Google may be the most data-rich company in the world. It owns and operates at least 15 data centers around the globe to support its products (1), and Google search alone receives 2.3M+ queries/minute (2). Data and algorithms may drive the success of Google’s products, but they also drive the success of Google’s HR function. Armed with company-supplied laptop stickers that state “I have charts and graphs to back me up. So f*** off,” Google’s people analytics team uses data to enhance the workforce component of Google’s operating model (3).
People Analytics at Google
Google’s focus on “people analytics” began when Laszlo Bock was hired as SVP of People Operations in 2006. Bock, who believed that data could unlock ways to improve the workplace (4), created a team of PhDs and ex-consultants to analyze workforce data and support the People Operations mission: “find them, grow them, keep them” (5).
Project Oxygen may be the most well-known people analytics project. After a failed “experiment” with eliminating managers in 2002, Google wanted to better understand the role of managers (6). The people analytics team encoded more than 10K observations from performance reviews/upward feedback and compared it to productivity metrics to prove the value of managers (7). Mixing this with double-blind interviews, they identified the “8 Behaviors for Great Managers at Google,” which now govern Google’s feedback and development processes (8).
Another project was the development of an algorithm–possibly built using data on current employees (resumes, productivity, etc.)–that reviewed rejected resumes for high potential candidates. Moreover, the people analytics team uses data–possibly attrition rates, encoded exit interviews, surveys, productivity, etc.–to guide decisions about benefits, such as the decision to increase maternity leave to 18 weeks.
The Value of People Analytics
Data and algorithms are a core part of Google’s DNA, so it’s no surprise that this seeps into workforce management. It’s also not surprising because of the value it brings to a company rooted in innovation.
Talent management is critical to how Google creates value: through technological innovation. Google’s use of data to recruit and develop employees has proved its effectiveness; training programs based on Project Oxygen, for instance, achieved a statistically significant improvement in manager quality for 75% of underperforming managers (7). Strong talent management also helps Google capture value through cost savings. Attrition is costly to Google, not only because they must recruit, but also because ex-Googlers accumulate valuable organizational knowledge. Analytics help Google reduce attrition in cost-effective ways; the decision to extend maternity leave reduced new mother attrition rates by 50% and “was much better” for Google’s bottom line (9). Google also uses algorithms to predict which employees may quit, which allows for an intervention that may retain the employee (10).
Challenges and Competitors
If it’s so beneficial, why aren’t all companies doing it? The answer is that it’s challenging. It requires a large amount of hiring, performance, productivity, etc. data; an all-star analytics team; and buy-in from employees. Google was better poised to tackle these challenges than many companies. In addition to its digitized application process, Google collected workforce data on a frequent basis (e.g., 2 reviews/year) and has a workforce that is fluent in data. To bring the right talent into People Operations, Google created a People Innovation Laboratory. To give them direction while an innovation process was designed, Google focused initial efforts around themed “projects.”
Looking ahead, one major challenge is the fact that its competitors are following suit. Experts believe that Google’s people analytics team was unique as late as 2009 (7). While competitors may have relied on intuition and descriptive analytics to manage their workforces, Google was using predictive/prescriptive analytics and even running experiments. A quick search, however, revealed that Facebook, Microsoft, and Amazon may now have people analytics teams using these techniques. If the competition is following suit, will Google maintain its edge in being one of the best places to work?
Another challenge for Google’s people analytics team is ensuring the proper balance of human-machine collaboration in workforce management. Repeated success will certainly build trust in these algorithms, but blindly accepting the output of machine learning algorithms can be dangerous. Contrary to popular belief, algorithms can exhibit biases just like humans (11). Given its expertise in designing algorithms, I imagine Google will develop best practices to avoid this risk.