Eric Schmidt talks about what worked and what didn’t while scaling up Google – Part II

Note: Since this is a long session, we have divided it into two blog posts. Link to the first post here.

A group of entrepreneurs taught a class at Stanford called “Technology-enabled Blitzscaling” and this is Freshteam’s attempt to present the key learnings/takeaways based on the lectures. 

In this class, Reid Hoffman, Executive Chairman and co-founder of LinkedIn, interviews Eric Schmidt, who was the CEO of Google from 2001 to 2011, its Executive Chairman from 2011 to 2015 and the Executive Chairman of Alphabet from 2015 to 2017. The interview is focused on Google’s scaling journey.

What worked and what didn’t during Google’s scaling

Reid: When an organization is growing so fast, it is a nightmare to figure out how to manage and how you should run it. What are the things that worked and what were those that didn’t?

Eric: Larry and Sergey were running things. It was chaotic. We would have a meeting called “60 minutes” on Mondays, which always lasted for two hours. You had to be offline and if you were found typing on your Blackberries, you had to pay a 25-cent fine. This strategy eventually failed as we found that it was impossible to to get people off their computers for one hour per week.

On Tuesdays, we had something called “Google Product Strategy Meeting”, which was a review process of all the products we were launching because no one knew about these. The rule was you couldn’t announce a product without going through the launch review a week before. The problem was that engineers would do something, which the support people or the lawyers weren’t aware of.

The best example for this was when a young person in his 20s walked in and put up this demo of a product where he had added the ability to geolocate where your friends were and to predict where you and your friends were going to be. 

My face went ashen and the lawyers and the marketing people went, “Oh my God!” Larry and Sergey loved it. And, I pointed out that there’s a minor problem of predictive behavior being illegal. Even if it wasn’t illegal, I didn’t want to deal with all the real-time subpoenas that were going to come in because of this. I said we couldn’t even afford lawyers to deal with the current subpoenas.

Just to be clear, Larry and Sergey were doing this for effect. They were just trying to be disruptive and have a good time. This is part of the culture – to draw it out and be provocative in a good way. I’m saying this in a good way and I think it was a very healthy process. We finally launched this product called ‘Latitude’ where users could lie about their location that would make the legal records unreliable.

Another example was we had launched Chrome and it was beginning to be successful. We were debating over HTML5 and the exclusivity of browsers and the rate at which we add features to the browsers. At that time, Chrome was well behind Safari and Internet Explorer. Since it was clear the three of us were not seeing eye to eye on this and we had a rule to not have any huge arguments in public, we sent everybody out of the room. The three of us had it out.

I told them to go back to their offices and talk it out. If they didn’t have a decision by noon the next day, I said I’d make the decision, which was clearly the worst option from their perspective. The next day they had come up with a much better solution that was far better than all the previous ideas.

In yet another instance, we had this engineer who had invented this free Wi-Fi. Nikesh Arora, who has since left the company, started asking the engineer a lot of reasonable business questions, none of which he could answer. The review was only getting worse and it got to a point where we had to put an end to it. 

We had two choices of what to tell the person: 1) “You’re an idiot for having brought such a terrible idea. You’ve clearly embarrassed yourself and the company” and so forth (the wrong thing to say at Google). 2) “I think this is the most incredible idea we’ve ever heard. You need to come up with a better approach. The objective of free Wi-Fi is incredibly strategic to the company”. Six months later, we re-introduced it and got a huge benefit out of it. 

Google’s Disorganization

Reid: There are a bunch of stories where one day, everyone worked for Larry and another day, it was like a flat organization. How did you manage people organization?

Eric: I have to tell you about the dis-organization. In the first year, I’m a normal engineering manager; we have five engineering directors and we’re reviewing the plans. Wayne Rosing, the engineering vice president, walks in and says that Larry and Sergey are on a rampage.

We had something called Snippets where the engineers had to write down what they did every week. Larry had been reading the Snippets, which wasn’t correlating with what the managers had been saying. He said we needed to get rid of all the managers and I said he couldn’t do that. However, the five managers became individual contributors and all of them were still in the company (in 2015).

Ten years later, we had 120 people working for one engineer vice president. We ran it this way for two years. Bill Cohen, a successful venture capitalist, jokes about what it was like, but the lesson here is you can do extraordinary things in these situations.

CEO’s role in Blitzscaling phase

Reid: In this kind of Blitzscaling circumstance, what’s the role of the CEO?

Eric: I can tell you my role was to manage the chaos. You have different kinds of CEOs and there’s no single answer. In any successful company, you have to have somebody who can run very fast, has very good product sense and has the intellectual emotional leadership of the key stakeholders.

In that sense, it was like a faculty where the key engineering talent sort of puts up with the management and my job was to organize the world around them. When Larry and Sergey hired me, the former told me that they didn’t need me then but that they would need him in the future, which was roughly correct.

As a small company, they were doing fine but in order to scale, they needed someone who understood replication and who could put systems in place. I set out to hire non-essential executives, which was everyone else. The founders worked on the technical side of things while I managed all the other functions.

There are other models where you have a very tactical founder who won’t give up control and cedes it partially to a business associate. Mark Zuckerberg, who won’t give up control of Facebook, is a good example of that. By the way, Larry and Sergey didn’t either. However, Mark allows the key stakeholders (in his case, the CFO and Sheryl Sandberg) to run large parts of the company.

Reid: One of our theses in Blitzscaling is that part of what you (CEO) do is that you also make decisions on what not to do in order to move fast. Were there any such key decisions that you took?

Eric: Surprisingly, no. Since the ambition was so broad, the only lever I had was to slow some things down and emphasize other things. I don’t believe you should narrow your focus. What I believe is that you get the best outcome when you make the broadest appeal in terms of leadership and excitement and we worked hard to make that happen.

We would refuse to do exclusive deals with anybody and tell them that we had the capacity to work with one and that was with you. I’d be careful to conclude that you should do a small thing. All success starts with doing one thing really, really well but you’ll recruit better with a broader vision that’s credible and which you can articulate. If you can’t sell a dream, you’re not going to be successful.

Innovating while scaling

Reid: What were the key things that kept Google innovating intelligently as it scaled?

Eric: There were actually two interesting rules when I joined: one, “Don’t be evil” and two, the 20% rule. I thought the first one was a joke because nobody wants to be evil. Once, I was part of a conversation on ad targeting and about a particular ad click in the conference room in the small building when we started. One of the engineers pounds the table and says “That would be evil”. I was like – “What just happened?!”

The whole mood of the room changed. There was a huge debate on whether or not that was evil and finally, they didn’t make that change. The analogy here was the Japanese Kanban system where any employee can stop the production line if they see a poor quality product. It works because it’s a shared value.

Strong shared values and a strong buy-in for principles take you far. I think that helped us.

It may or may not be true that the company still has the feeling but it’s definitely a part of its history.

The 20% time-rule was that we would ask an employee to do everything they can in 80% of their time and the rest of the time, they could do whatever they were passionate about. Many of the features we offer today began as explorations in employees’ 20% time. Part of the management job was to aggregate the things done in the 20% time.

Marissa Mayer, founder of Lumi Labs who held executive leadership positions in both Google and Yahoo!, maintained a top-100 list and tried to get these ideas to talk to each other.

Google’s way of grooming managers

Reid: Marissa is a good example. How did you end up growing managers?

Eric: I would argue, immodestly, that Google’s Associate Product Manager (APM) program is the largest source of entrepreneurs in the Bay Area today. Marissa, who did her undergraduation in Symbolic Systems at Stanford, was one of the three product managers when I started at Google. 

She had this idea that we should hire people right out of college who had technical degrees, particularly Computer Science, who did not themselves want to be programmers but wanted to work with them. She identified them and trained them and they would go on trips for weeks and sleep four in a hotel room which forged incredible bonds among these people. All the executives running key parts of Google came in through this program and Sundar Pichai is an example of this.

Though she was too young to understand it, this program was originally invented by Microsoft. It was called Technical Leaders. We give Microsoft the credit for this program and to Marissa for inventing this scalable model at Google.

Another story related to the APM was Marissa worked for Jonathan who said we needed a chain gang, a bunch of people waiting for work and who’d be assigned projects. We wrote down a list of questions we had about our products and businesses and all of a sudden, we had answers to every question we had about competitors, positioning, evolution and costs, etc., through the chain gang. These were 23-year-olds who were doing this as part of their development learning and such models are scalable. 

Another thing we did happened after we hired somebody called Shona Brown, formerly with McKinsey, and gave her the title of “Chief Business Operations” to manage all the functions we didn’t know what to do with. She said we should hire people from McKinsey and get them to do the McKinsey function at Google.

The idea was that employees would study the business they would go into later, which would then become their career paths. One of the earlier members of the program went on to run half of YouTube later. Dennis Woodside studied salesforce effectiveness, we put him in charge of the emerging world and later, all of Europe. Later, we put him in charge of Motorola and he left to become the chief operating officer at Dropbox.

The people who went through the program have business analytical skills and we’d take one of these people whenever we had any problems involving business units and how the businesses interact. They are an eternal stable of talent who would inevitably fix the problem.

Counterintuitive signs of talent and potential

Audience Member 2: What are the most counterintuitive signs of talent and potential at a very young age?

Eric: I would describe somebody called Noam who went to the math department at Berkeley University at 19, got bored and applied to Google but didn’t meet any of our criteria. Somebody mentioned we should talk to him. He had figured out a way to do spelling correction. Somebody suggested we should hire him to do spelling corrections and we did. He invented the spelling corrector, the first time it has ever been done.

So, we have another rule, which is if the person is super smart, hire them anyway but they’ve got to be able to invent something like a new spelling corrector.

Six months later, I’m at the cafeteria getting my dinner on a Friday evening and Noam who was next to me says he needs 10,000 machines immediately, which was a lot then. I asked him what he was going to do with them and he said he was going “to solve general knowledge by the weekend”. I authorized him to get as many machines he wanted.

 Later, Noam realized he didn’t need more than 2,000 machines and I was saved. Today, Noam is trying to solve general AI but if there’s anybody I can think of in the world who’s likely to do it, it’d be him. The way to handle strange/odd people is to ask them what they’re going to work on and give them a shot if they have a path.

We try to hire people who can coexist with other human beings because we tend to work in teams but there are cases where we’ve hired them and kept them separately. Usually, the measure if they should be so valuable that you’re willing to put all infrastructure around them. 

One of the things that Jonathan and I say in the book is hire the divas. Management text books say don’t hire the divas because they’re a pain in the ass (they are) but they are the ones that will drive the culture of excellence and they’ll drive you to excellence. Steve Jobs was a diva. Bill Joy, my colleague for years, is a diva and I mean this in the most flattering way.

They expect a lot, they drive people hard, they’re controversial and they care passionately about things. If you find those people, you’re probably going to work for one of them. So, be nice to them. Pay attention or be that passionate.

Role of separate entities in a big organization

Reid: Back to innovation, how important is it to have separable groups to accomplish something big within an existing company? You know, the Android group, Google X…

Eric: This problem has never been solved. It’s just a mess. The correct answer is two people consisting of a graduate student and a faculty member at Stanford and they go off and change the world. Every project I’ve been associated with over the last 40 years has always started with two people: one, roughly the equivalent of a graduate student and two, roughly an assistant professor who’s working for tenure. 

Some of the examples for this are Windows (was started by one person whom I know), Unix was two people, Java was by one person, Gmail was started by Paul Buchheit and his colleague, Android by a very small team and Linux by Linus Torvalds

One day, I wanted to meet the Google docs team and realized it consisted of 150 people. The reason for such a large team is because it requires a lot more research because of internationalization, integration and so forth. 

One of the complaints I have is that teams that are working on products are far larger than they should be and that’s a failure of the architecture. When you have that many programmers working, it means they don’t have the right libraries to depend on and they’ve not generalized the phenomena. My hope is that machine learning will fix that problem.

Reid: Is there anything you think is important about separability so you don’t have to interface with a cross-coordination with the rest of the company?

Eric: Computer scientists understand the concept of interface and AI. One of the problems we have internally is that our system is incredibly powerful but so co-mingled with itself because of its co-dependence. I never want to do a platform product without an appropriate end-user app that goes with it where I want a well-designed interface between the two. If you can maintain that discipline, you can build great things.

For example, the vast majority of the startups are likely to use Amazon Web Services (AWS) and its associated features. What’s unique about AWS is that they’re programmable by mere mortals as they are straightforward and intuitive. There are limitations to what it can do but it represents good design.

Scaling up separate entities

John Lilly: How would you view scaling all this up? If you go by organizational design, is this Alphabet?

Eric: I think Google has done particularly well because its two founders are so strong-minded, technical and brilliant. Amazon, Facebook, Apple and Google are driving the industry and each of them either has a strong founder or culture and a very strong product focus. 

Steve Jobs told us that the problem with Google was that we had too many things going on and that his vision, which has been realized by the success of his company, is a very small number of incredibly-done products. We can debate that but the fact of the matter is that Apple’s success speaks for itself.

In our case, I think we hit some kind of mission-scale problem. We argued two years about changing the logo. I don’t really care about it but everybody else did. They said the mission was all the world’s information that we could index and use. How do you match that with cars and biology and so forth?

We had visited with Warren Buffet. Imaging the three of us with Buffett, a fascinating and impressive man in every regard, on the 14th floor of a building in Omaha. You can understand Alphabet as an attempt to build a holding company that looks like Berkshire Hathaway out of an existing large company. This has never happened in the history of America – it was a big bet.

It’s typical of Google but we announced this without actually knowing what the companies were. We’d ask employees as to how many Alphabet companies we had and what they were. However, there wasn’t any correct answer because we hadn’t announced it. No rational company would ever do that. Google’s attitude is to just get started and work it out later.

We’ve said publicly that Google’s life sciences company, Calico and Access are some of the companies that look like they’d be successful. The instantaneous situation is we’re trying to push the Alphabet companies not to be divisions but separate companies, which means severing ties and becoming autonomous. 

This is inspired by Warren Buffett’s rule that he loves to hire people who would do it whether he hired them or not. He doesn’t have to worry about them and knows they’d wake up in the morning, show up at work and put in their all. The recognition of our scale was we needed to have an organizational structure that would attract such leaders.

Factors that made Google win

Reid: If you’re giving advice to entrepreneurs who are thinking about building game-changing companies, what do you think the key things were that made Google win?

Eric: First, you have to have the right founders – who are impressive, smart, passionate and committed (this has to be their life’s work). Two, you need to have some luck. In our case, a search and the ads we developed turned out to be a goldmine of revenue. Once I figured out it wasn’t a sham and that people were actually clicking on the ads, we set out to maximize our revenue.

You maximize that by making the ads more relevant and useful. The fact that we wound up in a very high-gross margin business very early, gave us the ability to take these risks. I ask founders about how much gross margins and flexibility they have. I’ve been so lucky that I forget there are other businesses that are relatively low-margin businesses. 

The ideal business is Microsoft – a monopoly software company with hardware competitors, who are competing for good treatment by you in a growing and global industry. Another example is Uber, which spends an awful lot of time talking about how the drivers don’t work for them and that they don’t own cars. 

This is not due to legal and liability reasons. It’s because if you think of them as a software infrastructure company rather than a business that owns those things, then they have very different economics. In that sense, it’s a modern version of the franchisee model. Why does Four Seasons hotel not own buildings? It’s better to be the operator with a revenue share and let somebody else ride the real estate up and down, which is vicious if you haven’t spent time on it.


Hope you found this post useful. Feel free to reach out and we can discuss some of the points Eric spoke about in this cool class at Stanford University! 🙂