Unless you need to look up exact values or do some calculations, visualizations are almost invariably better than tables of numbers. If you are not already visualizing your data, you are missing important insights.

Research has shown that data visualization can lead to both better and more efficient interpretation and analysis of data compared to numbers in a table or in text. My PhD research looked at the impact on data literacy, and I found that visualizing data can make you more data literate. That means that it can enable you to understand and analyze a more complex set of data than you would otherwise be able to. It is like moving up a reading level. And to use reading as an analogy, imagine being able to read and understand a complex academic article, rather than a highly simplified media brief of the same article.

Visualization can do the same for your ability to understand your data. But if the visualizations are incorrect, inappropriate or misleading, then you might actually go down a level in your understanding. Unfortunately, such visualizations are very common. So visualization is not inherently beneficial!

What types of data/information is best suited for a visual presentation?

All quantitative data is suited for visual presentation, and certain types of qualitative information as well. Sometimes qualitative information can be quantified. For example, there is a lot of innovative work on interpreting and intelligently quantifying text – analyzing the complexity, sentiment, etc.

Another approach is to apply labels – reducing inherently qualitative information to a limited set of categories. For example, opinions can be captured through a Likert scale (e.g. strongly disagree / disagree / etc.). This kind of information can then be quantified. Geospatial categorical data can be visualized very well through a map – see for example this map of political systems around the world: https://en.wikipedia.org/wiki/List_of_countries_by_system_of_government#Map

Steps to follow when creating a chart or graph

First think – what is the purpose of the visualization? Too often they seem to be added as a matter of habit, without much thought as to what the visualization contributes. For example, My presentation is about sales, so let me add a time series graph showing the most recent sales data. But what is its purpose?

Identify the most important pattern or trend shown in the visualization. This becomes the key message that needs to be emphasized in the visualization, and then communicated. For example, a change in trend after a new sales strategy was implemented, which can be emphasized through a label, arrow, change in line color, etc. And if a visualization does not contribute anything useful (e.g. clearly show a meaningful trend in the data), it is better to just leave it out.

Another recommendation is to explore and dig deeper. For a particular data set you can visualize it in an infinite number of ways. Trying a few different approaches will increase the chances of finding an interesting trend. When it comes to maps, mapping at different administrative levels (e.g. state-level and county-level) can reveal completely different trends and insights.

Once an interesting trend is found, make that central to the story you tell around that visualization. A visualization is incomplete without a story around it – whether verbal or written, or even as a thought process in your mind (if the visualization is not going to be shared). Unfortunately, the story telling aspect cannot be automated … yet.

What makes a visualization stand out?

A good visualization requires a good story, and a good story requires a few elements. The first is that it has to be newsworthy, of interest to most readers and preferably using recent data. The second is that there must be a clear message. Something to take away from the story that could be expressed in one or two sentences. And third is the element of surprise, which makes you think – gosh I didn’t know that!

A recent favorite which does all three is a New York Times map showing life expectancy of 40-year-olds in the USA with household incomes below $28,000. The story has a simple but powerful message – which is in fact the title – “The Rich Live Longer Everywhere. For the Poor, Geography Matters.”

What types of organizations would benefit from automated data visualization?

In my view there are three kinds of organizations for which automated creation of maps and charts, using tools such as StatPlanet, is a must. 

• The first are large companies – such as multinationals – which need to monitor and analyze their performance across different geographic levels (national and sub-national).
• The second are marketing and consultancy companies which provide data analysis services.
• The third are organizations which deal with large amounts of public data, such as government agencies, media organizations, universities, UN agencies and INGOs.

That said, automated data visualization is useful for companies of all sizes. Anyone familiar with creating charts using Excel will know that they often do not come out right. Fixing and customizing a chart can be a tedious process, let alone if you need to create dozens, or hundreds … or thousands. StatPlanet generates interactive maps and visualizations on the fly for any of the indicators in your spreadsheets or databases. No tweaking is needed – it comes out right because it understands your data (such as by determining the data structure and distinguishing the time and spatial dimensions). It then creates the interactive charts and maps around that understanding. But of course you can customize the results to your heart’s content if needed.

A final piece of advice

Take visualization seriously: a good visualization is worth a thousand words as the saying goes, but a bad one is worth less than nothing – it will do more harm than good.