HR Data Insights

2024

Data Analytics

LLM

UX Research

Usability

Transforming complex HR analytics into an AI-assisted, self-serve experience that improved insight comprehension by 46% and platform navigability by 67%.

Irene L, Shadiah G. - Design & Product; Tanaya D, Jun Xue, Emil G - Data Sci; Anthony I, Kyle O - Fullstack Eng

Context

Using AI on HR data to uncover bias in the workplace

Sigma Squared, founded by two Harvard economists, uses econometrics and AI to help organizations identify and address disparities in their talent processes. As the early-stage startup grew, it aimed to scale its impact, turning rigorous analytics into clear, actionable insights for HR leaders.

Challenge

Low usage hinted at usability issues

Despite the strength of Sigma Squared’s analytics, engagement across the platform was low. HR leaders weren’t returning regularly or exploring the data deeply, and our data scientists were spending significant time walking customers through results. We suspected that usability — rather than lack of interest — was the root cause, but we didn’t yet know where the friction lay.

Our goal was to uncover why users struggled to extract value from the platform and identify the most impactful improvements to drive adoption.

Approach

Build Empathy Internally

To surface usability issues efficiently, we ran an internal empathy exercise. Non-design teammates role-played as HR professionals investigating rumors of bias, revealing where users struggled to interpret data or navigate the platform.

Build Empathy Internally

To surface usability issues efficiently, we ran an internal empathy exercise. Non-design teammates role-played as HR professionals investigating rumors of bias, revealing where users struggled to interpret data or navigate the platform.

Identify High-Impact Themes

We organized insights from the empathy exercise into design principles, then mapped them by impact and effort. High-impact opportunities centered on making data visualizations easier to interpret and helping users easily navigate their way to answer specific questions they had.

Identify High-Impact Themes

We organized insights from the empathy exercise into design principles, then mapped them by impact and effort. High-impact opportunities centered on making data visualizations easier to interpret and helping users easily navigate their way to answer specific questions they had.

Co-create solutions with data science

Partnering with the data science team, we analyzed why our metrics were hard to understand, and brainstormed ways to present a story of bias and disparity that was conceptually sensible and actionable.

Co-create solutions with data science

Partnering with the data science team, we analyzed why our metrics were hard to understand, and brainstormed ways to present a story of bias and disparity that was conceptually sensible and actionable.

Design and Validate

We translated our ideas for improved comprehension and navigation into rapid iterations—wireframing, testing with customers, and validating hi-fi designs through A/B tests on UserTesting. The results confirmed clear gains in both understanding and navigability.

Design and Validate

We translated our ideas for improved comprehension and navigation into rapid iterations—wireframing, testing with customers, and validating hi-fi designs through A/B tests on UserTesting. The results confirmed clear gains in both understanding and navigability.

Solution

A snapshot of your HR health and top opportunities

Each organization now sees a concise overview of its HR "health" — a composite of efficiency scores across hiring, compensation, performance, engagement, promotion, and retention — along with cards highlighting employee groups most affected by bias.

A walkthrough of the updated experience, starting from Home, navigating to a top opportunity, and exploring the Deep Dive insights.

A walkthrough of the updated experience, starting from Home, navigating to a top opportunity, and exploring the Deep Dive insights.

Visualizing Disparity and Bias at a Glance

We restructured the dashboard such that any metric had an associated "deep dive", which visualizes bias and disparity side by side and clarifies how they relate.

AI-powered explanations

We integrated OpenAI to generate summaries that explain what’s shown in each data visualization. These dynamic explanations use natural language to help HR leaders feel confident interpreting their data.

Outcomes

Early indicators of success

In hi-fi testing, the redesigned experience improved platform navigability by 67% and insight comprehension by 46% compared to the previous version. Participants reported feeling more confident exploring the data without assistance from data scientists.

General availability and ongoing monitoring

The MVP launched to general availability in August 2024, and the team is now tracking monthly active users and engagement with the new Deep Dive feature to validate long-term adoption and scalability.

Strategic impact

By reframing analytics into an accessible, self-serve experience, Sigma Squared freed up internal data science resources, strengthened its value proposition, and laid the foundation for sustainable growth in enterprise HR analytics.