Secret of Effective Data Visualization

For last couple of years I have been extensively working on dashboard design & data visualization. I have experienced how hard it is to decide the right type of chart or graphs for my data, many a times. Since the same data can be represented with more than 1 type of chart, that does not mean any type of visualization will communicate the information as efficiently as the users are looking for.

Let me walk you through some examples.

The above example (Figure: 1A) shows the number of product sold by sales representatives over a period of 6 months. The line chart here almost works all right by showing the change in sale over the month for each representative. It also shows the overall trend, whether the sale is going up or down. However if the users are interested in looking at the total number of product sold by each representative in each month & comparing their performance, this may not be the right choice. You can see here how difficult it is to detect the difference in the number of product sold by representatives in a particular month. In this case using a bar chart (Figure: 1B) would be the right choice since the bar chart emphasize individual values & enable easy comparison between adjacent values.

The chart above (Figure: 2A) shows net income from operation (NIFO) % variance to budget over a period of time. When you read this chart to understand the variance you need to do the extra work of calculating the gap between a column & the grid representing 100% value. If understanding the deviations from the target or budget is the need, plus minus chart (Figure: 2B) works the best. This chart shows the variance over time or by category.

In above example (Figure: 3A) the chart shows the number of open position at a company by job type. The whole bar represent the total number of open position & each segment represents their proportion within the group e.g. number of open position for skilled nurse. When you look at this chart carefully you will find how hard & time consuming it is to understand the difference between the number of open position for Speech Therapist  & Occupational Therapist. Also reading the value of each segment is somewhat difficult except the one that appears at the bottom of each bar. A regular bar chart (Figure: 3B) works much better in this case. Stacked Bar Charts are the right choice only when you must display multiple instances of a whole  and its parts - with emphasis primarily on the whole.

So the point here is that we should not always rely on the brief manuals that explain when to use which chart or on the aesthetic appeal when visualizing our data. The secret of visualizing data which works, is truly about understanding how those data would be used in real life from users point of view.   

Thank you for your time.