Work Rules!: Insights from Inside Google That Will Transform How You Live and Lead

by Laszlo Bock

a model that prescribed four levels of measurement in learning programs: reaction, learning, behavior, and results.

Level one—reaction—asks the student for her reaction to the training. It feels great to teach a course and get positive feedback from the students at the end of it.

But how students feel about your class tells you nothing about whether they have learned anything. Moreover, the students themselves are often unqualified to provide feedback on the quality of the course. During the class, they should be focused on learning, not on assessing whether the balance of presentation to team exercise to individual exercise is correct.

Level two—learning—assesses the change in the student’s knowledge or attitude, typically through a test or survey at the end of the program.

The drawback is that it’s hard to retain newly acquired lessons over time. Worse, if the environment you are returning to is unchanged, the new knowledge will be extinguished.

If you don’t have an opportunity to do it again, you’ll lose the new skills you acquired—and you certainly won’t refine them.

Kirkpatrick’s third level of assessment—behavior—is where his framework becomes powerful. He asks to what extent participants changed their behavior as a result of the training. A few very clever notions are embedded in this simple concept. Assessing behavioral change requires waiting for some time after the learning experience, ensuring lessons have been integrated into long-term memory, rather than hastily memorized for tomorrow’s exam and then forgotten. It also relies on sustained external validation. The ideal way of assessing behavioral change is not just to ask the student, but to ask the team around them. Seeking external perspectives both provides a more comprehensive view of the student’s behavior and subtly encourages him to assess his own performance more objectively. For instance, if you ask most salespeople how good they are, they will tell you they are among the best in the business. If you ask their clients, and tell them you will ask their clients, you’ll get a more modest, and honest, response. Finally, level four looks at the actual results of the training program. Do you sell more? Are you a better leader? Is the code you write more elegant?

It’s much more difficult to measure the impact of training on less structured jobs or more general skills. You can develop fantastically sophisticated statistical models to draw connections between training and outcomes, and at Google we often do. In fact, we often have to, just because our engineers won’t believe us otherwise! But for most organizations, there’s a shortcut. Skip the graduate-school math and just compare how identical groups perform after only one has received training.

What can be counterintuitive and frustrating about this experimental approach is that if you have a problem, you want to fix it for everyone, now. As I shared in chapter 8, managers who take our “Manager as Coach” class improve their coaching scores by 13 percent. We spent a year waiting to see if the class really had an effect, and in the meantime thousands of Googlers were not benefiting from a program that could have helped them.

But putting everyone through a solution that you think will work doesn’t mean it will. A thoughtfully designed experiment, and the patience to wait for and measure the results, will reveal reality to you. Your training program may work, or it may not. The only way to know for sure is to try it on one group and compare it to another group.

As a pragmatic matter, you can accelerate the rate of learning in your organization or team by breaking skills down into smaller components and providing prompt, specific feedback. Too many organizations try to teach skills that are too broad, too quickly. And measuring the results of training, rather than how much people liked it, will tell you very clearly (over time!) if what you’re doing is working. But we don’t just want to learn. We also delight in teaching. You have to look no further than your own family. Every parent teaches, and every child learns. And if you’re a parent, you appreciate that often the child is the teacher, and you are the one who learns.

“the best way to learn is to teach.”

Because to teach well, you really have to think about your content. You need mastery of your subject and an elegant way to convey it to someone else. But there’s a deeper reason to have employee-teachers. Giving employees the opportunity to teach gives them purpose. Even if they don’t find meaning in their regular jobs, passing on knowledge is both inspiring and inspirational. A learning organization starts with a recognition that all of us want to grow and to help others grow. Yet in many organizations, employees are taught and professionals do the teaching. Why not let people do both?

Engage in deliberate practice: Break lessons down into small, digestible pieces with clear feedback and do them again and again.

Pay Unfairly Why it’s okay to pay two people in the same job completely different amounts

if you’re hiring people who are better than yourself, most other people issues tend to sort themselves out.

When there are only a few hundred people in a company, stock is a strong motivation, [Brin] says, because everyone gets enough options to have the chance to really make a lot of money. But “at thousands, it doesn’t work that well as an incentive,” because there are so many people that the options have to be spread too thin. “And people want the chance to be really well rewarded.” Even though Google now has some 3,000 employees worldwide, he says, “I feel the compensation should be more like a startup’s [with lower salaries and more stock options]. Not entirely, because there’s significantly less risk. But more like one. We provide the upside—maybe not the identical upside, maybe a little less—and higher odds of being successful.”

We also wanted to make sure employees stayed hungry and ambitious enough to keep striving for big impact.

in addition to having all the right environmental factors and intrinsic rewards in place (our mission, a focus on transparency, a strong Googler voice in how the company operated, freedom to explore and fail and learn, physical spaces that facilitated collaboration), we fine-tuned the extrinsic rewards as well. It came down to four principles: Pay unfairly. Celebrate accomplishment, not compensation. Make it easy to spread the love. Reward thoughtful failure.

Our founders have always been generous. They believe in sharing the value the company creates with employees. As a result, it really is possible to earn, or be awarded, tremendous amounts of money at Google.

At Google, everyone is eligible for stock awards, at every level of the company and in every country. There are differences in the target award you’re eligible for, based on your job and the local market, but the biggest determinant of what you actually receive is your performance. We don’t have to include everyone, but we do. It’s good business, and it’s the right thing to do.

proven that what people think will bring them joy may not always be what does.

Pay unfairly: Your best people are better than you think, and worth more than you pay them

The best performers not only command the highest compensation, they will also deliver sustained exceptional results.

people are on average underpaid relative to their contribution early in their careers, and overpaid later in their careers.

Internal pay systems don’t move quickly enough or offer enough flexibility to pay the best people what they are actually worth. The rational thing for you to do, as an exceptional performer, is to quit.

Fairness in pay does not mean everyone at the same job level is paid the same or within 20 percent of one another. Fairness is when pay is commensurate with contribution.liv As a result, there ought to be tremendous variance in pay for individuals.

human performance actually follows a power law distribution171

We equate the average with the median, assuming that the middle performer is also the average performer. In fact, most performers are below average:

exceptional contributors perform at a level so far above that of most, that they are able to pull the average up well past the median.

“Ten percent of productivity comes from the top percentile and 26% of output derives from the top 5% of workers.” In other words, they found that the top 1 percent of workers generated ten times the average output, and the top 5 percent more than four times the average.

“Industries and organizations that rely on manual labor, have limited technology, and place strict standards for both minimum and maximum production” are places where you’ll see a normal distribution of performance. In those environments there’s little opportunity for exceptional achievement. But everywhere else, this distribution holds sway.

How many people would you trade for your very best performer? If the number is more than five, you’re probably underpaying your best person.

At Google, we do have situations where two people doing the same work can have a hundred times difference in their impact, and in their rewards.

For example, there have been situations where one person received a stock award of $10,000, and another working in the same area received $1,000,000. This isn’t the norm, but the range of rewards at almost any level can easily vary by 300 to 500 percent, and even then there is plenty of room for outliers. In fact, we have many cases where people at more “junior” levels make far more than average performers at more “senior”...

To make these kinds of extreme rewards work, you need two capabilities. One is a very clear understanding of what impact is derived from the role in question (which requires a complementary awareness of how much is due to context: Did the market move in a lucky way? How much of this was a result of a team effort or the brand of the company? Is the achievement a short- or long-term win?). Once you can assess impact, you can look at your available budget and decide what the shape of your reward curve ought to be. If the best performer is generating ten times as much impact as an average performer, they shouldn’t necessarily get ten times the reward, but I’d wager they should get at least five times the reward.180 If you’re adopting a system like this, the only way to stay within budget is to give smaller rewards to the poorer performers, or even the average ones. That won’t feel good initially, but take comfort in knowing that you’ve now given your best people a reason to stay with you, and everyone else a reason to aim higher. The other capability is having managers who understand the reward system well enough that they can explain to the recipient, and to others who might ask if...

It’s hard work to have pay ranges where someone can make two or even ten times more than someone else. But it’s much harder to watch your highest-pot...

As Sergey wrote in our 2004 founders’ letter to shareholders: We believe strongly in being generous with our greatest contributors. In too many companies, people who do great things are not justly rewarded. Sometimes, this is because profit-sharing is so broad that any one person’s reward gets averaged out with the rewards of everyone else. Other times, it’s because contributions are simply not recognized. But we intend to be different. That is why we developed the Founders’ Award program over the past quarter. The Founders’ Award is designed to give extraordinary rewards for extraordinary team accomplishments. While there’s no single yardstick for measuring achievement, a general rule of thumb is that the team accomplished something that created tremendous value for Google. The awards pay out in the form of Google Stock Units (GSUs) that vest over time. Team members receive awards based on their level of involvement and contribution, and the largest awards to individuals can reach several million dollars.… Like a small start-up, Google will provide substantial upside to our employees based on their accomplishments. But unlike a start-up, we provide a platform and an opportunity to make those accomplishments much more likely to occur.

The following year we awarded over $45 million to eleven teams.183 As crazy as it sounds, the program made Googlers less happy. We’re a technology company, and the greatest value for users is created by our technical Googlers. Most of our nontechnical staff, all of whom do exceptional things, simply don’t have the infrastructure at their disposal to touch over 1.5 billion users every single day. As we launched more and more products, the bulk of Founders’ Award recipients were therefore engineers and product managers. So right off the bat, the half of the company that weren’t in technical roles viewed Founders’ Awards as demoralizing, since they were unlikely to win. It turned out that many of the technical people didn’t view the awards as attainable either, because not every product has the same impact on the world, rolls out as quickly, or is as easily measurable.

Over time, many technical people started viewing Founders’ Awards as just a bit out of reach, reserved mainly for a handful of core product teams.

With every award, management did their best to figure out who deserved it, and invariably missed a few people. As a result, each award was accompanied by teeth-gnashing among the near-winners who were working in the right areas but just missed the necessarily arbitrary cutoff. Ah, but surely the winners were happy? Not so much. Because of the hype surrounding the program, people assumed every winner received $1 million. In truth, awards could be that high, but most were not.

And then some (though not all) of our best, most creative, most insightful technologists, who had built some of the most impactful products in our history, would realize that they were unlikely to win a Founders’ Award for the same work twice, and would immediately try to transfer to new product areas. Without meaning to, we had created an incentive system that made almost everyone in the company less happy, and even the few happy people ended up wanting to stop doing the essential, innovative work that had earned them the award in the first place!

You should absolutely provide exceptional rewards. But you should do it in a way that’s just. The error we made in our Founders’ Awards was that we were celebrating money, even though we didn’t mean to. We announced that we were going to offer “start-up-like rewards.” We told Googlers that the awards could be up to $1 million.

Compensation systems are based on imperfect information and administered by imperfect people. They will inevitably have some errors and injustice in the margins. The way we ran the program focused too much attention on the money, which then naturally led to questions of whether the process was just, and to unhappiness.

“Fairness perceptions are very powerful. They affect how people think about almost everything at work, but especially how valued they think they are, how satisfied they are with their jobs, how much they trust their supervisors, and their commitment to the organization.”

It’s essential that extreme reward systems have both distributive and procedural justice.

We decided that our public, top-down reward programs would truly be open to the entire company. Rather than just asking our technical leaders for nominations, we turned to our heads of sales, finance, public relations, and other nontechnical departments and encouraged them to nominate teams.

We also shifted these programs from providing monetary awards to experiential awards. This was a profound change for the better. People think about experiences and goods differently than about monetary awards. Cash is evaluated on a cognitive level. A cash award is valued by calculating how it compares to your current salary, or to what you could buy with it.

And because money is fungible, as often as not it’s spent on staples rather than splurging on a pair of Christian Louboutin shoes or a massage, and fades from memory. Non-cash awards, whether they are experiences (a dinner for two) or gifts (a Nexus 7 tablet), trigger an emotional response. Recipients foc...

When we surveyed Googlers about what they wanted, they unambiguously preferred cash over experiential awards by a margin of 15 percent, and revealed that they would find cash 31 percent more meaningful than experiences. Or more precisely, this is what Googlers thought would make them happiest. But as Dan Gilbert explained in his terrific book Stumbling on Happiness, we’re not very good at predicting what will make us happy, or how happy it will make us.