Scaling Google and Yahoo with Marissa Mayer — Class 17 Notes of Stanford University’s CS183C
Here is an essay version of my class notes from Class 17 of Stanford University’s CS183C — Technology-enabled Blitzscaling — taught by Reid Hoffman, John Lilly, Chris Yeh, and Allen Blue. Errors and omissions are my own. Credit for good stuff is Reid, John, Chris Yeh, and Allen’s entirely.
This class was an interview by Chris Yeh of Marissa Mayer. Marissa joined Google in 1999 as employee number 20. She was the company’s first female engineer, who went on to be one of the original product leaders at Google, and is now the CEO of Yahoo!.
Video of the class, notes are below:
Chris Yeh: You yourself went to Stanford and graduated in 1997. Can you tell us about your time at Stanford and what you learned while studying symbolic systems?
Marissa Mayer: I loved Stanford and Symbolic systems. For me, I came to Stanford assuming I would be a doctor and got really deep into chemistry and biology but I noticed everyone who was on the same track as me was taking the exact same classes. I wanted to do something more unique.
Stanford had a strong computer science program and psychology program and I ended up switching to symbolic systems. One of the interesting applications of Symbolic systems is artificial intelligence and I spent some time thinking about how to create a brain that operates the way ours does.
Chris Yeh: Outside of school, what else were you involved in?
Marissa Mayer: I was an RCC, RA, and did a lot of work in our residences. I also was in the debate team (which I also did in high school), and just really enjoyed my time here. There were other activities as well, but those were the ones that stood out.
Chris Yeh: When you finished your Masters you went to go work for a small company (Google) — how big was it when you joined?
Marissa Mayer: Before Google, I spent the summer building a program that would look at what websites you would go to and what websites other people would go to — and built a collaborative filtering program that helped you find related sites to look at.
When I came back, Eric Roberts hired me to teach a class (CS 106b) for the first time. He asked me about my summer research program and he told me I should go talk to these two guys named — Larry and Sergey — who were working on a similar problem but taking a different approach via link structures. At this time, Google had just incorporated a week before then, so they weren’t looking to hire yet and I was too busy to do a startup.
Fast forward to Spring, I got an email from one of the first employees at Google who asked me if I would consider working there. I had procrastinated on picking the company I wanted to join but I said I would interview with them.
On the day of my interview, everyone left because Larry and Sergey went to go pitch Kleiner Perkins and the team was small enough that everyone went with them. The office manager told me to come back tomorrow because there was no one to interview me.
This is the now famous VC pitch that Google gave in which they raised $25M — split between Kleiner Perkins and Sequoia Capital.
Chris Yeh: You joined Google as an engineer, but overtime you became a product person, what was your evolution there?
Marissa Mayer: It was very gradual. I came in as an engineer and worked on artificial intelligence at Google. I worked on related sites and matching advertising to queries with some of our earliest ads.
Along the way, we realized we had a hard time finding someone to make user interface (UI) changes. We kept interviewing and looking for a strong systems engineer who could also develop on the UI but it was impossible to find this within one person.
After months of trying to find the UI/systems engineer, my boss pulled me aside and said I had an interesting background in symbolic systems — and to take a day out of my week to work on UI changes.
Every week, I would meet with the team and figure out — when we would roll out new features, what that feature needed to do, how the team wanted it to look like, what the data was telling me — and I would turn all of this into a product and code the front end of the feature.
Two to three years into doing this, we looked at all of the other successful tech companies at the time and found they all had this role called “product management” — but Google didn’t have this yet. Even though they didn’t have this function Larry and Sergey noticed there were a few people doing this unofficially — Salar, Susan, and myself — even though it wasn’t in our job description. We started the product management function within Google.
For me, the move to product felt more gradual because I had already been doing this type of work already even when I was doing engineering.
Chris Yeh: This ties in with something we talk about in this class, in that in the early stage, it’s important to bring in generalists that will do everything that needs to be done.
What were some of the things you noticed where you needed to change your process because you were scaling so quickly?
Marissa Mayer: Eric Schmidt talked about this a lot at Google. One of the things he said was, at every order of magnitude (1, 10, 100, 1,000, etc) — expect every process to break. You would need to change the process each time at each level of scale.
I saw this happen all of the time with all of our processes: our hiring process, code deployment, etc. We would go one step further and everything would break.
Another good lesson about scale was when Eric joined, we were 200 people and our plan was to grow to 400 people at the end of the year. Eric showed up and said there was no way Google could double and keep our quality and culture at our high bar. He made the decision to only let the company hire 50 people in total for the whole year (down by 4x) and enforced this using these things called “Larry and Sergey dollars.”
Eric literally had laminated bills printed out and when you hired someone you would have to hand in a Larry and Sergey dollar along with their paperwork. There became this huge black market for Larry and Sergey dollars which actually made things more efficient. For example, if a sales person needed something done by engineer — they could now negotiate using Larry and Sergey dollars to make sure the work got done.
As painful as it was, it made us much more thoughtful on how we scaled in the early days on prioritizing which was the most important people to add. Hypergrowth is fun but you have to be careful.
Chris Yeh: One of the most successful programs at Google was the Associate Product Manager (APM) program which you started. Can you talk about where the program came from and what made it so successful?
Marissa Mayer: In 2002, Jonathan Rosenberg joined (8 months into the product management function being created) — he was hired as the VP of Product Management and became the boss of Salar, Susan, and I.
We worked together to hire more product managers and we had this interesting issue where our PM’s needed to be deeply technical. After 4 months we hired just 2 product managers. At the same time the head of engineering hired 8 engineers a week.
As a company, you want to scale both engineering and product at a certain ratio. We have found that between 8 engineers / 1 product manager to 12 engineers / 1 product managers is a good ratio. We just couldn’t keep pace.
At the same time the two product managers who we did hire went to support Salar and Susan, so I didn’t have a junior PM to work with. Instead of complaining, I made a bet with Jonathan that instead of going through him, I could hire better product managers than he could — and I could do this any way I wanted to.
My approach to win the bet was to create the APM program where we would: hire great symbolic systems/CS students right out of college, make it a rotational program, build a strong community of peers, and teach the product management process.
One of the most important factors was the rotational aspect. Given we were hiring these APM’s right out of college many of them didn’t know what areas they wanted to do. Instead of hoping they got assigned to the right team, we would have them try out all different kinds of product management — enterprise products, advertising products, partner products, consumer programs. With Google, they could stay at the same company and try out many different roles for 2 years then at the end they could pick what they want to do.
Question from the audience: What personal traits did you look for in the first class of APM’s?
Marissa Mayer: We looked for people who were technically excellent but more importantly we looked for people who understood how to apply technology — not just people who knew how to code.
We designed questions to test for this such as “what is the coolest thing you have seen in the past 6 months and how will that affect the market”?
The other tough part is, we had to acknowledge that we were hiring someone who was taking on a leadership role who had no previous experience. We needed to find someone who was humble and also a great listener.
The humble piece was important because these APM’s had to win the respect of the engineers on their team. The engineers on the team might be 15+ years the APM’s senior, so you can’t just go in and run the meeting — the right way to win them over is: let me take notes, let me do scheduling for you, let me get the machines you need — by doing all of the hard work and the process work.
Our APM’s had to be willing to roll up their sleeves and work with the process, listening to engineers and winning the team over by helping the team get organized. Another aspect is while making decisions you can’t point to your previous experience because you don’t have it — so you need to be very data driven. The way to win over engineers is to say I read all of the logs and have seen that our users do x and then do y and we need to make this easier.
Chris Yeh: Do you feel programs like the APM program are special to Silicon Valley or could they be applied more broadly?
Marissa Mayer: We actually based the APM program on other rotational programs. We looked at Viacom which had this special rotational program across their subsidiaries and McKinsey which has a 2 year analyst program.
We drew on the notion of giving broad exposure, based on training, all while giving the APM’s real products they could work on as they learned on the job.
Chris Yeh: Moving on from Google, you are now the CEO of Yahoo!. What are some of the things/people you learned from — Eric, Sergey, and Larry — which you apply today?
Marissa Mayer: I was very lucky to have great mentors being at Google for 13 years. One thing to keep in mind was, I joined Yahoo! at a very tumultuous time — I was the 7th CEO in 61 months.
One of the big things I learned from Eric was, he said — executives confuse themselves when they think they actually get to do things. CEO’s don’t code, they don’t build product, they don’t design things, etc. Instead CEO’s set direction — their job is defense and to remove things in the way from their team. CEO’s listen and help clear the path to make their team and company more effective.
On my second day on the job, an employee at Yahoo asked me “when will you present my grand vision for Yahoo?” Every CEO who had come in before me between day 2–5 would have a big meeting to present their big plans for the company.
I told him and the rest of the company that I wasn’t going to do this “big dog and pony show”. Instead I was going to go into the cafeteria and listen to you and learn about the ideas that have worked, what hasn’t worked, and what ideas other people had. I said we would do a big strategy meeting the next year but I needed help from the existing team to shape the strategy. By doing this, it was a much more comfortable way for the organization to accept a new CEO.
Chris Yeh: What kinds of questions did you ask to get useful information when you first joined?
Marissa Mayer: There were a lot of people who were eager to bring their viewpoints to me and share their ideas.
One day I was in the cafeteria and this guy asked me “Is it time to go?”
I responded “no no no, I just got here, please give me a chance.”
He responded “no, not leave the company, look I’ve been here for 5 years and I have all of these ideas, but the board hasn’t given us any direction — is it time to go and get started?”
I told him by all means to go ahead and get started. Then I started to dig into why people didn’t feel empowered to do things and what was getting in their way.
One of the things we started was called PB&J — process, bureaucracy, and jams. It was a way for employees to tell us what was getting in their way and people could vote it up or down to see if this was actually a problem. People told me there were 1000’s of things that need to be fixed and there was no way we’d be able to address them all.
There were small things we fixed:
- There were parking gates inside the parking lot — no one knew why they were there so we took them down
- Employee badges didn’t open the gym door — unless they took a 1 hr orientation class — we changed that.
There were a lot of small symbolic things where if we got 50 votes on PB&J we would change it. There were bigger things as well such as the product shipping process — it was hard to get code into production and the existing process was very cumbersome.
In the first year with PB&J we changed 1,000 things across the whole company. We found people who were passionate across each department and empowered everyone to report problems, fix problems, and prioritize problems through the voting system.
Chris Yeh: It’s taking a community based approach — something we’ve talked a lot about in this class — to fixing organizational problems. It sounds like you were also changing the tone of the company — can you talk about some of the challenges in turning a large battleship like Yahoo to a new direction?
Marissa Mayer: There is a view of culture that I learned from Brett Taylor one of the early APM’s who became the CTO of Facebook. I asked him what was different about working at Google vs. Facebook.
His response was — he couldn’t really articulate, but both companies had a very strong culture. If you were at Google, even if you took away all of the logo’s and all of the company info — you would still know you were at Google. Same thing with Facebook.
Given my medical background I think of it in terms of DNA. With genes you can get certain genes to over-express or you can turn off certain genes — but it’s very hard to inject new mutant DNA. Culture is very much the DNA of the company.
When I came to Yahoo, it was important that we didn’t change Yahoo into something else. I wanted to make the best version of Yahoo and have the best traits of Yahoo to hyper-express. Then for the thing that weren’t working — I wanted to turn those off. This is why PB&J worked so well because it helped express the good traits that were already there.
Question from the audience: What were the road blocks for shipping and deploying new products?
Marissa Mayer: There were many things. One is we had many teams which were under-resourced. One thing we implemented was a tech counsel comprised of our 12 most senior engineer and architects at Yahoo and have them help show us how to do continuous integration, quick deployment, regression testing, and generally how to make new releases smoother. We didn’t have many of these processes in place when I joined so we tried to institute the best practices.
Another thing was, people didn’t know what we were shipping and when we were shipping them. So we implemented a launch calendar (similar to Google) which was helpful in letting other departments in the company know what we were doing — PR, legal, support, marketing, etc. This helped everyone in the company know what we were shipping.
Chris Yeh: How did you work with your management team when you made these changes?
Marissa Mayer: I really liked the way Eric ran his staff at Google:
- Staff meetings on Monday — what we did in the previous week, what we are doing this week, a way to get cross functional teams working together.
- Strategy reviews on Tuesday and Wednesday — deeper dive into specific product launches and operations.
- One-on-ones on Thursday
- Full company meeting on Friday — a way for anyone in the company to ask anything and understand our thought process.
After I said I’m not doing the “dog and pony show” of laying out the vision for Yahoo I had asked — When do I get to talk with all of the employees at Yahoo?
They responded: I would get to talk with all of the employees at the quarterly all hands meeting.
I responded: Yes that’s where we share all of our metrics — but when do I get to talk about product with all of the employees?
They responded: At the quarterly all hands meeting.
I responded: Oh I see. I get it now.
When I was hanging out in the cafeteria many people came up to me and said “I’m surprised you’re here, no one from management ever talks to us.” Before I joined, there were only 4 meetings a year where executives would go out and talk with employees — now we do this on a weekly basis.
We do deep dives on new products and talk about current events which are affecting the company. Overall it’s been a great communication tool, community building exercise, and it has helped us to bring transparency to the company. One of the things that was important to me was to demystify management and the decisions we make. Now when we have a board meeting, we show all of the board slides to the company beforehand. We try to show as much of the decision making process as we can in those meetings.
Question from the audience: For the strategy sessions on Tuesday and Wednesday — what do those actually look like?
Marissa Mayer: These strategy meetings tend to be free-form time. If we are talking about a new product we would talk about the metric of success, what legal concerns this new feature has, support needs, a thorough look at each part of the product, etc. This would look different if we were talking about a new benefit structure we were rolling out — so these meetings change quite a bit.
In general they are 1 hour long sessions for exec members and key company leaders to come with a topic where they need help with a decision, we look through the data with them, and the outcome is a decision we can make for that topic.
Question from the audience: One of the things you did was ban working from home, how is that not a cultural change?
Marissa Mayer: I have nothing against working from home — in fact my brother works from home. We made this change based on listening to what employees in Yahoo were saying.
Many people had said that I loved all of the changes we were making but we had this one team member who wouldn’t come into the office so we would have to spend a lot of time trying to get them caught up. These sorts of actions really let the team down.
This is generally easier when people formally work from home or remotely but in Yahoo before I joined, people would say “oh it’s raining really hard, I don’t want to come into the office” — and then email their team about this on the day of.
We grant exceptions to this, but in general we wanted to send a mission that at this point in time in Yahoo’s history — while we are going through so much change — it’s not a good idea to work from home.
Question from the audience: In terms of culture and hiring, one of the things that matters is the people you hire. How is this different from starting a company and being able to choose everyone vs. taking over an existing company with people already there?
Marissa Mayer: We did change much of our overall hiring practice.
Before I joined, managers could hire whomever they wanted. On top of that, there were no official performance reviews — bonuses and raises were based purely on feeling. There were many cases of managers hiring their friends and giving them promotions.
In terms of performance one of the biggest drivers was — if the company isn’t performing great — we should measure the performance of the employees within the company.
So we set up a goal system, a performance system, a hiring and review process, etc. We didn’t want to change the ability for managers to hire but if we are hiring a great person — management should be able to justify why they are hiring that person.
Moreover, companies are very much living and changing organisms and people come and go — so from this perspective you can change culture. Now Yahoo is 10,700 people and of those, 6,000 are newer than I am. Greater than 50% of the company is new and has gone through all of the new hiring process.
Chris Yeh: One of the things you have done at Yahoo is acquisitions (M&A), can you talk about what has worked and what hasn’t with acquisitions?
Marissa Mayer: We’ve done several dozen acquisitions and classify them into three groups:
- Talent acquisitions
- Building block acquisitions
- Strategic acquisitions
Talent acquisitions — One of the interesting things that has worked well for us is we could bring in really terrific people and group hire 4–5 people in small sets through talent acquisitions. The nice thing is these people were already working as a team and they could hit the ground running fast.
When I first joined a big pitch to the board was to double down on mobile. I was glad to hear we had an existing mobile team but I was shocked to find out the whole mobile team at the time was 30 out of 14,000 people. The mobile team needed to be more like 500 people — not 30. We acquired a lot of teams which have helped us to reinvent our app and mobile strategy.
Building block acquisitions — Yahoo is a 20 year old company, so there is a lot of code that is 10+ years old and the original contributors who understand that code are no longer there. Some of our acquisitions such as Xobni fit this building block type of acquisition.
What Xobni did very well was contact management — they would parse all of your emails to know what has changed (phone number, address, email, etc) and update the records dynamically and keep everything in sync. We brought this technology in to replace the address contact book within Yahoo Mail and now we have a better technology that is easier to maintain and better.
Strategic acquisitions — These are ones that push us into new directions. We’ve done a few that fit within this category: Tumblr pushed us into social, Brightroll pushed us into new ad formats, and Flurry pushed us into mobile analytics and monetization.
The key strategic areas for Yahoo are mobile, video, native advertising, and social. Strategic acquisitions push us forward in one or more of these areas for us.
Chris Yeh: Given it’s very busy to be a CEO how do you invest in yourself? — Do you have a coach, read, meditate, or what do you do?
Marissa Mayer: I do all kinds of things outside of work. I spend a lot of time with my 3 year old and I love to ski and travel.
One question I ask is “what is your rhythm?”
At Google, people on the outside think the company magically was successful, but everyone in the company worked 100–130+ hours a week. In order to do this, people had to sacrifice a lot of other activities, work weekends, etc.
One thing I noticed was all of the best people at Google had an activity they would do regularly that kept them sane. I called this “finding your rhythm” and I would ask people “what is their rhythm? — What do you need in order to not feel resentful?”
For some people it would be: on Tuesday night they would have dinner with their friends — and if they missed this they would be resentful all week. For another, who was on my team, she was a mother of three and had to do calls with Bangalor each night — for her it was ok to miss nights but she couldn’t miss piano recitals or soccer practices.
For these kinds of activities we would make sure these individuals could do the one thing they needed to do that week — no matter what was going on at Google. It wasn’t so much about work life balance but more about keeping a rhythm and knowing what was most important to each person.
For myself, I noticed my rhythm change. In my 20’s I loved traveling and I wanted to travel to a new location once every 4 weeks. If I had to cancel one of these trips I would feel really resentful for the rest of the time. Now it’s much more about having family time. The things that matter to you will change over time — you can’t be too structured about it.
Question from the audience: What are the tradeoffs of working at a small company vs. big company?
Marissa Mayer: I loved the perspective my mom had on this. She would tell my brother and I that she loved every phase of our life — when we were babies, toddlers, children, teenages, in college, adults, etc. Each phase was more fun than the last and she had a real enthusiasm for us. I always thought this was special when my mom said that to us.
For me looking at my career — I loved Google when we were 20 people, I loved it when we were 100’s of people, I loved it when we were 1,000’s, I loved being an individual contributor, I loved managing a single digit team, I loved managing multiple teams, etc.
I think it’s important to try small, try big, try different functions, try different teams, etc. I think it’s good to experience a bunch of variety to understand what you like and what you are good at.
Question from the audience: With decision making, how do you know when you are making the right decision? Has anything been helpful in helping with decisions?
Marissa Mayer: In 1999 when I was looking for a job, it was a crazy time in Silicon Valley. Google was my 14th offer and I had a hard time deciding. I had offers from startups, to teach, in management consultant, etc.
I made a list of all of the best decisions I made (going to Stanford, switching to symbolic systems, working at SRI, and working at UBS’s research lab) and tried to understand what each of these decisions had in common. For me the best decision were ones that:
- Made me work with the smartest people I could find — Smart people challenge you, make you think differently, make you justify your decisions, and ups your game.
- Did things that I wasn’t quite ready for yet, which pushed me further than where I was today
When you do things you aren’t entirely ready to do — you find out a lot about yourself. I’ve always surprised myself and learned about things I was good at which I didn’t realize I would be good at. These two things helped guide me to my decision to join both Google and Yahoo.