Data Visualization Foundation – Communicating with Charts & Graphs

Data Visualization as a Tool of Communication

Despite the number of articles and books on data visualization, many people are still intimidated by the process of creating and formatting charts. Why can’t we have a precise formula that works? There is only a limited number of chart types, especially in the business context, after all. That is because data visualization, more than choosing a chart, is the art of communication, a rather intricate human process. Our first problem is that we don’t know the questions and answers we are looking for, not the right chart types. Secondly, most of the articles on choosing the correct chart use the bottom-up approach, which suggests chart type based on the properties of data. What is missing is the viewer’s perspective: what information he is looking for from the visualization. That context will determine how we design the message and how the user interprets what he sees.

To master the art of data visualization, first let’s revisit its purpose:

The goal of a simple chart is to communicate some information to the viewer. So, what is information? Here is an example. Look at the table below with some data points and make as many conclusions as possible:

The information we get from this data is

•   Andy ate the highest number of apples, while Dylan ate the least

•    Andy ate half of the groups’ total count of apples

•   The group consumed a total of 18 apples

•   Andy ate twice as many apples as Bob

Information is what we learn from observing our environment, or, in this scenario, data, and reasoning. Based on the same data, viewers can generate different kinds of information and insights depending on the context of their mind (which is shaped by knowledge and experience). There can be much information to be interpreted from our example, how can we determine what is relevant and what isn’t? It all depends on why we seek out that information in the first place. If that data comes from an apple eating contest, we need the first conclusion. If we need to know how much each person is paying for the cost of apples, the second information is helpful.

Who wins the Apple Eating contest?

How are we splitting the bill?

 This simple example illustrates two part of the communication process – the intention (why) and the message or information delivered to the receiver (what). When designing data visualization, first and foremost, it is essential to understand the reader’s purpose and the type of information he is inquiring. This knowledge will help us craft the quantitative message and then encode it in an appropriate chart type.

The Advantage of Data Visualization – Why We Need Charts to Tell the Story

The second aspect of data visualization is that it efficiently delivers the message. That is because it takes advantage of the nature of our brain, as mentioned by Jon Medina in his book “Brain Rule”:

Visual processing doesn’t just assist in the perception of our world. It dominates the perception of our world.

Sight is our most sensitive sense and our eyes are built to help us scan the environment for survival. The sophisticated mechanism of our eyes collects data rapidly and automatically, then sends them to the brain for processing. This collaboration happens unconsciously and saves us mental efforts. Additionally, we love charts because they convert the relationship among numerous data points into shapes and lines, which are more consumable for our brain.

We pay lots of attention to orientation. We pay lots of attention to size. And we pay special attention if the object is in motion.

-John Medina, Brain Rules 

Because the objective of data visualization is clear and efficient communication, visual elements in charts and graphs need to focus on bringing out a comprehendible message and eliminating noises.

The Context of Data Visualization: Reporting & Discovery

In the BI and analytics world, there are two ways to approach data visualization: discovery and reporting. Each scenario requires a different approach to communicate efficiently with charts.

Reporting

When it comes to reporting, the viewers already know what goal they have in mind and what questions they need to answer. These reports often follow specific templates that cater to business customs so that users can regularly gain insights as quickly as possible (Daily, weekly, or monthly) To a user, a report or dashboard provides the information he needs to perform a task or make a decision. To an organization, the goal of reporting is to establish a consistent and universal BI language, hence providing everyone with the knowledge of where the business is now and where it is heading. In this scenario, data visualization needs to be

•   Simple so a broad audience can comprehend that information

•   Consistent in UI and UX, especially verbiage and labeling, across all reports and dashboard

•   Instructive, providing clear definition and clarification as needed

In a reporting context, we use the top-down approach by first understanding users’ questions before creating a visualization.

Discovery

In the discovery process, while the goal is defined, the work-flow is ambiguous. The user does not necessarily have a specific question in mind. By interacting with the charts, the user generates different queries on the fly. In this scenario, we need to design a self-service experience that let users slice and dice data with different visualizations in an intuitive and human-oriented environment.

An example of a good self-service experience that provides the tool to users from their perspective is SAP Analytics Cloud (SAC).

Instead of just giving them the chart types such as lines, column, and pie, SAC suggests the type of quantitative relationship that can be described by that chart type - comparison, trend or correlation. These labels are more meaningful to users because the ultimate question they have is not “what chart should I use?” but "what do I want to know about these metrics?". While this can be considered a bottom-up approach because the chart type depends on data, the experience is designed top-down from the user’s perspective. The languages used in the labels is more relatable to our analytical reasoning.

In a data discovery process, data visualizations need to be

•   Appropriate for the nature of data (number of values, unit or scaling)

•   Relatable to user’s thought process, inspiring users to formulate questions on their own

•   Dynamic and flexible, catering to different possible scenarios

To summarize, charts and graphs are a useful tool to communicate information from data. However, to communicate effectively, we first need to understand our as well as viewers’ goal and objectives. Only then we can formulate the messages to end-users and deliver them with an appropriate chart type. When designing a reporting dashboard or providing a self-service experience for the user, our goal is to help users either answer questions they already have or generate questions most intuitively and efficiently.