How to Win with Data Visualization

The choices for data visualization tools are vast. Their ease of use allows most anyone to at least dabble very quickly in adhoc analysis. The end results are impressive.

For instance, look at the interaction rate of TackLocal rich media ads for the first 6 months of 2015. A nice histogram showing campaign interaction rates peaking around 20%.  But this result only came after much data scrubbing and process analysis. The actual visualization piece is easy. But that is exactly my point. Had the visualization been done without the data cleansing and analysis the peak result was 12%, a considerable and misleading difference. 

Proper data analysis starts with understanding the data first. What is the process that led to the creation of the data? Are there cases when the data is recorded improperly or differently than you might expect? Are there outliers that can skew the data significantly?

Data Visualization has potential to bring insights into many otherwise difficult to comprehend sets of numbers. But just like Excel, it has the power to magnify errors as well.

So how do you win with data visualization? You ensure that your data source is well known. If it comes from a 3rd party, you need to understand how it was collected and if it was massaged where there may have been continuity gaps.  Are the units the same over long periods of time? If you are in a fast moving startup, did the measurement techniques change or the system architecture of the platform recording the data change in an impactful way?  Without that knowledge you are like an auditor trying to assess financial results recorded without accounting standards.

Data visualization tools are democratizing data analysis, which is a good thing. Just make certain that in exercising your new found power that you employ rigor in understanding your data sources. Happy analyzing !