6 Tips for Creating Effective Data Visualizations (with Examples)
How do you know if your data visualizations have the intended impact or make your point as effectively as possible? Are your insights easier to grasp as a result of your visualization, or is there so much clutter that your point gets lost? In this article we’ll share 6 tips to help you create more effective data visualizations with clear examples of common mistakes. By implementing these best practices, you strengthen the message of your visualization and create a stronger call for change or action based on the data.
6 Tips for Creating Effective Data Visualizations:
- Data visualizations should have a clear purpose and audience.
- Choose the right type of viz or chart for your data.
- Use text and labels to clarify, not clutter.
- Use color to highlight important information or to differentiate or compare.
- Avoid misleading visualizations.
- Keep your visualizations simple. Less is more.
Tip #1 - Data visualizations should have a clear purpose and audience.
The point of data visualization is to communicate insight from data more quickly to the viewer and to make trends more apparent. All too often visualizations miss the mark on communicating insight because they aren’t created for a clear purpose and audience.
Whether the visualization is for yourself or for an internal or external audience, narrow the purpose for the visualization and understand the audience that it is going to impact. This should guide both your data exploration and your data presentation. A clear purpose will help you know where to begin when selecting different data sets to explore, deciding how to compare different data categories, or creating the final visualization and deciding what final touches to add.
A clear purpose ensures that you don’t wander aimlessly through your data. That purpose and clarification on the primary “question” you’re trying to answer will help you determine what secondary questions need to be answered as well. As you answer the secondary questions, you will have a journey of discovery that will help you formulate your conclusions with the data. A clear purpose will help to clarify those key factors and data elements that will be needed for creating visualizations that most effectively answer your primary questions.
If you feel like you yourself are getting lost in your data, it’s probably because you need to refine your purpose first and then work outward from that purpose to find more information that directly relates to it. Your data visualization will be reflective of your own data journey. The more clearly you hone in on the purpose of your data journey, the better you will be able to present it.
Both when exploring the data and creating your final data visualizations, keep your audience in mind. What do they need to know? What is important to them? What other questions would they want answers to that you could answer through your visualization or through an additional view? Considering your audience increases the chance that your visualization will communicate actionable insight. How does your visualization answer your audience’s need?
The visualization on the right makes it easier to compare sales by category or by region by representing all amounts visually (not numerically) and grouping them by region in one chart. Test this out by comparing furniture sales in the Central and West regions in each of the charts and noticing how long it takes to gain the same insight.
Tip #2 - Choose the right type of viz or chart for your data.
The type of chart you use will either contribute to or detract from your data story. There may be multiple types of charts that could be effective; that’s why you need to be very clear on your purpose and audience for your data visualization, because those both can tip the balance of which chart would be best.
Match the data literacy level of your audience. Your audience’s data literacy level will greatly impact the type of data visualization you should create, and if you want to get your point across clearly and truly pave the way for action you need to cater it to the people who will need to act. With an audience that is extremely data literate, a more complex chart like a chord diagram may be effective and tell your audience so much with just one visualization. But if your audience doesn’t know how to read that type of chart, your point is lost and you might have been better off to just place the numbers side by side and let them compare.
The best strategy is to choose a chart that you know everyone will understand. In most cases, data visualization isn’t about making the most intricate, impressive chart you can; it’s about getting the job done and making an impact for the organization. That is best achieved by catering your visualization to the audience who needs to take action. (We do suggest doing company-wide exercises and competitions around data to increase your employees’ overall data literacy and build community around data, but if you’re trying to motivate action you should meet your audience where they’re at with their understanding of visualizations and work on growing data literacy later.)
Certain types of charts are hard to read accurately. Consider that certain charts are naturally less effective at communicating insight accurately. We don’t suggest using pie charts, as the human brain can’t perceive the area of angled objects as quickly as it would compare two rectangles, as in a bar graph. If pie chart portions are too small, viewers won’t be able to compare the difference. Additionally, charts like tree maps or bubble charts that use area instead of height can be challenging to read for accuracy. If the viewer will need to make more accurate, concise comparisons between different data sets, we suggest using a simple bar graph or histogram.
Form always follows function when it comes to choosing the right chart. Consider what is the most important takeaway that you need to convey? Exact values? Comparisons? Trends? Let that guide what chart you choose.
A bar chart allows the reader to more quickly and accurately compare each category, focusing on comparison over knowing exact amounts.
Tip #3 - Use text and labels to clarify, not clutter.
Text can either be used to your advantage, or it can clutter and detract from your point. Text on data visualizations may come in the form of labels, a brief summary paragraph, a title, legend, etc. On all occasions text is there to enhance the interpretation of the visualization, not take attention away from the data.
Label for clarity, not as a rule. When labeling your data visualization, too many labels or too few labels can interfere. Step back from the visualization and ask - is this visually easy to read based on how the labels are laid out? Are there too many labels? Look at it as if you’re seeing it for the first time and consider what information you have presented most strongly. What label jumps out to you first when you read it?
Keep the title simple and clear. Does the title clearly make your point? Or does the title cause more confusion? Titles don’t need to be catchy or distracting. They should make the point of the visualization immediately clear.
Don’t use a legend unless you have to. If you use a legend, the labels will be far from the actual data, and it will take your audience longer to read the visualization. If you have the space, put your labels right next to the individual data point they relate to.
Label precise values only if the exact amount matters. If it’s necessary to label the visualization with precise values, do. But if it’s a matter of comparison where the audience doesn’t need to know an exact number, then you can leave the exact amount out and they could read the axis if needed. For example, a visualization might just need to convey who had the highest sales, not an exact sales volume of that salesman. As a result, you can avoid labeling with an exact amount and cut down on visual clutter. (Always label the axis; there is just no need to label the exact amount near the data point to draw attention to it.)
On the other hand, if your biggest takeaway for your audience is to share Sales Person A’s exact sales volume, then label with an exact amount above their data point for quick reading. Always consider your intended takeaway and let that guide your use (or removal) of labels.
All text and labels should be legible. Ask yourself, “Is the label too small? Am I using a font that is hard to read? Are they so close together that it’s confusing for the reader?” These seem like simple considerations, but they are easy to miss when you’re deep in the data and quickly creating a visualization.
Avoid the temptation to clutter up the visualization with text. If you find yourself needing to explain more with text, consider if your data is saying enough. After asking yourself that, if you still need to explain some with text, be sure to highlight the most important point for your reader within that text. The goal is clarity. Use a short and sweet callout sentence or paragraph, and consider highlighting the most important takeaway within that in a different color so that the viewer notices it first. If you include text or other annotations such as arrows, consider using them to guide the reader's eyes through the viz and toward the key insights.
We love Geckoboard’s advice that “Adding the right text that’s brief and relevant helps people use their brainpower to understand the data rather than figure out the chart.” Get straight to the point by adding text and pave the way for the reader to quickly move onto interpreting the data.
Ask a colleague for their outside perspective. When the visualization is finished, consider taking it to a colleague and ask them, “What do you get from this? What is the first thing you notice? Is anything hard to understand? Is anything hard to read? What is your biggest takeaway?” Don’t tell them what the takeaway should be until you hear their answers. Then you’ll know how you might still need to refine your use of text and labeling.
Removing the labels for each state declutters the map and causes the reader to focus first on the states with greatest and least sales. Because state shapes are commonly recognizable, labeling each isn’t the priority and could distract from the data.
Tip #4 - Use color to highlight important information or to differentiate or compare.
Color can make data visualizations more interesting and engaging. But, we have definitely seen data visualizations where color is overused, leads to misinterpretation, or just adds confusion rather than clarification. Color should always clarify and draw attention to the purpose of your data visualization. If color doesn’t add meaning or enhance interpretation, you probably don’t need it.
Here are a few considerations to keep in mind in regard to color:
Color can make your point more apparent. The strongest way to use color in your visualization is to highlight the data that the viewer should pay attention to most. When your brain perceives an extreme color difference, it assigns meaning to that difference and pays attention to it more. Use this to your advantage to help the viewer read your visualization efficiently and understand your conclusion.
Use colors to contrast different data sets. For example, if you’re comparing sales for 2020 and 2019, use one color for 2019 consistently and one color for 2020 rather than labeling them with text. Readers will recognize the color differences more quickly than the differences between text labels.
Use colors to show outliers or anomalies. If there are anomalies and outliers in the data and your goal is to point those out, highlight only those with color. If you suddenly notice a dip in sales in February every single year for the last few years, highlight those differences and look at other data to find if there are any other changes that occur regularly in February that you were unaware of.
Colors should be intuitive. Use colors that match your subject. For positive sales values, use black; for negative use red. For temperatures use red or blue. Consider how different colors make people feel and if that is the message you would like to convey with your visualization. If the visualization doesn’t need to use color representatively, we suggest sticking to your company’s brand colors for consistency.
Resist the temptation to add more color for fun. Resist the temptation to add color for the sake of adding interest to your visualization. The point of your visualization and the data should steal the show, not the color palette. The viewer may try to assign meaning to color even if it is just decorative, so only use color if it will enhance or simplify interpretation.
Only use color gradients if the data is actually increasing or diminishing in intensity. It’s tempting to use color gradients to add interest, but gradients are not useful unless they add value to the interpretation of your data visualization.
By using only one color, you draw attention to the most important information first and quickly present the main takeaway of your visualization. The example on the right gives the color more meaning and purpose, but in the ineffective example the color is purely used as embellishment.
Tip #5 - Avoid misleading visualizations.
We assume that your intent in creating a visualization is not to mislead your audience. All too often though we see people change their visualizations in ways that don’t allow for an honest presentation or interpretation of the data. Here are a few common mistakes to watch out for that lead to incorrect interpretation of your data:
Always start with 0 on your axes. Most people don’t intend to mislead by changing the labeling of their axes, but oftentimes people delete a portion of their axes and don’t start their axes at 0 in an effort to save space. When reading charts, most people assume that the labeling starts at 0, so ensure correct interpretation by never changing the axis starting point.
Never change the aspect ratio of your visualization. If you complete your visualization and use it as an image in a presentation, never stretch the image disproportionately- whether to fit the size of your presentation or to make a trend line or comparison appear more drastic. Doing so creates a possible inaccurate interpretation.
Hiding important data. Sometimes in order to make a point seem stronger it is tempting to leave important data out of the visualization or to hide unfortunate data that is still very pertinent to the purpose of your visualization. Your goal though should be to present all applicable data honestly. Don’t try to mask or sugarcoat data by surrounding it by other information or more positive information, even if the data isn’t what you would hope for. Being honest with your data will lead to real improvement in your organization. If an area of your organization is struggling or if the data doesn’t support what you want to say, don’t change what you’re trying to say to match up with the data. Don’t be afraid to share the tough information.
Too much information can confuse. If you present too much data at once, people won’t know what to focus on, and they might choose the wrong data to focus on. It’s your responsibility to guide viewers through your data visualization and communicate the data clearly and accurately. They likely have not done as much research into the data as you have, so you need to lead them to make clear, accurate conclusions with your data visualization.
Only group elements that actually relate to each other. According to Gestalt principles of design, the viewer will naturally relate items that are close in proximity. Make sure that no items could be visually related that don’t truly affect or relate to each other.
Don’t use colors that mislead. We mentioned this in our tip about color, but check your visualization to make sure that the colors intuitively make sense and wouldn’t cause accidental misinterpretation. Color use should either implicate meaning accurately or, if inferences couldn’t be made based on the color, use your brand colors.
Know that trust in data is extremely important in your organization. When data is trustworthy, there can be true change and it creates a positive impact for your company. By avoiding these misleading visualization techniques, you share information that is strong enough to motivate action without the need for enhancement or embellishment.
The first visualization gives a false impression that shipping times have dropped drastically, but it only appears this way because the axis does not start at 0 and the graph is stretched vertically to make the decline look more dramatic. Once adjusted, you see that average shipping times have only improved minimally in the past 4 years.
Tip #6 - Keep your visualizations simple. Less is more.
French writer Antoine de Saint-Exupery wasn’t a data scientist, but he did understand the effectiveness of clarity and simplicity when he said, “Perfection is achieved not when there is nothing more to add, but when there is nothing left to take away.” In an age with so much information available, you have to be selective about what you present. Anything that doesn’t reinforce the point of the data should be removed from your visualization. If it doesn’t add meaning or apply to the point of the visualization, feel free to remove it.
Remove information that doesn’t have meaning to your audience. Does it have meaning to your audience? Is it something they would like to know or have a question about? If they would have a question about it, by all means include it in your visualization. If not, there is no need to include it.
Don’t clutter with extra design elements. You don’t need to add additional design elements to make your visualization or your point more interesting. Trust that the point made with the data is strong enough. It will take your audience longer to read the visualization if they have to visually process unnecessary design elements.
3D elements don’t help. 3D elements look more interesting but are harder to interpret. Consider this - can you compare the area of two spheres more quickly than you can compare the area of two circles? You might know that one is bigger or smaller, but it will be difficult to tell by how much. If you’re creating a bar graph visualization, resist the temptation to make the bars 3D. Your audience will have to take more time to read a 3D graph visually and compare accurately.
Unbalanced visuals are hard to read. Take time to refine your point visually. Take a step back from your visualization. Is the most important part of the visualization the first thing that jumps out at you when you look at it? Can you quickly tell what the point is? Remove anything that adds visual clutter that does not contribute to the conversation or interpretation. Additionally, make sure there is enough white space between elements. Cluttered visualizations are harder to read. If needed, make text smaller or increase the space between certain elements to make your visualization less crowded.
Notice how cluttered and difficult to read the visualization on the left is. The reader is tempted to jump back and forth between both axes and the data to read it, and the bar graph does not quickly convey that the point is to compare sales over time for two different categories. The line graph on the right quickly communicates the purpose of the visualization and makes it simpler to compare sales over time and for each month.
Data visualization has an immense power to persuade honestly- to enhance an argument based on the data and show the honest state of any area of your company. When you’re looking at the data, you aren’t subject to interpretation or opinion, and that’s the strongest, most effective argument you can make for your organization. There’s no reason to debate if the answers are in the data.
take a deeper look at creating more powerful data visualizations
read our complete guide to data visualization tools for business owners: