EquityVoice

Inspiration

Pay gaps hide in spreadsheets. Many companies don't know where theirs are, and even when they do, the hardest part comes after: someone has to sit across from an employee and talk about it. We wanted to cover the whole path, from finding the gap, to fixing it, to talking about it. That fits W@CC's "Data for Change" theme.

What it does

EquityVoice is a pay equity and transparency tool with two parts:

Equity dashboard: Analyzes payroll data and maps pay gaps by gender, race and ethnicity, and disability across teams. A heatmap and trend chart show where the gaps are, and a "Top fixes" panel suggests corrective actions with their cost and the number of people affected. Applying fixes shows the gap closing.
Compensation Conversation Simulator: An ElevenLabs AI voice agent role-plays tough pay conversations, such as an employee who has discovered a gap, so managers can practice and get feedback on handling them with empathy and transparency.

Employees can also anonymously share their role and salary range.

How we built it

We split the work four ways: analytics, backend and privacy, the frontend dashboard, and the voice simulator.

Analytics: We measure each group's adjusted pay gap relative to a comparison group in comparable roles:

where $\bar{S}g$ is the average salary of group $g$ and $\bar{S}{ref}$ is the average for the comparison group, adjusted for role, level, and tenure. [Describe your actual method and stack here.]

Frontend: A responsive app-style dashboard in plain HTML, CSS, and JavaScript, with an interactive heatmap and trend chart, built to connect to our API. [Add your backend stack.]
Voice simulator: [Describe the ElevenLabs agent, scenarios, and feedback scoring.]

Challenges we ran into

Privacy: Pay data is sensitive, and small groups can identify individuals. We added a minimum group size, so any group below the threshold is hidden.

Real data: We used synthetic payroll data for the demo rather than real salaries.
Intersectionality and disability: We included women of color and people with disabilities as groups, which needed careful handling because disability disclosure is voluntary.
Making gaps easy to read: A dense grid of percentages overwhelms people, so we designed the heatmap and fix panel to make the biggest gaps stand out.

Accomplishments that we're proud of

A full path from data to action. Most pay equity tools stop at showing the numbers. EquityVoice takes you from finding a gap, to choosing a fix, to practicing the conversation with the people affected.

A dashboard where you can watch the gap close. Selecting a fix updates the heatmap, the headline gap, the trend chart, and the cost in real time, so the impact is easy to see and explain.
Inclusive by design. We measure gaps by gender, race and ethnicity, and disability, including women of color, and made disability disclosure optional and voluntary.
Privacy built in. Groups below a minimum size are hidden, with an admin control to adjust the threshold, so no one can be singled out in the results.
An app-style experience. A responsive interface with a bottom tab bar on phones, a sidebar on desktop, light and dark modes, and a clean teal design that makes the biggest gaps stand out.
A voice agent that makes tough conversations practicable.
Building it as a team of four in a short time. We agreed on a shared data format early so everyone could build in parallel, then connected the pieces.

What we learned

Finding a pay gap is only half the problem. The data has to lead to a specific fix and a conversation people are prepared to have. We also learned how much privacy design matters when handling this kind of data.