Creating Effective Data Visualizations
Data Visualization can be a highly effective way of communicating essential insights to your target audience. Regardless of whether you are designing data visualizations for corporate, government, non-profit, or general audiences, the same general guidelines can be applied. Below I share some of my views on how to think about and ultimately create an effective visualization.
Keep a Narrow Focus
One of the great temptations in this era of big data is to provide an overload of information to the end-user. As data sets have become increasingly detailed and more comfortable to access, it becomes effortless to fall into the trap of attempting to display all sorts of related or even unrelated data elements in a visualization. For those of us who are immersed in data, this is an especially tempting prospect.
To wind up with a useful visualization, we need to step back and understand what it is we are trying to convey to the end-user. If we cannot quickly define the intended output, our data visualization is likely to end up unfocused and unused. In the case where there are multiple possible outputs, it is wiser to design a data visualization for each distinct output.
Provide the Right Level of Detail
Another of the great temptations in the age of big data is to show the deepest level of detail within the data. So just as we discussed keeping a narrow focus (limited breadth), we also need to provide just enough detail (limited depth) for the end-user. A prime example would be displaying a national map at the zip or postal code level when county-level data would provide more insight for the end-user. A general rule of thumb is to restrict data display to the minimum practical level. If your end-user can’t act on data at a given level, don’t show it in your visualization.
A related consideration lies in understanding who your end-user is; this is especially key in large organizations. The CEO or VP of Marketing will be interested in national or regional patterns that are critical to their business. A regional manager, on the other hand, will need to see information at a slightly lower level so that she can explain patterns within her region. Finally, a field salesperson will require information at the customer level. Therefore, we need to either design separate visualizations for each of these audiences or provide an interactive framework that starts at the highest level and allows for detailed exploration at lower levels.
Design Using Best Practices
After defining the appropriate breadth and depth of our visualization, we can move on to the visual delivery of the information. This is where it becomes critical to design a visualization using best practices. Think of the prior two stages as providing the foundation and structure of a house; now, we need to design and decorate the home to make it appealing to the end-user. To do this effectively, we need to manage the design elements of size, space, position, and color.
Size in this context refers to how we draw attention to specific elements within a data visualization. For example, I make it a standard practice to use very large font size to highlight primary KPIs on a dashboard, while using a slightly smaller size for secondary elements. This practice provides a quick visual signal to the user about the hierarchy of the information - abundant equal essential.
One of the great temptations in this era of big data is to provide an overload of information to the end-user
Space refers to the non-data portions of a data visualization. Appropriate use of space between elements in a data visualization allows the user to focus on each element without being distracted by an adjacent piece of information. A full visualization is the enemy of clarity, so we must resist the temptation to deliver too much information via too many data elements in favor of a clean, focused, well-spaced display.
Position refers to the general visual layout of a data visualization. For Western cultures, this means putting the most critical elements at the top left of the visualization and then working across and down the page. Secondary elements should be on the page below the essential high-level information. Far too many visualizations waste space displaying non-essential data in the most valuable real estate on the screen.
Finally, the color should be used judiciously. My practice is to use as few distinct colors as possible to effectively display the information in your visualization. The human visual system is typically limited to being able to distinguish between just 6 to 8 colors simultaneously adequately, so if your visualization uses more colors, the user will be forced to spend extra time attempting to decode the information.
If we design our data visualizations using best practices and provide the appropriate levels of focus and detail, our end users will be able to quickly and effectively understand their business. This should be the goal each time we create data visualization.