The current state of business intelligence tools in the marketplace makes it easier than ever to display data visually. It wasn’t always this way. Older tools like SAP BusinessObjects and Microsoft SQL Server Reporting Services allowed business users to build reports that displayed data, but it is difficult to build visually pleasing dashboards in those tools. With new data visualization tools comes a new way of thinking about how to visually communicate data to a decision-maker.
With tools like Tableau, Looker, and Sisense, analytics professionals have more capabilities than ever to tell a story with data, not just display it. However, using these new tools means introducing a new set of rules for how to communicate information appropriately. Just slapping a bunch of numbers into a text table is not a good design pattern; it’s difficult to interpret what’s important in a wall of numbers. Luckily, experts in the field have researched how humans see and understand data to produce a set of attributes that dashboard designers can use to communicate information quickly and easily. These are called “pre-attentive attributes.”
According to Stephen Few, an expert in the data visualization field, these pre-attentive attributes take advantage of a type of memory called “iconic memory.” Iconic memory is where information gets stored very briefly before the brain moves it to either short-term memory or discards it. In short, a person sees pre-attentive attributes and starts processing information before cognition even fully begins. The pre-attentive attributes may be classified according to their form:
- spatial (length, width, orientation, position, curvature, enclosure)
- color (color hue, color intensity), size and shape.
Alternatively, we could classify the attributes according to what type of differences in data they most naturally emphasize:
- differences in quantity (length, width, color intensity, size)
- differences in quality (orientation, position, curvature, enclosure, color hue, shape)
For instance, it may be most appropriate to represent the difference between sales figures with a bar chart utilizing length, or a geographic map utilizing color intensity. Examples of each of the pre-attentive attributes are in the screenshot below, again from Stephen Few.
Using these pre-attentive attributes as dashboard designers build visuals allows faster processing by users by taking advantage of the brain to communicate what is most important the fastest.
To present an example of using pre-attentive attributes to draw attention to data, here is a Tableau dashboard developed by Kansas City native Sean Miller.
The “Top & Bottom 5” chart is known as a diverging bar chart. There are multiple pre-attentive attributes meant to draw your eye as you look at the chart. The first one you likely notice is a version of the orientation pre-attentive attribute (or 2-D spatial positioning if you prefer) – some bars go left and some bars go right. Then you may notice that some bars are shorter, and some are longer (line length) or you may notice that the bars going left are purple and the bars going right are orange (hue). Lastly, you may notice that the colors closer to the vertical center are a lighter color, and the colors toward the bottom and the top are a darker color (intensity).
In one chart, Sean uses four different pre-attentive attributes to help you orient to the data story being told. Very quickly you can see that Slough County has the highest percentage of increasing mental health cases and Barnet County has the highest percentage of decreasing mental health cases. You can also quickly see that the orange bars are all larger than the purple bars, signifying that mental health cases at the extremes are increasing more than they are decreasing.
Lastly, that color scheme of purple/orange is consistent throughout the whole dashboard. The text description near the top-left introduces the color legend directly, rather than using an actual widget to display a color legend. The map uses the same gradient as the diverging bar chart. This consistent use of color is not a pre-attentive attribute, but when introducing color in your charts, it is crucial to keep colors *consistent* throughout the entire dashboard so that users do not need to keep re-orienting to a color legend or risk interpreting data incorrectly.
In just two charts using four pre-attentive attributes, Sean communicates a data story about generally increasing mental health cases in the UK but gives the user the ability to dive in and explore the data for their own data stories. It takes less than a minute for a user to orient to this dashboard and understand what it’s trying to say. There’s not a text table to be found. Using pre-attentive attributes makes it easy to build dashboards both visually pleasing and packed with insights.