The rise of the Big Data approach to finance and treasury challenges facilitates more advanced ways of analyzing data that numbers, or simple charts, could not ever deliver.

First pioneers emerged over a decade ago with solutions like Tableau, a business intelligence software I started to work on while it was still an OEM by Hyperion. Tableau broke the 3-dimension limit with elegance and efficiency. By being able to plot 3 dimensions (X, Y, Z) with size, color and shape and potentially within a 2-dimension matrix, it was able to stretch a user’s ability to comprehend multi-dimensional analysis.

Figure 1: A multi-shape bubble chart

I also worked with Trendalizer’s creators while at Google as their solution became embedded within the Google Docs suite. It added motion to these charts to cover time related evolution.

Figure 2: Trendalyzer with time slider at the bottom

These solutions aimed at displaying more dimensions but were falling short in representing specific problems. How do you feature things such as process flows, behaviors and connections? This gap created a critical hurdle for execution. If the execution arm of the process could not read or understand, or in other terms “visualize” the outcome of an analysis, how could the actual work be carried out?

While finance and treasury are comfortable with numbers, this simply isn’t the case for everyone in operations.

The D3 open source project drove the popularity of these new charts by making them available via open source libraries which made them immediately pluggable on top of a user’s data. Suddenly, you can instantly recognize process variants where most transaction flow would converge out of large datasets. So many charting options blossomed to accommodate the need of data representation to boost decision support and adoption.

Figure 3: Chord Diagram with D3 to represent trip patterns at Uber (https://bost.ocks.org/mike/uberdata/)

Finally, purposely built applications started to deliver composite visualizations to not only represent but also demonstrate. Being able to show a correlation from a cloud of dots with a curved line would have more impact than any R square number. Showing how clusters span across a population by representing circular shapes on a cloud of points can clearly demonstrate the different populations of elements to go after.

Figure 4: Correlation of a driver to a performance level

Figure 5: Clustering of a population

As execution and support for execution remain key elements to an analytics process, understanding these options are crucial for Finance and Treasury teams. Striking the optimal balance between message completeness and understandability affects overall success.

Now, there is a point where any chart will fail to reveal the complexity of a situation. At a certain stage, visualizing the issue will have diminishing returns. This is when numbers must be resuscitated and a classic ratio or index will reveal more than an inexplicable chart.

I’d still recommend the good ole’ whiteboard discussion before making any visualization decisions—it can do wonders each time ;).

If you are attending the AFP conference in Orlando Florida next week, I will talk more about how Visualization Technologies make big data practical. As a panel, we will be talking about where you will begin on this journey based on real experiences. We will also share observations on how organizations can begin to build analytics skills from within. So, if you will be at the AFP Annual Conference in Orlando later this month, please join us.

Can’t make it to Orlando? Then let me know if this is a topic you want me to continue to write about. Please like, share or comment – it’s how I know you want to hear more.