HR meets science at Google with Prasad Setty

Summary

Prasad Setty discusses the integration of data and analytics in HR at Google, highlighting the balance between people and science in decision-making. He shares key insights from Google's People Analytics team, including the use of promotion models, Project Oxygen's impact on management, and the gDNA longitudinal study exploring work-life dynamics. Setty emphasizes that while data informs decisions, human judgment and context remain crucial. He concludes by stressing the importance of combining science with personal reflection to improve workplace decision-making without relying solely on algorithms.

Takeaways

• 🍽️ The audience had positive feedback about both the food and conversations during lunch.
• 💼 Google integrates data and analytics into all of its people decisions, including hiring, promotion, and compensation.
• 🤖 Initially, Google tried to use algorithms to make promotion decisions, but engineers preferred to make decisions themselves, valuing the discussion and decision-making process.
• 📊 The purpose of people analytics is to provide relevant information to help decision-makers, not to replace them with algorithms.
• 📈 Project Oxygen identified eight key behaviors that differentiate great managers from poor ones, and Google uses these to improve management across the company.
• 💡 Google applies research and data to improve processes like selection, onboarding, development, and organizational design, focusing on evidence from academic literature.
• 🧪 Google conducts internal research and experiments, like the longitudinal gDNA study, to understand the long-term impact of work on life and happiness.
• 📚 Google emphasizes the importance of using academic research over business consulting to guide its people-related decisions.
• 🔬 The gDNA study helps Googlers gain personal insights, and many have found it valuable for self-reflection and improvement.
• 👨‍💼 The future goal is to enable all employees to make informed decisions, using data for reflection rather than relying on algorithms to dictate outcomes.

Q & A

What was Prasad Setty's main message in his introduction?

Prasad Setty emphasized the importance of using data and analytics for people decisions at Google, while also recognizing the need to let humans make the final decisions. Analytics should provide relevant information but not replace human judgment.

Why did Google engineers resist using the promotion decision model?

Google engineers resisted the promotion decision model because they did not want to rely on a 'black box' for making important decisions. They preferred to own the decision-making process and use models as a tool to reflect on their decision-making, not to replace it.

What was the initial goal of the People Analytics team when it was first formed?

The initial goal of the People Analytics team was to base all people decisions at Google on data and analytics. This included decisions related to hiring, promotions, and compensation.

How did Google’s Project Oxygen impact management at the company?

Project Oxygen identified eight key behaviors that differentiated the best managers from the rest at Google. It led to programs for selecting and developing managers based on these behaviors, resulting in improved manager performance and team outcomes across the company.

What major learning did Prasad Setty and his team take away from the early experience with Google’s promotion model?

The key learning was that people should make people decisions, not algorithms. People Analytics shifted its focus to providing better, relevant information to decision-makers rather than automating decisions through algorithms.

What did the gDNA study aim to achieve at Google?

The gDNA study aimed to understand the long-term impact of work on people’s careers and lives. By conducting a longitudinal study, Google sought to uncover insights related to work satisfaction, happiness, and productivity over time.

What was one unexpected finding from Google’s gDNA study regarding gratitude?

The study found that Googlers who had an innate sense of gratitude maintained higher levels of happiness at work over time compared to others. Their job satisfaction did not decline as typically expected over the course of their careers.

How did People Analytics approach the use of academic research at Google?

Instead of looking at what successful organizations were doing, People Analytics prioritized academic research to inform their decisions. They combined this research with internal experiments to solve people-related challenges at Google.

What was the main reason Project Oxygen succeeded at Google?

Project Oxygen succeeded because it translated academic research into personalized, meaningful actions for Google’s managers. The research was made relevant to individual managers and their teams, which encouraged them to improve their management skills.

What role did Prasad Setty envision for analytics in decision-making?

Prasad Setty saw the role of analytics as a support tool, providing relevant data to enhance decision-making. Analytics should arm decision-makers with insights, allowing them to make informed decisions without relying solely on algorithms.

Outlines

00:00

🎤 Introduction and Audience Engagement

The speaker, Prasad Setty, begins by engaging the audience with a lighthearted survey using a thumbs-up, neutral, and thumbs-down scale to gauge their post-lunch experience and feedback on the quality of the food and conversation. He humorously acknowledges their attentiveness during the morning session and seamlessly transitions into introducing the session's theme: the intersection of HR and analytics. He outlines the importance of data-driven decision-making in people management at Google, setting the stage for the discussion on how HR and science can collaborate to transform people operations.

05:03

💡 The Evolution of People Analytics at Google

Prasad shares the early challenges faced by the people analytics team at Google. Initially, the goal was to automate people decisions using data models, but they encountered resistance from engineers during the promotion review process. While the model provided high accuracy, the engineers valued human judgment and wanted the freedom to discuss and consider critical factors that were missing from the model. This experience taught the team that their role wasn’t to replace human decisions but to enhance them by providing valuable insights and fostering reflective decision-making.

10:04

🧩 Project Oxygen and the Impact of Good Managers

Prasad introduces 'Project Oxygen,' an internal research initiative that analyzed whether managers matter at Google. Contrary to initial hypotheses, the study showed that good managers significantly improved team performance and retention. Through double-blind interviews, they identified eight attributes that distinguished the best managers from the rest. This research influenced not just hiring and training practices but also provided continuous feedback and development for managers, leading to consistent improvement across teams. The success lay in showing managers the personal relevance of these behaviors, making the findings meaningful and actionable.

15:04

🔬 Embracing Science and Data: The gDNA Study

Prasad highlights Google’s ambitious longitudinal research project, gDNA, inspired by the Framingham Heart Study, to understand the role of work in people’s lives over time. Thousands of Googlers participate in bi-annual surveys that track attitudes, behaviors, and work-life interactions. Early insights reveal patterns like engineers being less susceptible to biases and how a sense of gratitude correlates with sustained happiness. The study aims to explore the long-term impact of work on individual well-being, providing rich insights to drive both organizational and personal development.

🛠️ Balancing Data and Human Judgment

Prasad concludes by emphasizing that while Google aims to use data and analytics to inform people decisions, the goal is not to replace human judgment. Instead, Google wants to empower people to make decisions with full awareness of the data available. He stresses the importance of understanding the nuances and context of each decision, reiterating that life and work are not governed by strict algorithms. The discussion then transitions to a panel featuring experts who explore cutting-edge technology, scaling people analytics, and new HR experiments.

Keywords

💡Analytics

Analytics refers to the systematic computational analysis of data or statistics. In the context of the video, analytics is used to inform and improve human resources (HR) decisions within Google. The script mentions that Google's people analytics team aimed to base all people decisions on data and analytics, highlighting its importance in driving HR strategies and decision-making processes.

💡Decision Making Model

A decision-making model is a framework or set of rules used to support decision-making. In the script, Google developed a simple logic model to assist in promotion decisions. This model was found to be reliable and stable, illustrating how decision-making models can be applied in HR to streamline and improve the accuracy of important organizational decisions.

💡Promotion

Promotion, in a corporate context, refers to the advancement of an employee to a higher position or status within the company. The script discusses the significance of promotions at Google, which are decided through a rigorous process involving committees and reviews, emphasizing the importance of such career advancements in employee development and organizational structure.

💡People Analytics

People analytics is the application of analytical techniques to HR data to improve business and talent management decisions. The video script describes how Google's people analytics team aimed to influence all people decisions with data, showcasing the growing field of people analytics in enhancing HR practices through data-driven insights.

💡Ownership

Ownership, in the context of the video, refers to the sense of responsibility and control individuals have over their work and decisions. The script mentions that Google's engineers preferred to own the decision-making process rather than relying on a model, indicating the desire for autonomy and personal investment in the outcomes of their work.

💡Project Oxygen

Project Oxygen is a research initiative by Google to identify the qualities of effective managers. The script describes how this project revealed eight key attributes of successful managers and led to the development of programs aimed at improving managerial effectiveness, demonstrating the impact of targeted research on organizational performance.

💡gDNA

gDNA stands for Google's DNA and refers to a longitudinal study Google initiated to understand the role of work in people's careers and lives. The script explains that gDNA involves regular surveys and data collection from Google employees, aiming to provide insights into work and life happiness, and to empower individuals with self-knowledge.

💡Longitudinal Study

A longitudinal study is a type of observational study that collects data from the same subjects over a long period. The video script mentions gDNA as Google's longitudinal study, which intends to follow employees through their careers to gain insights into career development and life satisfaction, highlighting the value of long-term research in HR.

💡Disagreement

Disagreement refers to a situation where there are differing views or opinions. In the script, an audience member mentions disagreement about how to proceed as a key takeaway from the morning, suggesting that diverse viewpoints and debates are part of the process of reaching consensus and making decisions in a collaborative environment.

💡Well-being

Well-being encompasses an individual's physical health, mental state, and overall happiness. The script discusses how Google uses insights from research to improve employee well-being, indicating the company's focus on the holistic health of its employees as a key component of HR strategy.

💡Positive Psychology

Positive psychology is the scientific study of human happiness and well-being. The script refers to Google's experiments based on positive psychology literature to prime gratitude in employees, showing how principles from this field are applied to foster a positive work environment and enhance employee satisfaction.

Highlights

Introduction to the importance of science in HR decisions at Google.

The goal for Google's people decisions to be data-driven.

The realization that analytics should support, not replace, human decision-making.

The process of promotions at Google involving senior engineers in decision-making.

Development of a decision-making model for promotions with 90% accuracy.

Resistance from engineers to using a model for promotion decisions.

The value of discussing and understanding the decision-making process.

The discovery that managers at Google significantly impact team performance and retention.

Eight key behaviors of successful managers identified in Project Oxygen.

Implementation of Project Oxygen findings to improve management at Google.

The increase in manager scores across all segments at Google post-Project Oxygen.

The importance of translating academic research into actionable insights for managers.

The launch of gDNA, Google's longitudinal study on work and life.

Findings from gDNA on the impact of gratitude on job happiness.

The decision to provide individual feedback to gDNA participants.

The impact of gDNA on Googlers' self-awareness and decision-making.

The vision for empowering individuals with data and analytics in decision-making.

Introduction to the panelists and their perspectives on science and analytics in HR.

Transcripts

[MUSIC PLAYING]

PRASAD SETTY: Hello.

Good afternoon.

Welcome back from lunch.

How was the conversation at lunch?

Because this is about science, let's do a show of hands.

We'll have a three point scale.

We'll do thumbs up, neutral, down.

Food quality?

All right.

I don't see any neutrals at all.

Stacy is even standing up.

Quality of the conversation?

Whoa, even better?

What's been the biggest take-away from the morning?

And you're not allowed to say Zingerman's candy bar.

Throw it out, throw it out.

AUDIENCE: Disagreement about how to go forward.

PRASAD SETTY: Excellent, excellent.

Yes.

Yes?

AUDIENCE: Ownership.

PRASAD SETTY: Ownership, yes.

AUDIENCE: How to get happy.

PRASAD SETTY: How to get--?

AUDIENCE: Happy.

PRASAD SETTY: How to get happy.

Perfect.

It's as if you were all listening

rather than checking email.

And that's amazing.

Thank you.

With that demonstration of sophisticated survey

techniques, we now conclude our HR meets science panel.

Thank you very much.

[CHEERING]

Now, we have assembled an amazing set of panelists

this afternoon.

We wanted to really showcase not just

the beginning of how HR and science get together but what

are the frontiers of analytics and science and the HR field.

And before I get all of our panelists here on stage,

I wanted to start off and share with you

our own experience here at Google

of how we have had science and HR meet.

So seven years back, when I joined Google,

one of our first exercises was to develop a rallying

cry for our newly founded people analytics team.

And we came up with something that resonates even today

and is going to keep us busy for a long, long time.

And for those of you who can't recognize this underused font

called Prasad Setty's handwriting,

it says all people decisions at Google

should be based on data and analytics.

After all, an organization our size

makes thousands of people decisions every year.

Who should we hire?

Who should we promote?

How much should we pay our best people?

And so on and on.

And in an innovation and people based organization like ours,

these decisions are no less important

than how we think about our product strategy,

how we think about our go to market strategy,

how we think about our business plans, et cetera.

So we thought analytics should help

influence all of our people decisions.

Now, initially we aspired for living up

to a very strong form of this mission statement.

We wanted analytics to spit out people decisions.

And an early experience at Google

changed my mind on what my team should actually be working on.

And to share this learning with you,

I have to describe one of the rituals that

make Google what it is.

And this actually takes place a few miles south.

Here's the Santa Clara Marriott, seemingly just

another chain hotel.

But in reality, what it hides behind its beautiful facade

is that it's the setting for some

of the most important people decisions

we make at Google-- promotions of our software engineers.

Promotions are a big deal at Google.

Twice a year, thousands of Googlers

get promoted to positions of higher responsibility.

And in our technical site, the process includes the following.

We bring in hundreds of our senior most engineers

from all around the world to the Bay Area.

We form committees of four to five people each.

And we have each committee review a stack of nominations.

And then they make a decision after lots and lots

of conversation.

We don't have enough conference rooms at Google.

And that's why we have to use up most of the Santa Clara

Marriott.

In engineering, people can self-nominate themselves

for promotions.

There are appeals committees is in case

you disagree with the primary committee's recommendation.

There are appeals of appeals committees.

This is like the Supreme Court process,

except that our committee members are a little younger.

And we thought that, given all this effort that

goes into making these promotion decisions,

people analytics could help our engineering brothers

and sisters make these decisions more efficiently.

So we looked at all the information that

was available to the committees, and we came up

with a decision making model.

A simple logic, nothing too fancy.

This is no E equals mc squared.

There's a pretty ugly formula.

But it works.

What we found was that it was reliable, in practice

did very well.

It was stable across multiple cycles.

And for almost a third of our promotion cases,

with 90 percent accuracy, you could just

use this model to make the right decision.

We were excited.

We thought that we would go to our engineers and tell them

we can immediately cut committee workload by 30%

if we use the model to make these promotion decisions.

We thought that these people who lived

in the world of certain algorithms all day

long would love this.

They didn't like it one bit.

They didn't want to hide behind a black box.

They wanted to own the decisions.

They didn't want to use a model to do so.

The one thing that they did find interesting about this model

was actually something that it was missing.

There was one critical factor that folklore had established

was crucial in informing promotion decisions.

And when we looked at our model and built it,

that particular factor was missing.

It was nowhere to be seen.

And so that little rich conversation

among the committee members about-- What

is the importance of that factor?

Are we all using it the right way?

What role should it play in the future?

That was the biggest learning for them.

They didn't want to use the model

to make decisions for them.

They wanted to use it to examine their own decision making

process.

I took away a very different learning for our team

about that.

And this is like a Homer Simpson "duh" moment, right?

We should let people make people decisions.

People analytics wasn't going to be

in the business of developing algorithms

to substitute for or replace place human decision makers.

Instead, what we were going to be all about

was to arm these people with much better,

relevant information so that they

can be capable of making better decisions.

What is the context behind which you

supply this relevant information?

What are the characteristics of doing so?

A second project illustrated for us

what that actually looks like.

And Laszlo brought it up earlier this morning.

And this is something that we have shared extensively

externally too.

But let me give you my version of it.

A few years back, we asked ourselves,

do people managers matter at Google?

And this wasn't just a theoretical question for us.

We actually wanted to show that they didn't, because then we

could design an organization with boundless levels

of autonomy.

And selfishly for me, I wouldn't have to listen to Laszlo.

No, I love Laszlo.

I always listen to him, always do.

What we actually found was the opposite.

We failed miserably at proving that managers didn't matter.

We saw that our best managers had much better retention rates

for their teams, had much better performance for their teams.

We then dug into and did some double blind interviews

to figure out, is this just happening by chance

or is it by conscious action?

And we found that our best managers were exhibiting eight

attributes, eight behaviors consistently that our poor ones

weren't.

And this is what we codified and published externally as well

as our Project Oxygen research.

But we didn't stop at the research phase.

All of People Operations got together and came up with ways

to live up to these attributes.

So we use these dimensions when we

are filling new people manager roles.

We have learning programs to help managers improve

their capabilities on these eight dimensions.

And for the past few years, every six months,

each manager at Google has been rated but their Googlers,

by their direct reports, on these dimensions.

And they get an individualized report like this,

purely for developmental reasons.

And after all these years, by any segment that you

cut our manager population-- across functions, regions,

across their tenure as a manager,

whether they started out being our best

or our average or our poor managers--

scores have increased all over, across all segments.

Oxygen has been very successful at Google.

What we asked ourselves was, why?

It's not like we came up with a new model of management.

All of these are principles that have already

been established in the literature, right?

But in the past, we used to tell our managers, hey,

you need to focus on improving your capabilities,

because that is what leaders of successful organizations do.

Our managers didn't listen, given all the priorities

that they had.

Instead, with Oxygen, we were able to go back to them

and say, this is why it matters for you

and your team and the organization.

And this is how you can personally benefit and learn

from it.

Our research that was internal to Google

and translating that research into something

that was personally meaningful for each of these people

trumped best practice.

And for all the academics in the room--

there are many of you-- you'll be

very happy to know that whenever we are faced with a new people

issue at Google now, we don't ask ourselves,

what does successful organization x

do with this topic?

Instead we ask ourselves, what does the literature say?

And if I have one piece of advice

to give all the businesses that are in the audience out here,

it is develop better relationships with academics

than with consulting firms.

We've used this approach where we

start with what's established in the academic literature

and then do our own research and our own experiments

at Google to great success in a whole bunch of areas.

It's not just about people management-- selection,

onboarding, development of all Googlers,

not just our managers, organizational design,

performance management, well-being.

I could go on and on.

So where do we go from here?

We continue to make sure that our organization is getting

more capable at making better people decisions.

We continue to arm our leaders and managers with information

that is rooted in research and relevant within the Google

context.

But we want to broaden our net further.

A few years back, Laszlo came to us with a crazy idea.

Laszlo comes up with crazy ideas all the time.

Usually I ignore it, because they

have a half-live of a minute.

But in this particular case, he was very persistent.

One of the unwritten key characteristics of my own job

is knowing plain crazy from serious crazy.

And in this particular one, Laszlo had a strong conviction

that we should do something along the lines

of the Framingham Heart Study.

In the interest of time, I won't go into the Framingham Heart

Study.

But it's one of these longitudinal studies that

has been around since 1948 and is the source of most

of our commonly known findings around the spread of heart

disease.

We didn't want to study heart disease,

but we wanted to use that type of longitudinal technique

to see, what role does work play in people's careers

and over their lives?

And what is the meaning that it has in that?

So gDNA, our own longitudinal study, was born.

Over the last couple of years, thousands of Googlers

have enlisted and opted in to be part of gDNA.

We plan to follow them through their careers

at Google and even beyond.

We hope to run this particular study for decades.

So we are very, very early in our project life cycle.

Every six months, our gDNA panelists get a survey

from us-- asking them about personal characteristics,

attitudes, perceptions, their beliefs,

we ask them to solve some problems--

on a wide variety of topics, anything

that's been associated with either work or life happiness.

We already have some population level statistics

that we just didn't know in the past.

For instance, we know now that our engineers

are less susceptible to some cost biases

than our sales Googlers.

We know that people who are better networked at Google

have better performance and higher promotion rates.

To Laszlo's point about the Office Space clip,

we know that people who have an innate sense of gratitude

at Google stay happier with work for a much longer period

in their careers.

Usually in organizations, what you'll see

is that people are happy at the beginning of their tenure

and then it slowly decays.

We tell our Nooglers-- our new Googlers-- savor this moment,

because this is the happiest you'll ever be at Google.

But when you cut the data by gratitude,

you find that those who have an innate sense of gratitude--

and this is what Sean was talking about too--

their scores stay stable over time.

And so we are now using this information

to do experiments, based on what's

been done in the positive psychology literature,

to see if we can prime gratitude in even those who don't perhaps

have an innate sense of it.

But more important than all these population

level stats is how Googlers have individually responded to gDNA.

One of the decisions we made about gDNA

was that we would send back information

to each Googler who took part in the surveys.

We would tell them how they scored in these dimensions,

how they related to Google at large.

We gave them links and resources to learn more

about these theoretical constructs

and how they could perhaps take action.

And in response, thousands of Googlers have come back to us

and said, this is what I've done because of gDNA.

I look forward to the survey.

It's fun.

It's engaging.

It helps me with time for introspection.

And I don't have time to go through all of these.

But as you quickly scan through them,

you can see the depth of reflection,

the richness of perspective, the commitment to action,

just based on taking a survey.

I hope this gives you a glimpse of where we are headed.

We want to empower every single person

to know much more about themselves,

be more aware about themselves as well

as the surroundings around them.

We want to not tell people what to do.

But we want them to be capable of all people making decisions

based on data and analytics.

And when I say that, we don't mean robotically

respond to this information-- oh, you showed me x,

and therefore, I must do y.

But instead, reflect on it and respond with all the intricacy,

the nuance, the judgment that we as humans are capable of.

We want people to be rooted in science.

But as one of my colleagues, Janet Cho likes to say,

life isn't an algorithm.

And we don't intend to make it one.

Thank you.

MODERATOR: You're going to keep going.

OK.

PRASAD SETTY: Before I walk off, I want to say-- sorry about--

I now want to introduce our panelists.

They're each going to do a little bit

of a talk about their own perspective

on science and analytics in the HR space.

And then we'll do a Q&A. But when

we assembled this amazing group--

as you'll find out soon from their stories--

as I mentioned earlier, we wanted

to see what is at the frontier of analytics and science here.

Ben is going to talk about cutting-edge technology that

is going to help us.

Paul is going to talk about, how do you scale this

to some of the biggest organizations out there?

Google at 50,000 people is a microcosm

of what the US Army is.

And Katie is going to talk about all the new research

and science and all the cool experiments

that she is running that are just going to be very, very

key for all of us HR professionals.

[MUSIC PLAYING]