As data collection and analysis increases, the demand to make sense of it all via visualization increases as well. Data, you might recall from previous posts, is nothing more than numbers. Given the vast majority of human beings are hardwired to process information visually rather than numerically, the phrase “a picture is worth a thousand words” is made refreshingly clear. Let’s go back to the rings on the Apple Watch. Three concentric circles layered above one another, each with a unique coded color representing “move,” “exercise,” and “stand.” Beautiful simplicity, it’s all one needs to know the status of her daily fitness goal. Behind those rings, however, lies a litany of data compiled from onboard sensors measuring the wearer’s movement, heart rate, and GPS location with some help from a connected iPhone.
When it comes to digital marketing, the data set becomes exponentially larger. It’s not at all unusual for a single data file to include millions of individuals, each with identifiable personality traits; purchasing and consumption habits; professional, social and cultural characteristics; income levels; ethnicities; geographic locations; and more. To illustrate the depth of the data abyss, consider this: DX Marketing has a database of 115 million addressable households in the US that we are able to combine with over 700 million offline and online behavioral profiles and more than 45,000 data segments. This is the very definition of Big Data and, as these numbers suggest, it can be tricky to understand all the information that it presents. Data visualization allows marketers to view Big Data in a way that’s easy to comprehend, so they can quickly solve problems, gain new understanding, and ultimately enhance the return on marketing investment.
Let’s look at an example of data visualization in action with this sample data from RawGraphs.
This snapshot of the data shows different types of music storage mediums, including the media, year, and market share. Although this is a small set of data with only 496 data segments included, it would nevertheless be difficult for most to take away more than one or two simple facts from a quick glance of the data alone. This specific data set is a time series, meaning the data is listed in a set of data points over a certain period of time. Specifically, it is quantitative, discrete data. This means the data contains numbers that cannot assume just any random value. Rather, the value of the numbers is important. Knowing basic facts like these can guide us to knowing which data visualization tools will best represent the information. For instance, the most effective ways to represent time series visualizations are Gantt Charts, Area Graphs, Bump Charts, Horizon Graphs, and Stream Graphs.
For our example, here I used an Area Graph:
From this example, we can quickly see that CDs have had the largest market share over a long period of time. Let’s say I wanted to see a bit more detail with my data, I could show all of the data points with a scatter plot:
From this, we are able to see that in 2010, CDs had the largest market share, however, in 1990 cassettes and CDs had the same market share. Electronics retailers would find this information extremely valuable to help determine the right mix of product offerings.
It is also important to present data without a bias. It is easy to display data in a way that can show a certain outcome if the proper rules for data visualization are not followed. For example, one rule is to always show bar charts starting at 0 on the y-axis. From our music data example, let’s compare the market share of CDs vs cassettes over all years with an improper y-axis starting at around 600:
Looking at this biased chart, it can appear that cassettes had very little market share compared to CDs. However, if we properly start the y-axis at zero, it becomes clear that while CDs have enjoyed the largest market share, cassettes once had a sizable market share, too. With the ratios of informational data now properly aligned according to basic rules of data visualization, the data is presented accurately.
Here is an example of D3js:
Word Cloud of words from http://dxmarketing.com/intelligence
For a more Interactive example, here is Tableau Public:
If Big Data is leaving you feeling more confused than confident, perhaps you need more reliable data visualization to help you see through the numbers and improve your marketing ROI.