Skip to content

Data Science Meets Design Thinking

Posted in Data Science, Data Visualization, and Design Thinking

In the O’Reilly article, Design thinking and data science: Solving problems with data necessitates a diversity of thought, Dean Malmgren from Datascope and Jon Wettersten from IDEO shares:

“Problem solving not only requires a high-level conceptual understanding of the challenge, but also a deep understanding of the nuances of a challenge… Solving problems with data necessitates a diversity of thought and an approach that balances number crunching with thoughtful design to solve targeted problems.”

Design thinking begins with understanding the needs and behaviors of the people you are designing for. In the context of designing data-driven solutions, you need to identify the correct problem to solve, explore what’s possible with data, and translate business problems into a data-driven design approach.

One approach is to collect feedback on design prototypes in an online survey, follow-up with in-person interviews, and then synthesize the results to inform future iterations of your design. This kind of research process is simultaneously quantitative, in that you’re using data from several survey respondents, and qualitative, in that you are getting a much deeper sense of how people might feel about your solution in the interviews.

And, rather than trying to summarize a group of people as being “males in their mid-40s with 2.3 kids,” develop a story on what the data might mean supported by more robust visualizations. Dig into your survey data and explore the survey results to generate hypotheses, which, in turn, guides your thinking during the interviews.

You get the best results when data scientists and interaction designers work closely together to brainstorm, sketch, mock-up, and build prototypes. By rapidly prototyping a concept, getting it in front of your clients early and iterating based on candid feedback, you can develop something useful in a matter of days or weeks that otherwise would have taken months or longer to build if you committed to building a fully-functional solution.

This process requires multi-disciplinary teams to apply their unique perspectives to identify user needs and to creatively use data as key design resources for the best experience design outcomes. Learn from one another and be inspired by each other’s skill sets and explore how you can better collaborate in the future. Take yourselves out of your comfort zones and extend your capabilities. Look at other projects where you can take advantage of the shared perspectives between Data Science and Design Thinking.


  1. Great recap. Although they’re often seen as orthogonal methods of inquiry, there is a lot of overlap between (good) data science and (good) design thinking. I really like the term “experimental design” as it serves as a bridge between these fields of thought.

    What is data science after all but — a science? And what is science built on but — experiments — which have to be designed.

    Data scientists can be lured by collecting more and more data without considering the biases and perspectives that may be hiding in the data. OTOH, designers working, for example, to improve customer engagement may be thinking creatively and rapidly iterating but perhaps not thinking in terms of falsifiable hypotheses.

    Good science requires both good data AND good design

    January 15, 2018
  2. Completely agree, George – it is the combination of good data and good design that is needed for good science – and great Design.

    January 15, 2018

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.