Telmar News & Insights

Six Best Practices for Data Visualization

August 17, 2016

With all the toggles, crosstabs, coding, graphing, analyzing and all around data crunching media planners and buyers need to do, making the data “look pretty” is a separate challenge. The right data visualization tool, though, makes sharing insights and results for reviews and market updates even easier.

Telling a visual story with metrics and measurements is its own art form; data visualization software helps someone become a Van Gogh of charts and graphs.

Choosing Your Data Visualization Palate

There are four basic ways to present data:

  • Comparison
  • Composition
  • Distribution
  • Relationship

Determining which chart is the best way to present these forms of data involves advance consideration of the number of variables, data points, and the way in which the values will be displayed (e.g. over time or among groups).

The bullet may be better for one type while bubbles or bars are best for another. Or perhaps you need a stack instead of a scatterplot? Or a pie instead of a donut?

Fortunately, relationship is generally best displayed in a scatter or bubble chart while distribution tends to work well in a histogram or scatter charts. Yet the more common types, comparison, and composition lend themselves to seven or more visualizations — each.

Ultimately, there are do’s and don’ts for each one, and with so many options, finding the right chart type can be difficult.

Best Practices for Data Visualization

Let’s look then at some general observations that can be made about data visualization best practices.

1. Start simple

What are you trying to show? It’s easy to get lost in the world of data. But hone in first on what you want the visualization to convey — its purpose.

2. Consider the audience

Will the layout pass the squint test? Who will you be presenting to and where? The size of the room and of the group could both be factors. Communicating data is not just about visualizing the data, but also presenting it in a fashion that is easy for an audience member to process. What questions will the audience have? What questions are you trying to inspire?

3. Embrace logic

Don’t get so creative with your visualizations that you are counterintuitive to the expectations of the audience.  For instance, when using time in charts, set it on the horizontal axis. Time should run from left to right.

4. Streamline

Remove any excess information that does not add value. This includes extra lines, colors, text, or information.

Not convinced? Consider this reader-submitted data visualization example from Junk Charts:

data visualization

5. Avoid Color Overload

Using more than six colors in a chart can also overwhelm the viewer, as we also see in the above graphic. Similarly, you should use the same color (from light to dark) to compare the same value at different time periods. Also, check how your charts will translate if the end user prints out in grayscale.

6. Try and try again

In your effort to emphasize the most important data and orient for legibility and all the above factors, you’ll need to be attentive to detail and open to trial and error. Take the extra time needed to not just visualize the data but present it in a way that is effective and easy to understand.

 

Telmar’s Data Visualization Software portfolio and expert support team can help you extract more information from the data you already have.

DOWNLOAD OUR  MULTIBASING CASE STUDY

 

Sources
http://junkcharts.typepad.com/junk_charts/2013/01/ruining-the-cake-with-too-much-icing.html

Subscribe to Telmar Insights

We would like to keep you about Telmar and our products. You can expect to receive the following types of communication from us:

  • Industry news relating to media planning, audience targeting and segmentation, data integration, and market research
  • Information on Telmar products, services, and company updates
  • Free downloadable whitepapers, infographics, and other long-format content

If you do not wish to receive such communication email from Telmar, then please opt-out here