Data visualization – the fundamentals – numbers

We live in a data-driven world, which means that the ability to create effective charts and graphs is now just as important as being able to write well.

However, while computer technology allows almost anyone to create graphics, only a few are skilled at it. A well-designed chart can ensure that the information contained within it is communicated effectively and easily to the viewer.

 In an attempt to beautify their presentation, those who are not well-versed in the art and science of information design frequently employ a variety of visual maneuvers that end up detracting from its content. They get carried away alternating clashing colors or incorporating 3D blocks in an effort to spruce up the look of their chart without giving enough consideration to the information that it is trying to communicate. This is something to watch out for.

The open secret is that content makes graphics interesting, not the other way around. Extra color layers and other enhancements should only be used where they do actual work. Otherwise, they only serve to obfuscate the information presented – not something we want!

In our previous blog post, Speaking Through Data, we discussed the various roles that information graphics play in business decisions. This second post of the series will teach you everything you need to know in order to distinguish between good and bad charts.

The fundamentals

Did you know that a single erroneous data point can invalidate the rest of the data and render the entire chart useless? Consider a color, font, or shape that misrepresents or obscures a single piece of data on a chart, and you'll understand what we mean when we say that good choices in data visualization are crucial.

The font, color, design, and depth of critical analysis displayed in a chart determine its effectiveness - or lack thereof. Is my data worthy of a graph, and have I accurately represented it? That is the question we must ask ourselves whenever we need to communicate numbers in a specific context. So, without further ado, here are the fundamentals, the backbone, the foundations of creating intelligent–and convincing–charts.


Words vs. Charts

Charts are a powerful communication tool. Within a story, numbers convey information narratively. In a chart, numbers communicate information visually, making it appear more concrete and precise and thus more easily digestible.

This is illustrated in the figure on the right: mapped-out information has a more significant impact than a series of numbers linked by words.

Let the data speak for itself

The best charts are those that are free of distractions and allow the viewer to compare and contrast the data before drawing conclusions.

Heavy gridlines and 3D renderings, for example, obscure the data and divert the viewer's attention away from the content.

A clean and crisp chart, on the other hand, allows the viewer to concentrate on the data, which is the true star of the show.

Same numbers, different stories

Embellishments are not a substitute for the proper organization and presentation of data. Filter and edit data to ensure that it is consistent with and relevant to your message.

Example: Credit cards issued by bank X in each country.

When the total number of credit cards is compared to the number of credit cards per person, two different stories emerge, and different impressions are formed – all from the same data.

Country C has the largest total market, and the chart on the left depicts the credit card market as a whole.

Country B has the highest card issuance per capita. Despite its smaller population, the chart on the right shows the success of the marketing effort in country B.

It's all in the framing

The way information is presented will influence how viewers interpret the data. For example, viewers frame information based on their expectations, and their point of reference determines what they expect to see.

Without context, it is impossible to say whether $100 is a reasonable price for stock X. If we knew more about stock X, such as its 52-week high and low values, we could answer this question.

Even if a number is given at random and without context, people will automatically create a reference point and assign meaning to it. When you provide the reference point, you control the message.

Creating reference points

A single number in a statement may not mean much to the viewer on its own. For example, the statement - Stock B is currently trading at $5 per share - does not provide much insight.

A series of numbers plotted together on a graph, on the other hand, has a more meaningful effect. When the prices of stock B are plotted over time, the graph clearly shows that the stock has lost half its value at $5.

Send the appropriate signal

We've already established that the same number can have multiple interpretations. For example, the agony of a $1,000 loss is more intense than the joy of a $1,000 gain. So you can chart a single set of numbers in a variety of ways, so choose one that best communicates the message you want to convey.

For example, consider the performance of stock X.

Plotting the actual share prices yields the following graph:

However, when the percentage change in price is plotted, the line enters the negative territory. This exacerbates the apparent drop in stock prices. The chart visually implies that the performance is poor by establishing a baseline.

Both charts provide a fair representation of the situation, but your charting choices create a framework that transmits a specific message to the viewer.

The message of your chart should always be consistent with all of the facts and evidence available. For example, a chart that excludes quarters with poor performance when plotting profit and loss, for example, would be misleading. However, presenting facts in your chart is entirely up to you.

Tell the whole truth

Do sweat the small stuff

Data is only as good as its source, which is why it is critical to obtain data from reputable and impartial sources. For example, to avoid bias and add credibility, market share data should be benchmarked against a third party.

Approach data with skepticism at all times. If something isn't quite right with a number, it's critical to determine what it is. A single erroneous data point can jeopardize the credibility of an entire chart.

Bad data + Good visualization = Bad chart

One size does not fit all

You might have guessed by now that the type of chart you choose will be influenced by the point you want to make. So make an informed decision: some chart formats are more convenient than others for communicating specific messages.

The bar chart on the left simply reports the revenues of all the companies in a specific market, whereas the doughnut chart on the right shows that company A accounts for 55% of the total market.

Be transparent

Increase credibility by presenting facts in a balanced manner. A 200-person hiring drive, for example, could represent 1% of one company's workforce and 10% of another's. Provide context for your numbers to be more transparent.

A percentage without a base number has no meaning. If you mention a 10% increase, be specific about when and where that increase occurred.

The market share for product X is shown below.

The only conclusion that can be drawn from the two pie charts is that A and B both have a 60% market share, but it is impossible to determine which has more sales without knowing the size of each market.

Round off at the end

Wait until the very end of your presentation to round off your numbers. During the analysis stage, rounding the numbers up or down can result in inaccurate final results and subsequent erroneous interpretations.

The distinction between companies A and B is lost in the following example. Meanwhile, $0.2 billion (or $200 million) is a substantial sum!

Edit at your discretion

A larger data set allows for more in-depth analysis during the research stage. In the editing phase, however, it is important that you assess whether all your extra information buries the story's main point or enhances the story and makes it more convincing.

After analyzing the data, the pie charts show that company B has a stronger market presence despite the fact that some details are lost when the smaller companies are combined—the viewer benefits from the editor's effort to highlight the underlying data.

An excerpt can tell the entire story

Extracting a few numbers from a series is permissible if the data points tell a story without leading the viewer astray.

For example, it is not deceptive to extract recent performance data when sales have increased steadily. However, it is preferable to display all eight quarters to emphasize how consistently strong performance has been.

Nevertheless, omitting the last three quarters from the chart below would be misleading. Because the poor performance data excludes previous quarters, the viewer comes to a different conclusion than if all of the facts were presented.

In our next blog post, we'll go over how to use typography in charts. Stay tuned and sign up for our newsletter to keep up with our blog posts!

Read More


Remote team collaboration: 10 effective ideas for your virtual team

Read More

How to troubleshoot data-driven web application issues: 4 effective steps

Read More