In 2020, over 64,200,000,000,000 gigabytes of data was produced online. This would take 1.8 billion years to download! With so much data at our fingertips, how can UX Researchers leverage it to better understand their business and user needs? This talk uses real-life examples of how the discipline of data science can (and should!) complement UX research to create better user experiences.
Grishma Jena, Data Scientist with the UX Research Operations team for IBM Software in San Francisco, USA, recently spoke at UXNZ, the leading UX and IA conference in New Zealand hosted by Optimal Workshop, on how Data Scientists can work in synergy with UX researchers.
In her talk, Grishma uncovers the benefits of bridging the gap between quantitative and qualitative perspectives in the pursuit of creating better, more user-centric products.
Background on Grishma Jena
Grishma is a Data Scientist with the UX Research Operations team for IBM Software. As the only Data Scientist in the organization, she supports 100+ user researchers and designers and uses data to understand user struggles and opportunities to enhance user experiences. She earned her Masters in Computer Science at the University of Pennsylvania. Her research interests are in Machine Learning and Natural Language Processing. She has spoken and facilitated workshops at multiple conferences including PyCon US (the largest Python conference in the world). She has also taught Python at the San Francisco Public Library.
She enjoys introducing new technical concepts to people and helping them use data and code to drive change. In her free time, Grishma enjoys traveling, cooking, writing, and acting.
Contact Details:
Email: grishma.jena@gmail.com
Why Data Science and UX Research Should Be Best Friends 🐰ྀི🐻ིྀ
Grishma highlights the beneficial and often necessary synergy between data science and user experience research. She first explains how data science fits into UX, and then briefly provides an overview of the data science process. Through this process, valuable insights can be shared with user research teams who can then interpret and share them with designers, developers, and product managers to create better user experiences.
Data Science in UX ⚛
Data science in user research involves using data-driven techniques to gain insights from user behavior and interactions, ultimately improving the user experience. Examples of data science in user research include:
- Understanding user struggles: Identifying user issues and preventing them from leaving the platform.
- Segmentation: Identifying distinct user segments within the product's user base.
- Usage patterns analysis: Studying how users engage with the product, including those who use it less frequently.
- User behavior prediction: Predicting how users will interact with the product.
- Feature prioritization: Helping product teams decide which features to develop and prioritize.
- Triangulation with qualitative research: Combining quantitative data analysis with qualitative insights.
- Personalization: Tailoring user experiences based on identified user segments.
The Data Science Pipeline 📊
Data Scientists generally start off with a question and a set of data, followed by a process of ‘data wrangling’, cleaning, exploring/modeling, and evaluating. Data Scientists use various processes, algorithms, and machine learning techniques, for example, to extract patterns and insights.
Generally, the process is as follows:
- Research question: Start with a research question that seeks to provide insights into user behavior or product performance.
- Data collection: Gather relevant data from structured, semi-structured, or unstructured sources.
- Data wrangling: Process and transform messy data into a usable format for analysis.
- Data exploration: Investigate data distributions and patterns to formulate hypotheses.
- Model building: Develop models to predict outcomes or behavior based on identified features.
- Model evaluation: Assess the performance of the model using metrics like accuracy and precision.
- Storytelling: Present the insights gained from the model in a meaningful way, connecting them to the initial research question.
The goal of the data science pipeline is to transform raw data into actionable insights that drive decision-making and lead to improved user experiences. The process involves iteratively refining the analysis based on feedback from users and other teams, and revisiting earlier stages as needed, to ensure the quality and relevance of the insights generated.
Generally, data scientists are more quantitative, whereas user researchers are more qualitative. But what if we were to combine the two? Grishma goes on to explain real-life examples of how these disciplines can work in harmony to achieve exceptional user experience.
Why it matters 💥
Data scientists delve deep into the numerical aspects of user behavior and product performance, while user researchers typically focus on understanding user preferences, motivations, and behaviors through direct interaction and observation. These two roles approach the same challenge – improving products and user experiences – from different angles, each offering unique insights into user behavior and product performance.
By combining the quantitative rigor of data science with the empathetic understanding of user researchers, a synergy emerges that can unlock a deeper, more holistic understanding of user needs, behaviors, and pain points. This collaboration has the potential to not only reveal blind spots in product development but also drive innovation and enhance the overall user experience.
To illustrate the power of this collaboration, Grishma describes real-life case studies from Airbnb, Google, Spotify, and ABN Amro. Below is a high-level summary of each case study:
- Airbnb: By combining data science with user research, Airbnb gained insights into host preferences based on city size. Data scientists helped develop predictive models for booking acceptance, enhancing the user experience. Additionally, a collaborative effort between data scientists, designers, and developers improved conversion rates, showcasing the power of interdisciplinary teams.
- Google: Google used deep learning to predict web page element usability, reducing the need for resource-intensive usability testing. This approach highlights how data science can complement traditional user research methods, especially in time-constrained situations.
- Spotify: Spotify's case exemplifies the synergy between data science and user research. They identified an issue where a power user misunderstood ad skip limits. Data scientists detected the anomaly, while user researchers delved into the user's perspective. Together, they improved messaging, demonstrating how combining data-driven insights with user understanding leads to impactful solutions.
- ABN Amro: In the case of ABN Amro, user research helped address an issue that arose from a machine learning model. User validation revealed the model's shortcomings, prompting collaboration between user researchers and data scientists to find a balanced solution. This case illustrates how user research can prevent potential failures and optimize product usability.
In summary, data scientists and user researchers have different perspectives, strengths, and weaknesses. Collaborating allows the two disciplines to:
- Gain a holistic understanding of products and users.
- Balance qualitative and quantitative data.
- Mitigate biases and validate findings.
- Compare user actions with self-reported intentions.
- Make proactive decisions and predict user behavior.
- Humanize data and remember the people behind it.
The synergy between data science and user research ultimately leads to a more comprehensive understanding of user needs, better product design, and improved user experiences. It ensures that both the quantitative and qualitative aspects of user behavior are considered, creating a more empathetic and user-centric approach to product development.
