Learn how to apply Design Thinking to any Data Science case

Imagine the world as a gigantic puzzle with millions of pieces. The puzzle is still not complete, some pieces are already placed, but there is still space for new ones. We want to make our solution a puzzle piece that fits into this huge puzzle, but we do not know-how. Design and Agile Thinking help us shape our solution into a puzzle piece that fits into this huge puzzle.
Creating a data product can be chaotic and hard. My advice is to not give up easily when it at first seems as if there is no solution, but to keep trying to build the pieces!
Following the Design and Agile Thinking as your guides in your trip can make the design, implementation, and arrival at your solution easier.
You have probably heard about the Design and Agile Thinking approaches since they are widely being used in many organizations. This article gives a brief introduction to both approaches and explains how to apply Design Thinking in a Data Science case.
If you are more interested in how to apply Agile Thinking in a Data Science case, check Part 2 of this article, otherwise, keep reading!
Design or Agile Thinking?
Difference #1
If you are in a situation that is blurry about what the actual problem is and what the tangible solutions are, then you can use the steps of the Design Thinking method to help your way through it.
If you are in a situation in which the problem is well-defined and you need to manage your way until you reach the solution on time, then you can follow the Agile Thinking method.
Difference #2
Design Thinking is used to redefine problems, understand users, experiment with novel concepts, and deliver optimal and innovative designs of solutions.
Agile Thinking is a project management method that transforms the design of a solution, which is the outcome of the Design Thinking process, to an actual solution.
Similarities
Both have similarities such as being:
- human-centered, setting as a priority the end user’s and stakeholders’ needs,
- iterative, consisting of multiple cycles and improving with the provided feedback in each iteration,
- option-focused, listing possibilities for exploration
The combination of those two methods can ensure high-quality solutions to real problems within organizations.
Design Thinking: Data science without data
In most cases, huge amounts of data are available and if we start immediately by looking at the data, we will probably get lost while trying to understand the data columns and fields and forget about the initial problem. So, we should take a step back, detach from the details and try to create a holistic overview of the organization’s processes and its bottlenecks.
Our desired outcome in this phase is to identify problems and design solutions. Since we want to bring a data-driven solution, the design should include an initial draft of the data requirements and a first concept of the data product. In this phase, we should also define the users of the end product and include them in the design process.
The Design Thinking approach consists of three steps; problem understanding, divergence, and convergence, as displayed in the diagram below.

Step #1: Problem understanding
This phase aims mainly to a deeper understanding of the current situation within the organization.

A. Interviews
Arranging interviews with different people from the organization, helps us gain insights into their processes, communication styles, and vision. From a Data Scientist’s perspective, the expected key outcome is to discover the data maturity level of the organization, which shows the current role of Data Science in their organization. To reach this level of understanding, we made a list of questions for the interviewees.
Personal Questions
- What is your background?
- What is your role within the company? In which department, do you work?
- How much do you know about Data Science?
Team Questions
- Are you part of or do you manage a team?
- What are the roles of the colleagues that you closely collaborate with?
- What is the function of your team?
- How is the team’s structure in terms of size and hierarchy?
- How frequently do you communicate within the team and with the other departments?
Business Questions
- What are the most challenging problems of your organization?
- What are the most challenging problems of your team?
Data Science Questions
- Do you have any examples of Data Science related projects within your organization? If yes, what were those projects about?
- Do you have any Data Scientists working for your organization? If yes, can you give some examples of their impact?
- Do you have any idea or suggestion about how data science could handle challenges within your organization?
Data Pipeline Questions
- Do you currently collect or possess any data?
- Do you perform tasks that involve working or analyzing the data?
- How is the data stored and extracted?
- What is the size of the data?
Technologies
- What type of dashboards do you use for data visualization/analytics?
- What kind of database systems do you use for data storage?
- Do you use any cloud technologies?
- Do you have any preferred programming language for data analysis/modeling?
Vision Questions
- Do you see your organization developing towards a more data-driven organization in the next ten years?
- How easily do the employees adapt to new technologies, such as a new piece of software or data product?
B. Key findings presentation
The presentation should include a summary of the outcomes from the interviews and summarize the major challenges that the organization faces.
For adding the Data Science perspective, defining the data maturity level of the organization is crucial for moving the data use to the next level. The maturity can vary from a company not using any data to a company trusting the data and basing its decision-making on the data insights.
The following diagram shows the journey to a data-driven organization.

- Data-denial: active refusal of data usage
- Data-indifference: lack of interest in the usage of the data
- Data-aware: data collection and visualizations of historical data for past events understanding (descriptive analysis)
- Data-informed: models based on the historical data for future event prediction (predictive analytics)
- Data-driven: advice on the decision-making based on the predictions (prescriptive analytics)
After creating awareness about the key challenges and the organization’s data maturity level, the presentation should also include some examples of how Data Science can help an organization handle some of its challenges and transform into a more data-driven one.
Step #2: Divergence
What if everything was possible? We should go through this phase with this question in mind and explore every possibility. The phase consists of two parts; the first draft of the stakeholder map and the brainstorming session.

A. Stakeholder map (first draft)
After the problem understanding phase, we gained a good overview of the parties involved in our project. To organize this view of different stakeholders as well as their involvement and impact, we need to create the first draft of a stakeholder map, as displayed below.

The two main variables that differentiate the different types of stakeholders are their interest and their influence level in our project, and all combinations of those two variables are possible.
For now, we don’t need to make a detailed map of all the stakeholders, but a first draft is enough to continue to the next steps of the Design Thinking method.
B. Brainstorming
Having completed the stakeholder map, we select the most representative stakeholders from the different parts of the map, so that we have a suitable combination of stakeholders with different impact and interest levels. Then we invite those stakeholders to our brainstorming session.
During the brainstorming, we create two or three groups with different stakeholders. We split the session into two parts; the ideation for challenges and the ideation for solutions.
For the ideation for challenges part, we ask the participants about the two major challenges they face, group them and let them vote for the most important ones. Then, for the most voted challenges, we request them to determine the stakeholders affected by them and group them together.
After that, we select the three most voted challenges and their most affected stakeholders and kick off the ideation for solutions part. Later, we ask the participants to imagine the ideal situation for the closely related stakeholders and to identify the bottlenecks that prevent the organization to reach this ideal situation. Then, we let them brainstorm about ways for managing those bottlenecks and group them by similarity.
Step #3: Convergence
After diverging and exploring a variety of challenges and their possible solutions, adding no limitations, it is time to converge. The aim here is to define a specific use case and propose a data-driven design concept to the stakeholders.

A. Concept creation:
Our priority here is to summarize all the challenges and solution ideas that were created during the brainstorming session and focus on the ones that seemed to have the greatest impact on the organization and to be more burdensome. For each one, we work on defining a first draft of the data requirements, a concept of the data product, and a high-level data-driven data pipeline.
For the first draft of the data requirements, we think about data availability, possible data collection, and data privacy issues. After that, we work on making a first sketch of the final data product, incorporating the initial data requirements. With the data requirements and the end product in mind, we design an overview of a data pipeline that includes the entire process from the data collection to the deployment of the final product and its delivery to the stakeholders.
Having a good first design of the solutions of the key challenges defined during the brainstorming, we arrange meetings with the main stakeholders. We use the time of these meetings to discuss those designs and converge to one design that solves a use case that seems to be more interesting for them and more realistic to be solved.
B. Stakeholder map refinement
After having defined a first design of the solution and a specific use case, it is time to review the first draft of the stakeholder map, which was created before the brainstorming session.
The primary purpose of this step is to refine the different groups of stakeholders and create three teams; the data, feedback, and management team. Those teams are essential for the co-creation process of the development of the product and for enabling effective collaboration with them.

The data team will include stakeholders that are closely related to the data such as Data or Business Analysts. This team will help with data collection and data privacy. The feedback team will consist of the end-users that will experiment with the prototype and the end product and provide their feedback. In the management team, the stakeholders higher in the hierarchy will be included. This team will serve as the steering committee and will have a high impact on the decision-making of the project direction.
C. Project plan
The last step before working towards implementing the design is to create a project management plan. This is a transition step between the Design Thinking and the Agile Thinking method.
It is essential to review all the information created during the Design Thinking process, including the data requirements, the design of the solution, and the different groups of stakeholders. Based on this review, we create an agile plan, after its execution will deliver the data product.
Wrapping up
Using Design Thinking as a guide makes the shaping of a suitable solution becomes easier. After the design, implementation comes and Agile Thinking can be our guide in this process.
Read Part 2 of this article to learn more about how to apply Agile Thinking in your Data Science case.