Advancements in technology have made data more available and accessible than ever before. However, data is not useful in itself. We must have the ability to interact with this data and make meaning from it by identifying and understanding patterns and outliers. And then we need to be able to present this data in the best way possible.
Data visualization is the graphic representation of data, and it's art is in understanding the best ways to reveal complex information at a glance. Data visualization is not about knowing which graph or chart is the best to use but rather understanding the layers of detail that generate insights for the viewer. Since art concerns itself on matters of beauty, and we are drawn to beauty much like moths are drawn to flame, our visualizations cannot be devoid of aesthetic considerations.
As described by Noah Ilinksky in his book, Beautiful Visualization, for a visual to qualify as beautiful, it must be aesthetically pleasing yes, but it must also be novel, informative, and efficient. Good visualization entails that data needs to convey information clearly, excite and engage observers, drive home the intended message and remain to-the-point.
Process Of Data Visualization
1. Intended Message
The first step of data visualization is not the nitty-gritty of format or structure; it is intent.
Is essential to envisage your intent and ascertain your goal. For instance, you determine that your intent is to sell your product to a new audience. Identifying a goal will then help you articulate a clear purpose of your visual. Your goal could be to devise a new strategy to gain engagement of the new audience.
2. The Data Itself
Keeping in mind the intent of your visual, the next step is to select what data is best suited to meet your goals. Not all data is relevant. Data needs to be culled to communicate a clear message. When trying to market a product to a new audience, relevant data could include demographics, what channels your target audience use, their purchase history and loyalty.
3. Graphical and Aesthetic Construction
Graphical construction includes layouts, axes, shapes, colors, lines and typography. The main purpose of the graphical elements is to lay out the information. When using graphical elements, any element that does not guide, communicate or highlight information is extraneous. It will take away, rather than add value to your representation. Your choice of graphical construction needs to be based on clarity and efficiency.
Aspects such as color, placement, size, icons, layout, and labels are all aesthetic considerations. These choices must take into account familiarity of their audience and minimize noise.
Data visuals are used for the same purposes as speech: to inform, to persuade or to entertain. Here is a quick summary of what different graphical representation tools are best suited to:
To compare sets of values: column charts, mekko charts, bar graphs, pie charts, scatter plots, bullets and line graphs
To show composition (parts that make up a whole): pie charts, stacked bar charts, mekkos, stacked columns, area charts and waterfall charts.
To showcase distribution of data: scatter plots, mekkos, lines, columns and bars.
To analyze trends: line charts, dual-axis lines, and column charts
To establish relationships between data sets: scatter plots, bubbles, and line charts.
Once you have the data, graphical construction and aesthetic choices go hand-in-hand and must be worked on simultaneously to determine the best way to portray your data. Here are a few things to consider when deciding on how to design visualization for your data:
Using size to differentiate is best used to represent few data points with enormous variations because it promotes intuitive differentiation.
Source: WHO Report on the Global Tobacco Epidemic, 2008, page 57
As shown in the graph above, there is a vast difference between the per-capita tobacco tax revenue and the per-capita spending on tobacco control. The difference in size of the circles, or the difference in the heights of the bars immediately grabs attention and showcases the disparity between the two components. On the same note, here are two graphs portraying the same information on opinions around animal cruelty.
Source: Animal Friends Croatia
If a marketer for a cosmetic company wants to push for cruelty-free branding, Figure B would be more effective in quickly showing the vast difference in opinion of the general public.
Data model matrices are useful for database project evaluation because they expose non-intuitive data properties that are hard to uncover by simply looking at common indicators. Models and matrices can be applied to help with long-term strategic planning and to solve real-world marketing challenges. Here is an example of a matrix that is useful in evaluating investment in marketing channels.
Source: Smart Insights
Based on the information collated on the matrix, a brand can better decide where to invest. For example, a brand could consider increasing investment in platforms under the low investment, high ROI square and decreasing investment in platforms under the high volume, low ROI square.
Using color to differentiate is best used in enormous, high-resolution sets of data. Different shades of colors are used to identify patterns in large datasets. However, color may not be an effective tool for reaching larger audiences because people may be afflicted with colorblindness.
When selecting fonts for a visual, they must be legible. That is most important. It should not be difficult to read the words because they look prettier in a fancier font - this is extremely counterintuitive. Let's say you are a cigarette company looking to expand your sales globally. Here is some data that would help you decide where to increase investment based on the proportion of smokers worldwide.
Source: WHO report on the Global Tobacco Epidemic, 2008, p19
The first figure would make it extremely difficult for someone with colorblindness to be able to differentiate between the countries. The font, despite its fanciness, would deemphasize the country which needs to be highlighted. Although the colors in the second chart do not serve a special purpose, they would help differentiate between the countries and show investors that China is the largest market for cigarette consumption.
Using a visual that factors in time (where the data is shown to change over time) must be backed up by context, and any sudden changes must be explained.
Location-based visuals (on maps) hook the audience if they are familiar with the location. Furthermore, interactive visualizations that use location must be targeted towards those who are deeply familiar with the location. For instance, the graph below tracks development across Brazil through education, longevity, and income indexes.
Only a person familiar with counties across Brazil would be equipped to navigate this chart.
Categorical data are best visualized through three techniques: treemaps, parallel sets, and mosaic plots (also known as mekko charts).
The Gloucestershire County Council use of parallel sets is a great example of finding the most efficient way to present data. The council was measuring the effectiveness of its marketing campaign by carrying out massive public surveys that asked two questions:
Have you seen the posters?
Do you think the Council provides great value services?
When they tried using pie charts, four of them were needed to show the same data set: one showing opinions of people who had seen the posters, one for those who had not, one to show the total exposure of the campaign, and one showing the overall opinion of services provided.
However, parallel sets allowed for plotting the data of these questions in terms of one another on a single graph.
The convenience and conventionality that comes with using standard formats, such as bar graphs, lines, scatter plots, flow/organizational charts, make them the most commonly used means of information visualization. However, a truly good representation requires the creator to leverage the familiarity that stems from standard formats and tailor them to create unique representations tailored to meet the needs of your information.
Locobuzz is a SaaS platform that converges with technologies such as Artificial Intelligence, Machine Learning, Big Data and Analytics to provide brands with a 360-degree customer experience management suite. Locobuzz’s powerful analytics algorithms have helped seasoned brands establish a strong foothold in the digital hemisphere and transformed their customer experience journeys. Visit our website www.locobuzz.com for more information on our Customer Experience management services that are catered toward businesses like yours!