This is the first of a series of five posts discussing the four pillars of successful visualizations. In this article I’ll introduce the four pillars and discuss why they’re in the order they’re in; and then in subsequent posts I’ll examine each pillar in depth and explain how to think about and use the concepts when building an effective data visualization.
A successful visualization:
> has a clear purpose and focus
> contains (only) the right content
> is structured correctly
> has useful formatting
Therefore, the pillars are, in this order:
Purposeis the specification for your entire effort. It defines what success means and how to get there, usually in the form of what knowledge you are trying to communicate. To be useful, the purpose must account for your goals as a designer, the needs and use cases of the customer of this information product, and the characteristics of the data itself.
Content is pretty straight forward. It’s the data and relationships that are represented in your visualization. In most cases, that does not mean all of the data and relationships that you have access, but rather the specific, concise subset that supports your purpose.
Structureis the physical layout of your visualization. It may be a line graph, scatterplot, entity relationship diagram, histogram, map, or any other spatial representation** of your content. The appropriate structure is informed by your purpose – the knowledge you’re trying to reveal, and by the content you’ve selected to convey this knowledge.
**By definition, if you’re visualizing knowledge, you’re placing it in space. An unformatted stream of text isn’t a visualization. A line of words is, however.
Formattingis my bucket term for all other visual treatments that go into your visualization. This includes both non-spatial visual encodings of the data (size, shape, color, texture, arrows, boxes, etc.), and all of the supporting labels, axes, grid lines, highlights, notes, etc. The formatting helps to explain the message, reveal what’s interesting, highlight noteworthy data points or areas, and generally make the whole thing easier to understand.
Without the right purpose, you’re headed in the wrong direction, shooting at the wrong target, or looking at a map to the wrong city (assuming you have a map at all). Lack of purpose means success is nearly impossible, as success itself isn’t even defined.
With the wrong content, you can’t hope to satisfy your purpose and convey the knowledge that you have decided is important. Too little content means there will be holes in the knowledge; and, too much content makes it harder to find the elements that are important. Selecting the right content depends on understanding the purpose.
Once you have determined your purpose and the right content, you’ve figured out what to visualize. Then you can begin to consider how to visualize it. That begins with selecting a structure.
The wrong structure can make it hard to reveal or recognize the patterns and relationships that matter, even when the content is correct. Specific types of relationships in the data require specific structures to reveal them, so structure depends on both purpose and content.
Formatting must work to complement what is presented by the chosen structure, to reveal the correct content, in service of purpose. Poor formatting doesn’t necessarily mean your visualization is doomed, just that at the very best it will be difficult to extract knowledge from.
Clearly, each of the last three pillars depend on all of the previous pillars, so we can represent the four as a stack of blocks, with purpose at the bottom. If any block is removed or destroyed, the blocks above are unsupported.
That’s the overview of the four pillars. The next four posts will examine each pillar, discuss ways of thinking about each and approaches to success, and provide examples.
For further discussion on this topic, download my recent whitepaper, “Choosing visual properties for effective visualizations.” In the whitepaper, I discuss the huge number of design decisions you need to make upfront before creating your visualization that will impact on the ability of the visualization to communicate knowledge accurately and efficiently. This paper addresses one key aspect of the design process: how to choose an appropriate visual property (position, shape, size, color and others) to encode the different types of data that will be presented in the visualization.
For more information:
Visit IBM’s visualization hub, IBM Many Eyes and join more than 100,000 like-mined visualization enthusiasts, academia and professionals, including additional insights from Noah Iliinsky and other IBM visualization luminaries.
Share this post: