The Client's Challenge

The primary challenge faced by YouTube was that the existing dashboard provided fragmented analyses of machine learning models and classifier decisions. Stakeholders across various roles found it difficult to navigate the platform and extract meaningful insights from the classifier outputs.

This lack of a centralized interface not only hindered users from efficiently reviewing classifier decisions but also impeded their ability to act on these insights, resulting in delays in decision-making and an overall reduction in productivity.

My Role & Contributions:

As the Lead UX Designer for the Kapla Dashboard project, I spearheaded user research to identify stakeholder needs and developed an intuitive information architecture. I created wireframes and interactive prototypes to gather early feedback. Collaborating with data experts, I designed meaningful data visualizations and conducted usability testing to refine the dashboard. I also worked closely with cross-functional teams to ensure alignment on project objectives.

Additionally, I provided training and support to users, facilitating the effective adoption of the new tool. My user-centered approach transformed the Kapla Dashboard into an efficient and impactful resource for YouTube classifiers, enhancing their decision-making processes.

• 1 Lead UX Designer

• 1 UX Researcher

Team Size:

• Figma for UI/UX design

• FigJam

• Google Meet

Tools Used:

Project Duration:

• 8 weeks

Understanding User Needs: A Shift from Data Analysis to User-Centered Design

Initially, the project was framed as a data analysis problem. However, through extensive research, we identified key user challenges by engaging with various user groups, including content moderators, classifier users, operators, machine learning model developers, and quality assurance teams.

This deeper understanding of user needs shifted our focus from solely analyzing data to creating a more comprehensive solution that addresses the specific pain points and workflows of each user group. Ultimately, this led to a more effective and user-centered Kapla Dashboard design.

Identified User Groups and Their Needs

• Classifier Users: Focused on understanding the verdict of the classifier and its accuracy, they needed clear insights into how classification decisions were made and their implications.

• Operators: Interested in evaluating how specific classifiers impact video performance, they sought tools that could help them correlate classifier outputs with key performance metrics.

By addressing the distinct needs of these user groups, the Kapla Dashboard was tailored to facilitate better decision-making and improve overall content management processes.

To address these challenges, a new Kapla Dashboard was developed, featuring a streamlined, role-based user experience that consolidates core functionalities into standardized workflows. The primary objectives of the redesign were to simplify navigation, enhance data visualization, and ensure that users could easily access and act on classifier insights.

The Solution

A User-Centered Kapla Dashboard Experience

The Kapla Dashboard has transformed how YouTube employees interact with classifier outputs, providing a cohesive and user-friendly tool that maximizes the utility of machine learning insights.

Proposed Solution:

By consolidating fragmented workflows into a streamlined experience, the new dashboard has empowered stakeholders to make informed decisions quickly, ultimately driving improved content management and user engagement on the platform.

This case study highlights the value of designing tailored user experiences that align with organizational needs, leading to enhanced efficiency and effectiveness

Key Learnings

• User-Centric Research Drives Effective Solutions

  Engaging with diverse user groups like content moderators, classifier users, operators, ML developers, and QA teams revealed that a one-size-fits-all approach doesn’t address the unique needs of each stakeholder. Tailoring solutions based on specific user needs is essential for usability and satisfaction.

• Moving Beyond Data Analysis to Experience Design

  Initially, the project was treated as a data analysis issue. However, in-depth research highlighted the importance of creating workflows that make complex data actionable and accessible, transforming the project into an experience-focused solution rather than a data-centric one.

• Role-Based Insights Enhance Usability

  Understanding the varied objectives of classifier users, operators, and other stakeholders allowed for role-specific dashboards and workflows. Catering to individual roles enhances efficiency and empowers users to make informed decisions quickly.

• Cross-Functional Collaboration Strengthens Solutions

  Working closely with data scientists, ML developers, and content managers provided a holistic perspective, enriching the solution by incorporating expertise from multiple fields. Collaboration not only improved functionality but also aligned the dashboard with real-world requirements.

• Iterative Design is Crucial for Refinement

  Regular testing and feedback loops with users were vital in identifying and resolving usability issues early, resulting in a dashboard that better meets user expectations and is adaptive to ongoing changes in classifier and moderation needs.

• Visibility and Transparency Build Trust

  By providing clear insights into classifier decisions, accuracy, and impacts on video performance, the dashboard fosters user confidence and transparency, improving trust in classifier outputs and, ultimately, the platform’s decision-making process.