Summary
Creating value from data requires data products, i.e. innovative analytical solutions that extract relevant business information from the data for decision makers. Like any innovation, most data innovations fail in one or more dimensions: economic viability, user desirability, or technical feasibility. The combination of design thinking and data science – often called data thinking – helps to minimize the product development risks by deepening the business, user and data understanding.
Over a three-part series we explain how design thinking is applied to data science projects to build data products that are economically viable, desirable for the intended users, and technically feasible. The three parts are:
- Part 1 introduces the design thinking approach and examines the difficulties faced in designing successful data & analytics solutions
- Part 2 shows why a deep business and user understanding is necessary and how the business model and analytics use case canvas are helpful tools to gain this understanding
- Part 3 details the third design phase and explains the usage of the data understanding and the data landscape canvas. It concludes with a discussion on how to blend the business, user, and data perspectives.
Design thinking – but what for?
Although design thinking’s application in the business world is a fairly recent development, it’s undeniably a methodology that’s evolved over many decades. Whether we go back to the 1950’s United States , where Buckminster Fuller coined the term ‘design science revolution’, or to the Scandinavian cooperative design pioneers of the 1960’s, we’ll find major influences on what we nowadays call design thinking.
Fuller’s interdisciplinary teams of highly trained experts worked in a very process-driven way, in their attempt to solve truly grand problems(i). Their goals were so challenging and complex, that systematic processes and the involvement of the best available experts from various fields where imperative. The Scandinavian cooperative design, or participatory design as it’s mostly referred to today, took a different approach. They invited people affected by the outcome of the design process to participate. Consequently, they invented tools and approaches that allowed the designer to take the role of a facilitator and the team members to actively participate in the creative process.
Design thinking as it’s known today still shows these influences. Whether it be the elaborate processes of the design science revolution, or the participative elements that go back to Scandinavian cooperative design, with their focus on the end user’s needs. The ingredients used to develop products and services have radically changed though. We’re working with data, often from multiple sources, and algorithms. Much of this data is highly complex and the algorithms are under constant development from an ever-growing global open-source community. What does it take to successfully apply design thinking in this new environment?
Dimensions required to design a successful data product
We’ve applied design thinking in a wide range of projects and across industries, where the objective was always to develop products or solutions that leverage data and algorithms as main ingredients. These projects typically come with an inherent invention risk. It’s basically unknown if the data and/or algorithms will be sufficient to address the identified user problem. We refer to these products as ‘data products’ and their ultimate objective is to generate user benefits – or in more general terms: business value – from data. A data product can be a commercial, external facing service or an internal solution for colleagues and other departments.
In our very narrow domain of analytics, data science, and AI development, we’ve seen many projects kicking off with a lot of enthusiasm and yet, only few of them made it to successful deployment. Brian T. O’Neill states in a post that 85% of analytics, AI and big data projects fail. O’Neil reviewed various posts and publications that focus on the challenges of such projects. He made another observation that we’ve experienced far too often, we refer to this one as the “solution looking for a problem” issue. Many data science, analytics, and AI developments fail to address relevant user needs.
How could this happen, given all the talk about design thinking and user-centered design? Maybe the user needs often vanish from the radar because teams still follow the so called ‘cross-industry standard process for data mining’, short: CRISP-DM(ii). CRISP-DM was developed in the mid 1990s by a group of 5 companies, funded by the European Union. It breaks down the data mining processes into six phases and was perceived the leading methodology for data mining projects for many years. Data mining has very little to do with the development of usable or even user-friendly products. At best, it describes the process of data exploration. It thus doesn’t come as a surprise that understanding the users, let alone inviting users to participate in the development, isn’t a necessary requirement in the CRISP-DM process. Business- and data understanding are not sufficient when we’re ultimately trying to create data products perceived as useful and actually used by real users.
A successful data product needs to address three core dimensions: it has to be viable from a business perspective, desirable for the end users and last but not least, it has to be feasible from a data perspective. We thus propose to add the business and data understanding perspective from CRISP-DM as explicit dimensions to the design thinking method. Arguably, design thinking has always been about creating desirable and viable products and services, so the major addition is the data understanding. The combination of design thinking and data (science) is by no means our invention, it’s rather a necessity taught by practical experience. Rachel Woods proposed a ‘Design Thinking Mindset for Data Science’. More and more blog posts and papers about the benefit of blending design thinking and data (science) are being published.
In the following articles we’ll explain how to approach ‘Data Design Thinking’ – or short: ‘Data Thinking’ and how ultimately all three dimensions come together, so our products are viable, feasible and desirable.
Part 2, showing why a deep business and user understanding is necessary and how the business model and analytics use case canvas are helpful tools to gain this understanding, will be published next week – 7 July.
References:
(i) R. B. Fuller, Nine Chains to the Moon: An Adventure Story of Thought (First ed.), Philadelphia: Lippincott, 1938.
(ii) C. Shearer, “The CRISP-DM Model: The New Blueprint for Data Mining,” Journal of Data Warehousing, vol. 5, no. 4, 2000.
