Smarter ideas worth writing about.

Data Visualization Best Practices Part Two: Mistakes to Avoid

Following the Cardinal rules for visualizing data detailed in the first part of this blog series should have you on track for creating great data visualizations. However, be aware that there are still pitfalls developers need to dodge. Listed below are a few common development mistakes and ways to avoid making them: 

  • Using the wrong chart. Developers often use one chart type where a different one would be more appropriate. The most hotly contested chart type is the pie chart. While the pie chart has its purpose, it is definitely misused and overused. The pie chart should only be used to show parts of a whole—think percentage data. If users need to compare quantitative data from several categories, use a bar chart instead. If users want to compare values over time, a line chart is the best option.

  • Inconsistency between visualizations. Consistency is especially important when compiling multiple visualizations into a report or dashboard. If “Number of Bikes Sold” is displayed as a blue bar in a bar chart, make sure that it is blue in any other visualization that utilizes color. If a specific amount is shown by thousands on one visual, do not place it next to a visual that displays the same amount in the hundreds. If all of the visuals in a report display the current year’s data, don’t throw in a visual that shows the previous year’s data without proper context. All three of these scenarios depict a situation where a poor design decision could result in users having a distorted perception of data. Keeping data representations and context consistent throughout a report or dashboard enables users to easily process information and gain truthful understanding from it. 

  • Displaying axis values in a counterintuitive manner. Always develop using scales and axis values that are precise and give an accurate picture of data. It is generally best to start an axis of numeric values at zero. Both Power BI and Tableau default every chart to have axes start at zero as a best practice. If you were to change this, you could end up distorting the chart. For example, starting an axis at three and then incrementing by one will make a value of five appear as though it is nearly double four, which is inaccurate. Additionally, be sure to increment an axis regularly. Don’t increment by one and then start incrementing by 50 later on. Much like the four and five example, an irregularly incremented axis results in a skewed depiction of data.

  • Displaying numbers in a format that is hard to digest. Don’t go beyond four digits when displaying numbers, and of those four digits, limit decimal places to two. For example, 6.2 million is much easier for the human brain to comprehend than 6,200,000. 

  • Ignoring how data is sorted. Pay attention to how charts and tables are sorted. Sort highest to lowest to emphasize the largest values. If one category of data is more important to users than another, put it first. 

  • Choosing a color scheme that not everyone can see. If you aren’t colorblind, this probably isn’t even on your radar. However, colorblindness affects approximately 1 in 12 men and 1 in 200 women, so there is a good chance that someone who will be using your data visualization is colorblind. At its most basic definition, colorblindness causes an inability to distinguish between two colors. To make a visualization easier for a colorblind person to read, instead of only using color to call out different sections of a visual, vary color intensity by making sections darker or lighter, and use different symbols where appropriate. Keeping this in mind will also keep you from overusing colors, resulting in a visualization or dashboard that looks overcrowded and busy. 

If you avoid these potential errors and adhere to the Cardinal rules of data visualization, you will consistently deliver effective data visualizations, reports, and dashboards to users. Remember, the best data visualization is one that can be interpreted quickly, is intuitive, and supports actionable analysis.  


About The Author

Data Solutions Consultant

Lindsay is a Data Solutions Consultant at the Columbus branch of Cardinal Solutions. She has experience in all aspects of the Business Intelligence life cycle, from data modeling and ETL to report and dashboard creation.