Instead of walling off your data scientists to crunch numbers all day, integrate them with your design team. A human-centered approach to data science is essential for developing smart new products that consumers can actually use. Instead of a version of data science that is narrowly focused on researching new statistical models or building better data visualizations, a design-thinking approach recognizes data scientists as creative problem solvers. We’re not suggesting that the disciplines of data science and design merge, but rather that if practitioners work together and learn each other’s art they will produce better outcomes. Many of the techniques used in design thinking approaches—such as user research, analogous inspiration, sketching and prototyping—also work well with data-driven products, services, and experiences.
Rise Science came to IDEO with a challenge. The young startup had built a robust data platform for college and professional athletes to track their sleep and adjust their behavior so that they played at peak performance. But for the players, the experience was challenging. Rise expected athletes to look at data-driven charts and graphs to determine what decisions to make next, but players struggled to find those insights. Rise was convinced they just needed easier-to-read charts and graphs.
As IDEO designers and Rise’s data scientists spent time with players and coaches, they discovered that Rise didn’t have a data visualization problem, they had a user experience problem. Charts and graphs were far less important than knowing when to go to bed each night and when to wake up the next morning. Within a few weeks, the charts and graphs moved into the background of their app and an alarm clock and a chat tool took center stage.
In the 18 months since relaunching their service, Rise Science has signed up over 15 of the most elite pro and collegiate sports teams, as well as several companies who hope to improve employee performance and wellbeing through better sleep habits.
This example shows how human-centered data science can result from interdisciplinary teams incorporating design thinking into their approach. Instead of a version of data science that is narrowly focused on researching new statistical models or building better data visualizations, a design-thinking approach recognizes data scientists as creative problem solvers. We’re not suggesting that the disciplines of data science and design merge, but rather that if practitioners work together and learn each other’s art they will produce better outcomes.
Many of the techniques we use in our human-centered design process at IDEO—user research, analogous inspiration, sketching and prototyping—work well with data-driven products, services, and experiences.
Data by themselves are inert—dumb, raw material. Making things smart will mean designing with data in a way that reflects and responds to the functional, social, and emotional behavior of users. If you start with the needs and insights of people rather than leading with data, you can gain insights through the combination of qualitative design research and exploratory data analysis. This hybrid approach can radically change the user experience for the better and be a true differentiator.
For example, Rise and IDEO visited college athletes in dorm rooms and training facilities to develop a deep understanding of their day-to-day needs – a common design-thinking practice known as user research. We observed that nearly every component of the player’s life is scheduled, measured, and optimized. Giving them more data to digest was simply too much to ask. We also learned that the onboarding experience and service touchpoints like in-the-moment sleep coaching were just as important to the athletes’ success as the data visuals tied to their sleep.
Data scientists see the world in unique ways, but they can only leverage that point of view when they have a chance to go out into the world and spend time with human beings. Engaging data scientists in design research produces powerful insights and, more importantly, unlocks empathy for the people who will be touched by the data engines they develop.
Analogous inspiration is another design-thinking technique that data scientists can add to their toolkit. That approach was key to the success of a project with Procter & Gamble. When the global organization deploys new technologies, they invite 50 key employees to a two-day training summit. When those employees return home, they are charged with teaching others the technology. While the training itself was effective, the process of hand-picking key employees was fraught. Turns out the people who were picked were not influential enough to spread their new knowledge throughout the company.
Our Data Science team found analogous inspiration in research on how diseases spread through social networks. In most cases of epidemiology, we are interested in disrupting and preventing the spread of a disease. In this case, we wanted the disease (a new technology) to spread as quickly as possible through the social fabric of Procter & Gamble. We wanted to engineer a technology pandemic.
The epidemiologic model allowed us to discover that the hand-selected employees, while close to senior leaders, were too centrally located to meaningfully spread something new throughout the organization. We suggested a different approach: Select participants who were distributed throughout the organization, but strategically located to spread the technology through the social fabric of the organization. We tried hundreds of millions of combinations of people that were best positioned to collectively spread this new technology. The initiative was so successful that the resulting group of ambassadors have been reused for several other change management programs.
Sketching and Prototyping
Data scientists can use sketches or prototypes to get user feedback, just the way product designers do. Although there’s a fog of ambiguity that descends during the exploratory data analysis phase, using the designer’s trick of making data visual helps your brain see patterns that suggest a few ways forward.
The key here is to not be afraid to iterate: A pattern might lead you to look at the data in a particular way which then causes you to look at patterns in a completely new light. You might also choose to return to exploratory data analysis to pivot or change directions altogether. Start with a lot of skepticism, as the data could be dirty or missing fields, hence indicating patterns that are off-course.
In another Procter & Gamble project, for example, we began with simple sketches which were shared with key stakeholders to get their input. Before ever writing a line of code we were able to learn what a desirable interface might look like.
It’s hard and expensive to build a data science team, so it’s no surprise that most companies task these teams only with “data science work.” But engaging data scientists in all stages of the design thinking process will pay off in incalculable ways. Interdisciplinary collaboration that pulls data scientists away from their screens and out into the world produces powerful results. It transforms data from a crude tool to measure your business into a sophisticated tool that helps your business grow.