UX + Data: An integral research approach
In this article we delve in the integral approach that combines UX Research and Data Science, its added value and how they complement each other.
“All data created by people. And all people create data…Today we divorce people from their data, and that gives companies a license to forget about the people behind the data…It allows us to divorce ourselves from the responsibility of what that data can do.” – Ovetta Sampson, Microsoft
Data Science and UX are highly related with decision making, whether from a business perspective, to improve the user experience or even to ameliorate product performance. In order to minimize blind spots when making decisions, it is essential to have as much data as we can on the table and extract the right insights that allow us to understand the problem to be solved and to execute the best solution. This can be obtained from user interviews to server logs, but in the end data is a representation and a way of measuring what real people are doing with our products in the real world (Ted Institute, 2014). Therefore, we must be conscious of what we do with that data; how we use it and with which purpose we are gathering it.
To guarantee a conscious human-centric process, we believe horizontal and cooperative collaboration is important, in which all teams are empowered. In this sense, even though UX and DS (Data Science) teams have their differences, there is a common denominator between them: the extraction of insights from data. This enables multiple collaboration spaces that, when enhanced, provide a holistic understanding of data and help mitigate the weakness of using a single research approach.
The circle of UX
The UX approach is human-centric. UX Designers and Researchers are more interested in knowing users’ intentions; why they use a feature and how they feel about it. Research teams work from empathy with users, delivering insights about them, seeking patterns in thick data and building hypotheses. Data is mostly obtained from systematic research methods such as user interviews, surveys or diary studies, among others. These insights intend to answer questions like what are the user’s motivations to use a product? Or what role does the product play in their daily life?
The circle of DS
Data scientists are more interested in how the product is performing or is expected to perform. They work with systematic gathering of large volumes of data, seeking for features’ correlations, classifications, making statistical models with applied Machine Learning, among others.
This approach allows us to answer questions such as How does a product feature change behavioral metrics, such as clicks or time spent? Or which features are used and which are abandoned?
UX and Data Science working together, a powerful intersection
Despite some differences that may exist in how they treat data, both disciplines have points in common, so multiple spaces for collaboration are enabled. As a common denominator, both teams seek to explain phenomena observed in data and generate insights for decision-making.
Both UX and Data Science teams follow the scientific method which includes the cycle of:
- Identifying the problem
- hypothesizing and exploring
- designing and testing
- and finally, error analysis or feedback analysis (Salehi, 2022).
This common ground enables interesting spaces for collaboration. An example may be the generation of data-driven personas (Xiang et al, 2016). UX Researchers usually generate archetypal users for having a shared understanding across different teams involved in the product cycle. These user personas rely on workflow data from surveys, self-reports, interviews, and user observation. However, this data, that is not directly related to user behavior, weakly reflects a user’s actual workflow in the product, is costly to collect, is limited to a few hundred responses, and is outdated as soon as a persona’s workflows evolve.
With Data Science techniques it is possible to gather behaviors from telemetry data directly when the user is interacting with the product. Then, by way of clustering techniques we can arrive at data-driven generated user groups and workflows that UX researchers can validate. This approach that triangulates qualitative and quantitative information can also contribute in minimizing bias when working with personas -and is time and cost effective-.
Another interesting space for collaboration is to understand what users say and do. A company that is ahead in integrating data science and UX teams is Spotify (Xiang et al, 2016). Through their simultaneous triangulation method, they combine UX Research and Data Science in the same process. For example, combining into the same experiment a diary study and data tracking for the same group of users.
They concluded that both teams should work together following this process:
- Define research questions
- Mix complementary methods
- Implement methods simultaneously
On the other hand, UX Research methods and findings from user interviews may be valuable to define features needed to extract from data, or to evaluate if a trained model meets the user’s needs and expectations (Nahirnyi, 2021).
Last but not least, to make Machine Learning powered applications more inclusive, decisions cannot take place in black box algorithms (Campbell, 2022). It is important to take the knowledge gathered from UX research about users’ mental models into consideration, so that they are able to perceive and understand how the algorithms are working for them. This implies both teams working together to minimize biases towards ethics and transparency of the product.
For us, it is crucial to provide the best possible digital solutions from beginning to end. We believe this integrative approach makes the difference in understanding users to design and develop high quality digital products that respond to your business’ users needs. Basing our decisions in accurate and varied data (as we have explained, data obtained from UX and Data Science offers different insights and type of information) is key.
References
Campbell, M. (2022, august 22). Ethics and bias: the UX of AI – Bootcamp. Medium. https://bootcamp.uxdesign.cc/ethics-and-bias-the-ux-of-ai-cba22a01a896
Nahirnyi, A. (2021, march 30). Every Machine Learning Team Needs a UX Researcher | UX Booth. Uxbooth. https://www.uxbooth.com/articles/every-machine-learning-team-needs-a-ux-researcher/
Salehi, B. (2022, june 9). UX + data science = smarter decisions – UX Collective. Medium. https://uxdesign.cc/ux-data-science-smarter-decisions-6ea847c7288f
TED Institute. (2014, december 23). Rochelle King: The complex relationship between data and design in UX [Vídeo]. YouTube. https://www.youtube.com/watch?v=YTRIeWI0EGQ&t=126s
Xiang, Z., Brown, H., & Shankar, A. (2016, may 7). Data-driven Personas: Constructing Archetypal Users with Clickstreams and User Telemetry. ACM Digital Library – Universidad Nacional de Cuyo. https://dl.acm.org/doi/10.1145/2858036.2858523