Top 5 data visualization best practices you need to remember
Top 5 data visualization best practices you need to remember
Data visualization can save us from drowning. The tsunami of data that floods our daily lives is on the rise. By 2025 we can expect up to 463 exabytes of data to permeate our daily existence! (Bear in mind that 1 exabyte = 1,000,000,000,000,000,000 bytes!)
Data visualization is the life-preserver we need to manage the heavy-data load, since human processing power falls slightly short of machine learning speeds! Humans do, luckily, respond to and process visual data faster – up to 60 000 times faster than text. The visual medium is natural to us. As a result, businesses can use data visualization to break down vast complex data into bite-size chunks we can understand at a glance.
And there’s potentially more to the subject, a recentForbesarticle observes: “Data analysis isn’t just about assembling, ordering and interpreting data; it’s also about educating, simplifying, clarifying, and persuading.”
This raises an important point—persuasion. A visualization needs to be informative but it should also persuade people to take a particular action or decision. The decision may pertain to everyday business or something that affects people on a global scale, like climate change awareness. The bottom line is, data visualization has the power to change people’s perception.
Creating a visualization that steers business decisions or changes the world, begins with a few basics. Here are five best practices to consider when you craft your next visualization:
1. Communicate to a specific audience
When you’re designing a data visualization, you need to know the specific audience you’ll be communicating with. Knowing this helps you to aim the data story at them and speak to their specific needs. Aim it at the highest level persona who will view the visualization and make sure their needs are covered and the common roadblocks they encounter.
Creating a visualization that communicates with a broad audience may result in it being ineffective and could reduce the likelihood of extracting key business insights.
Once you’ve established who your audience is, consider what decision you want them to make from the data. And the nature of their decision. Different roles or functions in a business make decisions differently. C-level executives tend to make strategic decisions. These are usually one-off type decisions. Whereas an Operations Manager has to continually make decisions based on daily operational choices.
Taking into account these decision-making elements, you can then align the data visualization accordingly. Strategic decision-making may require more complex visuals, while for operations, a binary-type interface would suit simple yes-no decisions. The time scale should also be adjusted to display a scale relevant to the decision type.
Now that we’ve got the audience nailed down, let’s look at the visual aspects involved.
2. Choose the best visual
It’s important to choose a visualization that will communicate the right message to your audience. When you’ve got this covered, your audience will quickly understand your point and extract the insights they need.
Each type of visualization works best with a particular data type. These data types could include:
Categorical data, also called qualitative data, represents characteristics like a person’s gender or age and is non numeric.
Ordinal data is a sub-category of categorical data. Ordinal data is also non-numeric but the order of the data matters like in surveys or questionnaires.
Quantitative data, also called numerical data, is measured in numbers, like length or revenue.
Let’s take a look at the visuals that pair with these data types.
Tables can display large quantities of data, categorical or quantitative. If the audience only needs to see a high-level view of the data, the table format may be overwhelming.
Bar charts measure categorical or quantitative data and are a good way to compare the quantities of different categories.
Vizlib Bar Chart
Line charts use ordinal or quantitative data. They track change over time and the relationship between two variables.
Vizlib Line Chart
Area charts measure ordinal, categorical or quantitative data. They are similar to line charts, with the area beneath the line shaded in.
Pie and donut charts demonstrate categorical data. They show the comparison between the parts of a whole. Make sure you label the individual wedges so it’s clear which pieces are larger. And don’t compare two pie charts alongside each other that have different totals!
Vizlib Pie Chart
Scatterplots reflect quantitative, ordinal or categorical data and show the values of two variables against two axes. The pattern of the results shows the relationship between the variables. You can also demonstrate the relationship by the size of the data bubbles.
Vizlib Scatter Chart
Heatmaps measure categorical and quantitative data. They show a graphical representation of data with individual values represented by a specific color.
3. Use proper design principles
It’s imperative to keep the dashboard simple and uncluttered. Much like any visual presentation, too much clutter will disperse the viewer’s attention and leave them feeling confused and frustrated. They should be able to grasp the high-level meaning of the visualization at a glance if you adhere to a few basic design principles.
Create digestible designs
Firstly, the design must be digestible – quick and easy to understand. Keep the design clean. Instead of trying to jam too much into one visualization, rather create an additional object. Another way to limit the clutter and deliver a better user experience is to include alternative dimensions and measures.
Secondly, consider adopting every good designers’ passion for white space. White space, or the space around objects, gives the eye a rest and allows your focus to fall where it should, on your visualization objects. It allows the viewer to understand what they see more easily and creates a better user experience!
Image credit: storytellingwithdata.com (showing the power of white space)
Color usage is an important design factor in visualizations. Use colors to differentiate visual elements and make your data story as clear as possible. You could combine colors and shapes for your indicators, like a green empty triangle and a red full triangle, for additional visual clarity. Perhaps try using different line styles and different colors with your line graphs. And remember to use a palette that is color-blind friendly to ensure there are no accessibility issues with your visuals. If you’d like to learn more about selecting the ideal color palette, read our guide here.
The layout of visualization elements, like icons and filters, should enhance learning. Ensure you position them in a way that reflects the correct hierarchy of information.
Placing them on the left-hand side of the dashboard is a good idea as users look to the left first for important elements like filters or navigation tiles. The further right you place objects, the less attention the viewer pays to them. The best layout of the main filters or icons is stacked vertically on the left.
A horizontal arrangement is also workable. Placing icons or filters on top allows more space for visualizations. Note that viewers will consider the first icon in the horizontal display as the most important.
Together with positioning, you could also show the importance of data by the size of the chart or text. Make the key objects larger to draw the audience’s attention.
Lastly, the design should maximize understanding and provide a great user experience that is not overwhelming. You could implement this principle by labeling data directly, rather than using a legend and including annotations on your visualization. And, only include visualizations that serve a purpose. This will cut down on distractions and allow the viewer to focus on the key message in your data.
Image credit: storytellingwithdata.com
4. Provide context for the visualization
The most effective data story-telling always contextualizes the data. Contextualizing refers to showing the bigger picture, the story that surrounds a situation. Framing the data, by including context like annotations, makes it easier to understand. When viewers grasp what they’re seeing, they can interpret it correctly. As a result, you establish their trust.
It’s vital that business leaders understand visualizations at a glance and trust the insights to make their recommendations. Once trust is established, they’ll be more likely to take the action you want them to.
You can contextualize data bycomparing your metrics to something tangible like a goal. If you can present metrics in comparison with dynamic thresholds, it’ll make the data easier to interpret.
An additional way to provide context is by including annotations. Annotations, like text labels, tooltips or icons, are visual cues for the viewer and make the visualization quicker and easier to grasp. Annotations help explain large shifts or changes in the metrics. They will also help the viewer stay focused on the visualization's main message.
Vizlib Line Chart with annotations
5. Avoid common pitfalls
Applying best practices will help you master your data visualization skills. And by avoiding common pitfalls you’ll be well on your way to even more effective data story-telling. Look out for these potential glitches when creating or collaborating on visualizations:
Avoid using too much color. Stick to the fundamental rule—keep it simple. Simplicity leads to understanding and keeps your audience on board.
Don’t try to embellish the data. Make sure it is transparent.
Ensure the data is correct! And that the parts add up to the total figure.
Label the data correctly to avoid confusion and losing your audience.
Don’t truncate a graph’s axes as it may distort the data representation and minimize the impact of your message.
Data visualizations should be simple and persuasive. Follow these best practices to shape the mind of your audience! You’ll be empowering them to take insightful action in their business or perhaps to change the world!
Here are a few additional examples of Vizlib data visualizations.