Data methodology

How we report our data.

Data methodology

Historical numbers may differ slightly due to rounding and corrections in methodology year over year.

Some data may be intentionally redacted due to security and privacy restrictions regarding smaller n-counts. In those cases, the data is marked N/A.

In some cases, due to rounding and how we count multiracial people, the individual percentages may not add up exactly to the overall percentage.

In past Diversity Annual Reports, we assessed retention parity by comparing the attrition rates of underrepresented communities to Google-wide attrition using an internally computed index.

Starting this year, we will shift to reporting on exits representation (the demographic representation of Googlers who leave the company) and comparing exits representation of underrepresented communities at Google to their overall representation at Google. This new methodology is more transparent because it provides real percentages instead of a calculated index. Plus, it’s easier to understand and compare attrition year over year, giving us the opportunity to see our data holistically across time. If you’d like to see our 2021 attrition data using our previous methodology, it's available in the data appendix of this report.


Communities methodology

All reporting on gender, unless otherwise stated, reflects global data. Google also reports on global non-binary gender, using global self-identified data. We do not collect data where it is expressly prohibited by local law or would put our employees’ safety at risk.


All reporting on race, unless otherwise stated, reflects U.S. data. Google also reports race representation data for APAC (Asia-Pacific), Americas (non-U.S.), and Europe, the Middle East, and Africa (EMEA), using global self-identified data. In these instances, some race categories have changed to be more globally relevant. We do not collect data where it is expressly prohibited by local law or would put our employees’ safety at risk.


In our 2019 Diversity Annual Report, we began counting multiracial people as a member of all the racial categories they identify with. This system used in the report is called the “plus system” (indicated by the + sign) because multiracial people are “plussed in” to each racial category they identify with. The “+” is not used when referring to an individual or community outside of our data methodology. To see this data using U.S. government reporting categories, view our EEO-1.

For the second time, we are publishing race data outside the U.S. This data has enabled us to expand and evolve our work in response to the unique historical and cultural contexts of race and gender around the world by creating custom and tailored programming and dedicated staff.

Defining racial and ethnic categories is particularly complex. In this report, the objective is to create categories that address significant global patterns of racial and ethnic dynamics. In some instances, this data set is limited due to various government protections around the world and the desire to protect Googler confidentiality.

“Native American+” includes Native Americans, Alaska Natives, Native Hawaiians, and Other Pacific Islanders as categorized by U.S. government reporting standards.

“Americas” includes all countries in North and South America in which we operate, excluding the U.S.

“Latinx” is an umbrella term that includes all those who identify as Latinx, Latino, Latina or Hispanic.

More inclusive demographic data

At Google, we build for everyone. We know that one of the best ways to do that is to have a workforce that’s more representative of the users we serve. Thanks to an initiative called Self-ID, Google gathers global data on race, gender, and other identities to help give us a more complete picture of our workforce. This data is helping to power our diversity, equity, and inclusion (DEI) efforts globally, and helps to make everyone at Google more visible - so that we can create an even more inclusive workplace.

Of employees who have self-identified globally, we see that:

6.7% self-identified as LGBQ+ and/or Trans+.
5.4% self-identified as having a disability.
5.0% self-identified as being, or having been, members of the military.
<1% self-identified as non-binary.

In 2021, the number of Googlers who self-identify (or “self-ID”) as members of these underrepresented communities grew — but at a slower pace than our overall growth.

Transparency and data sharing

Data transparency is a critical contribution to creating systemic, industry-wide solutions. External research shows that only industry-wide systemic solutions will create sustainable change. This is why we’re making it easier for researchers, community organizations, and industry groups to access and analyze our data by publishing it in BigQuery, an open source data warehouse.

We were one of the first tech companies to start sharing our diversity data publicly in 2014, and today, we are proud to provide one of the largest publicly available DEI data sets in the industry. We believe that data transparency and standardization is an important step in service of collective action.