Designing for Data Visualization

IBM designers share the unique challenges and opportunities of designing for data visualization

Created by Christine Song and Alexandra Grossi

Every business has a story to tell and these days that story usually involves data. At a time when people expect software to help make sense of their data, it’s the responsibility of the designer to ensure that data presentation is meaningful, compelling, and most importantly, easy to interprete. So, how to achieve these goals? There is no magic bullet, but at IBM we have a set of design guidelines that help.

People working in modern day enterprises require more than just static charts and pretty looking visualizations; they need to go beyond by drilling deeper, probing further, and understanding the forces that influence their business.

The IBM Hybrid Cloud Design team for IBM Business Analytics in Toronto spends a lot of time performing user research about how to create data visualization products that meets users’ needs. With a product portfolio including Cognos Analytics, Watson Analytics, Watson Analytics for Social Media, and Planning Analytics, they create a wide range of tools that focus on helping their users make the most of their data. To explain how design helps deliver great data visualization tools, I’ve asked the IBM Business Analytics design team to share some of their insights and perspectives on this topic.


Data Visualization at IBM

Our clients are from various industries as well as organizations of all sizes, from large institutions to lean start-ups. But regardless of size or industry, our users all have the same goal. They have data, they have questions, and they need an analytics tools that will help them make sense of their data and turn it into useful business insights, while reducing uncertainty.

When it comes to designing the details of a data driven product, there are a few things that we keep in mind to create the best possible experience for our user.

What is the power of data visualization?

Consider this: you receive a postcard from a friend in Venice. The glossy photo contains a typical Venetian scene — a view of the Grand Canal, a gondola navigated by a man in a white shirt who appears to be singing, and stone bridges that fade into the horizon. Your friend writes about how beautiful it is and ends the note with “you simply have to see it for yourself!”

Suddenly you’re overcome with excitement and you begin trolling travel booking sites looking for cheap flights and accommodation. One postcard just isn’t enough of an experience for you. You want to go there yourself, explore the tunnels across the stone bridges, and hear the sounds of the gondoliers singing as you venture off the Grand Canal down the backstreets of Venice. A snapshot is simply not enough to satisfy your need to explore and see things for yourself. You’ve heard about St. Mark’s Square and while it may not be pictured on the postcard, you really want to see it and it can’t wait any longer.

It’s the same with data visualizations. Users are not typically satisfied with simple postcards no matter how picturesque they may be. They need an experience that is as immersive as possible, while making it easy for them to uncover deeper insights and drill deeper into their data in order to make better business decisions.

How can you experience your data?

Our users are looking for a tool that not only presents a static view of their business, but one that also enables them to interact with that data in real time. Offering data visualizations that are flexible and change with the user’s thought process allows for true exploration.

For example, in Cognos Analytics we offer an experience in the user dashboard that provides side-by-side data comparisons and methods to quickly see how these data points relate to each other and what these discoveries mean.

The first dashboard of Cognos Analytics gives a good overview of data from Bikeshare Chicago’s overall ridership. By using the sorting, filtering, and brushing features, the city manager was able to see which neighborhoods have the highest percentage of subscribers in the 39 to 55 year age range from the previous year.

These type of user goals show that data exploration isn’t the end, but rather a means to gain as much use out of the data as possible, whether it be applying it to business models to maximize profit, or to creating more accurate troubleshooting techniques. This application of data demonstrates the true power of data analysis tools.

Thinking back to our Venice analogy, a tool like Cognos Analytics allows the user to really dive-in and and get immersed in the data, rather than only being able to see it at surface level.

Guided exploration with cognitive analytics

Insights have more value when you can act upon them, especially in business. The tools we build enable users to quickly uncover interesting patterns and relationships in their data without the need for any coding. But endless exploration can sometimes lead to analysis paralysis, where the user is continuing to search for every possible data correlation. This may be fun for some data scientists, but is not something that most businesses can afford.

To overcome this, a well-designed data exploration tool not only helps users explore freely, but also navigates them towards the insights they’re really looking to make.

Watson Analytics allows easy data exploration through natural language processing, which means users can ask simple questions about their data regardless of their analytics expertise. In this day and age, as designers we need to reduce the gulf between man and machine. With the predictive capabilities of Watson Analytics, users can ask questions like “what drives sales?” and swiftly be presented with key sales analytics to help them make decisions.

In this example, Sam, an airport operations manager, asks Watson Analytics “What’s driving overall satisfaction?” in English. Watson does the number crunching, creates a predictive model and returns a number of different fields and graphics associated with levels of satisfaction for airport customers, in addition to displaying their predictive strength. The user is easily able to obtain this level of data analysis without needing a statistics degree.

Again thinking of Venice, guided exploration in Watson Analytics is like having your very own personal tour guide who not only knows the best local restaurants to eat at, but also knows what to order and how to order it in Venetian.

3. Analytics your way

When it comes to analytics, not all users needs are equal. One of the challenges of designing a data visualization tool is making it intuitive to use for anyone. To define universal experiences, we believe in designing for diversity.

When designing our Business Analytics portfolio, we have multiple personas in mind that range from a novice analyst to a power user. This means the design needs to strike a balance between reducing the learning curve while conveying powerful analytic capabilities. Our customer experience strategy is focused on providing users with the tools and resources they need to be successful, regardless of their existing expertise in data analytics.

Our research process includes many direct and indirect partnerships with our clients and users, including deep observational studies to learn about their workflow as well as how they want to use a data analytics tool. All of this hands-on research helps ensure that we design meaningful experiences, with embedded support and guidance to help users succeed.

Analytics isn’t easy and it’s not as intuitive as booking a trip to Venice. Our users have many different levels of skill, experience and understanding. They consist of new managers who want to get a better understanding of the business, to power users who want to look under the hood to find out what statistical model was used in the analysis. In Watson Analytics we offer layers in our products that progressively disclose as much or as little of the magic that is used to generate visualizations.

Watson Analytics doesn’t simply show a chart of his data, it highlights statistically significant numbers and results, saving him the trouble of doing the calculation himself (B). It also surfaces a series of insights and follow-on chart suggestions in the Discovery Panel on the right (C). A data scientist working for a business professional, can open up the Statistical Details panel to confirm and investigate more closely the models and parameters behind the results.

Design plays a key role

Designing data visualization is not just about the visuals, but why those visuals matter in the data analysis process and how they can be of actual use for the user. We work on designing for iterative data exploration, a guided experience that helps the business user get to their business answers as quickly as possible, and a flexible work flow that supports analytics experts and novices alike. Design work in this field can have powerful implications for data users and effect on how businesses operate.

This is just a taste of our data visualization design approach at IBM Business Analytics. We could tell you more, but why not explore for yourself what design for data visualization can look like. First time visitors and locals are invited.


The Hybrid Cloud design team for IBM Business Analytics showcases just how much their design work impacts our products and helps our users achieve better results. Thanks to the incredible talent and skill coming from IBM Design, we are able to make all of our products more focused on a human experience.

A special thank you to my colleagues Anne Stevens and Lisa Marie Chen for writing this article with me and sharing their perspective.


Arin Bhowmick (@arinbhowmick) is Vice President, Design at IBM based in San Francisco, California. The above article is personal and does not necessarily represent IBM’s positions, strategies or opinions.

Like what you read? Give Arin Bhowmick a round of applause.

From a quick cheer to a standing ovation, clap to show how much you enjoyed this story.