Olga Pinchuk also contributed to this article.
We collect data at a mind-boggling pace. In fact, as companies, we’re hoarding it. But what good is data if it can’t speak to us? Fortunately, data complexity can be broken down through design and visualization – the charts, graphs and plots that show trends, outliers and opportunities.
How to convert complex data into easily consumable visualization is one of the most critical aspects of any analytics projects. After all, data is most valuable when translated into consumable and actionable information that can be understood by the end user. Too often, though, data visualization is an afterthought. Our point of view is that visualization is so critical to analytic success that it should be considered from the very beginning.
Visualization tells a story
Design is a tool that tells the story behind the data. A good storyteller will be able to provide the right amount of detail, leaving out unnecessary data ‘noise’ and help readers form the right picture of what’s going on. When a designer’s role is described as simply “making it look pretty”, there is clearly a misconception of the role design plays. Design tells the story of the data. Just as a bad storyteller may be difficult to follow, a bad design may not properly display key insights.
We tend to assume that data analytics is about math and technology, but it’s also about intuitive and intelligent visualization. A page of numbers by itself is useless. Data is only valuable if translated into intelligence that the end user can easily consume and understand—data per se, is useless. Yet people don’t give visualization the right emphasis. Instead, it is often treated as an afterthought, the last piece of the puzzle. However, if you think about it earlier in the process, it will inform the way you go about the entire project and you save time (and headaches) down the road.
Early involvement makes a difference
Design should be part of the execution plan throughout the project. We’ve found that early involvement of a visualization expert saves time, money, and frustration. For example, a client might have a broad, vague vision of how the data should be visualized. Let’s say, an intuitive dashboard that shows the performance of multiple business units of a large organization based on underlying metrics. For simplicity sake, let’s picture every business unit represented as a smiley or frowning face. Without early involvement of a true design expert paired with the data experts, this unrefined vision may lead to representing result-sets (smiley faces) that, by their very nature, can be dirty, imprecise and full of data noise. Critical design-thinking should be applied. Should a smaller business unit be represented with the same sized smiley face? Will a 5% revenue shortfall in one business be represented with the same type of frowning face as another business that just suffered a supply chain disruption that will have an effect on output a quarter from now? Once you start applying critical, sophisticated thinking by a design, business and data expert, you will soon realize how tricky the ultimate choices will be. There is no silver bullet. Only a tight, ongoing collaboration between these camps will find the right balance. Hence, we recommend bringing in a visualization expert early on to iteratively focus the vision, ensuring the necessary data becomes available and more precise to accomplish the vision. The result, an overall shorter, cheaper, more focused and intensive project.
Having a designer in on initial meetings, when clients are thinking about what questions they want to be answered by data analytics, and thinking about who needs the information, is foundational for data visualization. By contrast, involving a designer at the end of a project, means they have requirements but not a true understanding of the context. This leads to a lot of back and forth because the designer isn’t aware of everything the project team discussed with the client at the beginning.
Context. Context. Context.
Context is important for effective data visualization. More than a list of who and what, context is a deeper understanding of the user roles and user goals. Context helps a designer understand how to tell the story, including what kinds of data will be useful and how much detail to include. For example, if leadership needs to mitigate supplier risk, then the designer may choose to show high-risk suppliers first. If the audience is a data analyst looking at the same challenge, they might want to see supplier trends over time.
User context and needs are constantly evolving. An analytics system must be nimble enough to provide each user with the information they need when they need it—not more, not less. Irrelevant information is noise and may distract from critical insights. A designer, therefore, plays a pivotal role in communicating information with context, resulting in insights.
The role of time
Time is also a key factor. As context changes, so does the value of a particular data set. In many cases, users need near real-time data to be able to act upon it. Static graphs may accurately display past events but may be irrelevant after just a few days or weeks. Designers think about how time will affect your data and plan to provide it in the right form, at the right time.
We talk about data as being consumable, or understandable, in visualization. But there’s another dimension to data too: it should also be actionable. How actionable it is, will depend on how timely it is. The older the data is, the less likely the user will be able to act upon the information, no matter how consumable it is.
Pretty vs. useful
Clients often ask us to ‘make it pretty’ when they are thinking about design. But pretty isn’t necessarily useful. Of course, attractive visuals are nicer to look at, but if you can’t make sense of the data, making it attractive or colorful won’t help. The role of visualization goes much further. Making it pretty is the easy part and is often an end-stage. The design considerations you have to make early on, however, are about what data will be useful for whom, and how it should be displayed to be consumable and actionable.
Our philosophy about data visualization can be summed up by Albert Einstein’s famous quote: “If you can’t explain it simply, you don’t understand it well enough.”
Simplified vs real data
The quality of the end result is also contingent on one key aspect: the quality of data used for testing. Only data that reflects the complexity, depth, and richness of the data that will ultimately be used, will produce high fidelity designs. It doesn’t have to be the actual data; historical data with the same qualities will suffice. If the data set provided is simplified and doesn’t reflect the complexity and use case of real data, you’ll only see what’s missing once you run real data. Corrections after project completion tend to be expensive. There is no quick shortcut. You need ‘real-like’ data early on.
To get the most out of your analytics investments, we recommend involving a designer early and often to keep the project focused on who, what, when, how and why. Think of your designer as your end user’s champion, continually asking, “How will I communicate this?” Be realistic about data visualization and let the data guide the plot of the story and how it will be told. Provide the team with appropriate data sets, they’ll need it to ensure the mockups they’ll provide reflect the deliverable. And finally, commit to clarity. If data visualization can’t make the data simple to consume and act upon, then there is no value in it. Start over.
Data is your most valuable asset. Ensure that you can understand and use it by prioritizing visualization and adding designers to your data science teams and analysts.