How to unite data science with design thinking
A brief guide for design agencies who like to work with data, offering four challenges and solutions.
A designer's mindset is trained to solve problems by going broad and narrow down again; to think what is possible as a solution. A data scientist scopes the problem as small as possible and explores how to get a more concrete solution.
4 challenges when designers and data scientist work together in an agency:
Difference in mindset
Practical data constraints
Difference in mindset
Data scientists like to scope things as narrow as possible and then make things way to complicated. Designers, on the other hand, tend to first go broad and then narrow down towards a logical conclusion. This conclusion, however, is always way too vague for data.
While designers thrive on insecurity, ambiguity and 'wicked challenges', data scientists are trained to stay away from these dangers.
As explained by one of our junior data scientists - "During university, any insecurity needs to be kicked out, as we cannot prove any value otherwise."
While designers tend to have an abductive way of reasoning (what is a possible solution for my problem), data scientists have a deductive way of reasoning (forming a logical conclusion based on assumptions). Without better understanding each other's way of thinking, the chances of aligning both are low.
Most agencies work in a matrix structure, crossing projects over different disciplinary teams (like a design, business, and data team). Although project teams seem to collaborate on a day-to-day basis, the teams themselves are steered different in different ways. Although team members recognize the importance of contribution from other disciplines, both teams tackle client needs differently, have separate proposition offerings, exploit opportunity separately, and employees are developed independently.
A pragmatic culture supports an agency’s agility and ability to rapidly respond to contextual changes. However, this pragmatic culture also results in ad hoc processes and a lack of formalization. This results in people missing cross-functional learning, communication, and knowledge sharing.
In contrast to design projects, using data science in practice is far more constrained. Looking at availability, accessibility, and affordability (AAA), in practice, regulations (like GDPR) and technical constraints cause high effort in acquiring the needed data. Next to that, the first phases of data science focus on preparing and cleaning the data before any analysis can be made. Hence, using data for your design process demand far higher resources than thought of before.
Establishing synergy in findings of data-design collaborations is complicated. On many occasions, the findings of the data and design teams do not match. During one of our client projects, the design team developed a persona based around ‘experience’. For a designer, experience is a term where you can easily think of creative concepts.
While the data segmentation, in theory, is based on the same set of customers, there are no metrics for ‘experience’. This increased the complexity of the work and decreased the quality of the delivery. Without predetermining hypothesises and metric that both teams aim to align, the two insights will always be too far apart.
4 ways to tackle these challenges
Start with projects needs and constraints
Map and monitor touchpoints
Perform collaborative stand-ups
Meet each other at business value
Start with (and reflect on) project needs and constraints
Before a new project starts, the team should map all the project needs and constraints. Answering one question: what do we need to run our project successfully? These needs consist of standards elements like time, data, or expertise. But also factors like internal stress emotional status of the team members are important elements to consider.
This is especially important regarding the data resources (as these have high requirements). Each decision should be critically reflected on; Is the data there? Is the data accessible for me? Is the data affordable to prepare for analysis (time and budget)?
Next question: Do we have these elements? Yes. Great, let’s start. But unfortunately, most of the times it’s a no. Well, what do we need to do to get it and does the project timeline allows for this? Otherwise, the project needs a pivot on either time, expectations, or scope.
Synergy between teams: Map and monitor touchpoints
After all project constraints are clear, the project team should map the touchpoints between the two teams. What are the interdependencies between the teams? When do we expect to deliver results? During decision-making, the data and design teams must find synergy between their activities and results.
The teams need to answer three questions during proposal drafting: What are the activities the data and design team will perform? What are the results of these activities? How will we combine these results so that the sum of the parts is greater than the parts?
Collaborative Decisions: Perform collaborative stand-ups
Because these collaborative projects tend to switch scope a lot of times, daily contact is crucial to keep reviewing project constraints and touchpoints. Just a 30 min stand-up would suffice for updating and getting crucial information at the table. While it is common to hold daily stand-up scrum meetings in the morning, this might not be the best fit for your team. Many teams hold their meetings mid-day, or even at the end of the day. Choosing the best time to hold your stand-up can be difficult.
Keep in mind, a stand-up is not actual collaboration. Only knowing what the other team is doing, does not mean you’re actively collaborating.
Meet each other at the client’s business value
Data scientist and designers tend to speak two different languages and have two different mindsets. Therefore, the team needs to find common ground. As the main goal of the project in an agency (in addition to technical or customer values) is to add value to the client. Therefore, during all decision-making processes, the team should continuously place the client’s business value central.
What is the value to the client if we perform this analysis? Great, new UI design, but does it generate a higher click-through rate? Challenging each other on delivering value allows the team to communicate on the same level. And, of course, meeting project deliverables.
About the author: Nick is the co-founder and director of Yabba Data Doo. As a designer, his passion is to add value to people's lives on the intersection of business and data science. During his studies in Industrial Design Engineering at the TU Delft, Nick learned the ropes of strategy consulting during multiple internships, only to be seeing that projects ended in bullet points and lengthy reports, and no real action was being achieved. That is when he took measures in his own hands and after co-founding his first startup Hugo, he graduated on the intersection of innovation and data science and proceeded to start Yabba Data Doo.