Checking compensation through our pay equity analyses We conducted our first pay equity analyses in 2012; we have run analyses every year since then and shared our 2016 findings last year. We completed our most recent analyses in Q4 2017; we looked for unexplained pay differences by gender and race / ethnicity and made upward pay adjustments, where necessary, before Googlers’ 2018 compensation went into effect. Based on these analyses and adjustments, we have zero statistically significant pay differences.

Here’s how we got there: At the end of our annual compensation planning process (for salary, bonus, and equity) we ran rigorous statistical analyses to check the outcomes before any amounts were final. We conducted separate ordinary least squares (OLS) regressions to check for pay equity in each job group -- a job group is made up of job family (like Software Engineer) and level (like Level 4). The OLS method allows us to account for factors that should influence pay (e.g., tenure, location, performance ratings) and look for unexplained differences in total compensation (salary, bonus, and equity) across demographic groups. Specifically, we looked for pay differences based on gender (for which we have information worldwide) and, in the U.S., by race / ethnicity.

Our analyses covered every job group with at least 30 Googlers total and at least five Googlers per demographic group for which we have data (e.g., at least five men and at least five women). These n-count minimums ensure statistical rigor (e.g., higher statistical power, narrower confidence intervals) and allowed us to include 89 percent of Googlers (n=63,153) from entry through executive levels. We did not find statistically significant pay differences for 62,925 Googlers, but did for 228 Googlers across six job groups. We therefore increased compensation for those 228 Googlers, totalling ~$270k USD, before finalizing compensation planning and paying any Googlers. The 228 Googlers included women and men across multiple countries (e.g., U.S., UK, India) and functions, as well as Black and Latinx/Hispanic Googlers in the U.S. across functions.

Setting up people processes to be fair and equitable People processes at Google are interconnected. A Googler’s level, performance ratings, and promotion status all impact pay. So we set up these processes with controls for fairness along the way: we use data to make structured decisions, we train people to avoid bias, and we take into account multiple perspectives. When a Noogler is hired, we assess them against a consistent set of criteria for their level using rubrics and then trained, objective reviewers make hiring decisions (not any single hiring manager). When Googlers receive performance ratings, there are multiple steps to support fair evaluations. And when a manager plans a Googler’s compensation, they must document any change they make to the Googler’s compensation -- even if it’s $1.

We will continue to focus on fairness in all of our people processes, and want Google to be a great place for everyone to work.

Learn more about running a pay equity analysis from this re:Work guide: Structure and check for pay equity