The Data Product Design Thinking Process
How to use Design Thinking for wicked data visualization problems.
Curiosity is a major driver for all of us. We are constantly carrying out research into processes, causes and effects.
We want to understand how companies, operational processes and economic relations work, what steps and parts they consist of, why things happen and how everything is interrelated.
Once we have analyzed everything, we can use this knowledge to exert influence on the world in a positive way.
Data is an increasingly important factor in the search for correlations. The right data (smart data instead of big data) shed more light on the darkness. Without data, an accurate, systematic analysis and precise design of the world around us is no longer conceivable. Intuition and gut feeling alone are no longer enough.
But data in its raw version (binary, in tables, logfiles, databases, etc.) cannot actually be read by humans. If we want to be fast, we need data visualizations that make the right selection of data easily and precisely visible at the appropriate level of aggregation. This task is performed by so-called data products. They visualize our world in a suitable form and make it accessible for our analysis and our search for causes and effects.
Data products such as Google Maps, Google Analytics or mobile apps for weather, health or timetables provide us with this at a really high level. What we see as an attractive, clear display is a highly complex condensation and sophisticated visual summary of data, connections and structures.
What is a Data Product?
Data products are digital products that use and visualize data to support decision-making and help control complex systems.
A data product is the visually processed result, with the help of algorithms, of a statistical, mostly complex data analysis of systems. Systems can be processes, facilities, companies, cities, countries, infrastructure, solar systems, organisms, ecosystems, climate conditions. Indeed, our world consists of an infinite number of systems.
A data product describes historical, current or future conditions and developments of this system. Users use data products to describe complex conditions and developments of systems quickly and visually and to derive information and their own decisions from them.
Designing with Data
The visual interface between machine (data product) and human being is very important.
But how exactly do you develop them? How do we ensure that they fulfil their tasks effectively? Especially in professional environments, very different requirements are placed on a new data product by very different users. You can quickly find yourself in a situation where you have to solve a tricky problem (wicked problem).
Tricky problems are characterized by the following characteristics:
• there are many interdependent influencing variables
• there are no right or wrong solutions, only good and bad ones
• you can’t solve them with a formula and logic alone
• you can only solve them intuitively, not logically
• there are social, psychological and cultural aspects to consider.
HOW TO DEAL WITH WICKED PROBLEMS
The challenge with data products is…
on the content side:
· the complexity of the underlying system is high
on the data side:
· data from different data sources must be connected to each other
· the data quality is often insufficient
· the preparation of data (ETL) is complex and time-consuming
· data science techniques must be properly integrated
on the visualization side:
· the classic visualization methods (bar chart, pie chart, dot plot) are no longer sufficient
· the possibilities of visualization with common tools like Qlik, Tableau and PowerBI are limited
on the user side:
· the users have very different knowledge
· they have to do very different jobs (jobs to be done)
· decoding the information contained in data products and assigning it to your own knowledge is a challenge
· the working environments of the users (medical technology, system control, vehicle control) are highly complex
More details on all these aspects can also be found in our article on the Data Design Guide. here
What is Design Thinking?
Design Thinking offers a very effective approach to solving tricky problems. Here, an interdisciplinary team approaches a potential solution in individual, systematic steps and from the user’s perspective. This is then tested for suitability by users.
Design Thinking has become a well-known and well-proven method of developing innovation. Many large technology and consulting companies use it.
The internet is full of practical and descriptive explanations of Design Thinking, so here is just a brief overview of the four underlying phases in the Design Thinking process.
· Discover — Discover the environment and context for the question or problem under consideration
· Define — Formulate the requirements and needs of users
· Develop — Develop initial ideas to the solution
· Deliver — Create a solution and get user feedback.
Design Thinking for Data Products
We wondered whether this method could also be used for data products. Design thinking is all about the users and their needs. Now another very complex dimension is added: data or data science. We then adapted the classic design thinking method specifically for data products and have been working with it very successfully since the beginning of 2018.
The challenge of using Design Thinking in the development of data products for business applications lies in the following points:
• Data product development for Big Data and Data Science applications often breaks new ground. At this point, the target use (job to be done) is not clearly defined at the beginning. The context of use and user requirements must be developed equally with the data product. Sometimes they only become apparent after the first visualization of the previously invisible system.
• The data competence of the user is unknown or too low to understand the message
• Designers need a deep insight into the data structure, the raw data and the data pipeline. What is developed creatively must also be technically possible. A pure UX view is not enough.
• Many data products are based on data science algorithms. Data and the potential of statistics and data science are the raw material for the creative process. Data products only become efficient when they fully exploit this data potential. Here, too, designers cannot go any further without an understanding of data literacy and data science.
• Data products in a professional environment are aimed at users and experts in specialist fields. The data product thus intervenes deeply in so-called business domains — a further level of complexity.
Authors: Evelyn Münster / Christoph Nieberding
About us
Evelyn Münster and Christoph Nieberding are managing partners at Designation, a Munich based design company for data products and business design. www.designation.eu
