creating data visualizations, the design process will not only be more straightforward, but the outcome will almost certainly provide an upgraded user experience.
In dashboard design, less is more. Often, people creating data visualizations include redundant labelling and headers to make their insights clear when they are actually doing the exact opposite! When too much additional information is included in a visualization, even if it is repetitive, the user spends precious time sorting through this information for what is important.
Another issue that can arise with over labelling is incorrect or misspelled information. Not only is this confusing for users, but it can cause more damage in the end than not labelling at all. If two pieces of information in the visual contradict each other, the user my choose the meaning that makes more sense to them, ignoring all contradicting information even if this information is valuable. In addition, the visualization itself loses credibility.
By understanding what the user naturally gleans from a visual without any instructions, you can eliminate excess clutter from your projects. Because our brains are good at differentiating properties such as size, shape, and color, a visual aspect that is highlighted by using a different color or shape probably does not also need a distinguishing label.
Interactive features in most popular business intelligence software can also help to eliminate clutter with interactive options such as showing details of data points when the user hovers over these points with their cursor, eliminating the need to label every point on the visual itself.
Colors Have Meanings
As noted above, colors can have a huge effect on the user experience of a data visualization. Because of the weight they carry, colors should be used very intentionally within any visualization. Too much color can be overwhelming to the user, while too few or similar colors will make it difficult to distinguish any important insights.
Colors should not be used to convey more than one meaning. For example, in a visualization showing profits and sales, a single color should not be used to highlight negative profits and positive sales metrics. Our brains naturally assume that objects with the same color are associated. This means that using the same color to convey something negative and something positive at the same time will lead to confusion. A good way around this issue is to use a red to green or “stoplight” color scale. Red shades denote negative outcomes while green shades denote positive ones. An orange to blue color pattern is a great alternative when users may have visual impairments such as color blindness.
Regardless of whether a metric is judged as good or bad, avoid using one color gradient as an indicator for two different metrics. In the below example, the same color gradient is used to represent both profit and product quantity.
The color gradient can trick the user into believing that the two darkest colors indicate the same amount when in reality they are not closely related. In the chart above, it looks like similar profits and order quantity occur on Tuesday, but the numbers are offset by nearly fifty-thousand units.
Natural Eye Movements
In the English language, text is read from left to right, top to bottom. The same is true for people who speak English and other related languages and data visualizations. Our eyes naturally look at the top left corner of a visual before gradually moving down and to the right. To take advantage of this natural tendency, the most important information should be located at the top left corner of a visualization to receive the most attention.
Building a visualization to flow in a way that feels natural to a user is a great strategy to ensure the most critical information is not being overlooked.
10 Seconds of Understanding
As a general rule, a user should understand the purpose of a data visualization within 10 seconds of viewing it for the first time. If more time than 10 seconds is required, the visualization needs to be reassessed. Some edits that may help quicken user understanding are removing redundant labels and information, using color, shapes, or size properties of a chart to point out important information, and adding a clear set of instructions near the visualization title.
Steps to Consistently Delivering Designed Data
By using the tips above, your data visualizations are almost guaranteed to be more design focused and user friendly. Where do you go from here?
Like any skill, data design will take practice. The more visualizations you create and the more diverse use cases you attempt will strengthen your design skills. Opportunities to practice data design are everywhere. Makeover Monday is a great way to get involved and interact with data you otherwise would not have.
One of the best ways to grow your personal data design is to submit your visualizations for feedback. This can be through an anonymous survey, setting up designated feedback meetings, or asking people you know what they think about your work.
Creating beautiful and engaging data visualizations should not be something you learn to do only once. That would lead to implementing design in the same way forever. Design trends change, user needs fluctuate, and there is always room to learn more and push your creation ability to the next level.
Identify and Use Resources
Several resources for further learning have been included in this post. Taking the time to engage with these resources and identifying resources on your own is the key to mastering data design.
Bear Cognition Offerings
At Bear Cognition, we integrate data design into every application we build. Creating visually appealing and customized applications is what has set us apart from the beginning. If you are interested in seeing what we can do, don’t hesitate to contact us today!