CS183C Session 13: Shishir Mehrotra
Our special guest was Shishir Mehrotra, who talked about scaling, and then spoke with Allen Blue. You can watch the recording here.
Scaling: 10 things that matter
Shishir went to school at MIT. Left school to start Centrata, then went to Microsoft and Google. Left Google last year to start a company (which he won’t talk about today).
Joined YouTube in 2008. Just purchased by Google, 100 employees, still not clearly a success.
Officially, responsible for Product and Operations, but ran YouTube as a triumvirate (Salar, Shishir, Robert Kinsel).
- Tailwinds Matter
1988: #1 show = Cosby Show; 25 share, 26 channels
1998: #1 show = Seinfeld; 22 share, 52 channels
2008: #1 show = American Idol; 12 share, 110 channels
This wasn’t because people stopped watching TV; people were watching more TV in 2008 than 1988. Networks with <1% share collectively accounted for 50% of viewership.
“We believed that online video was going to do to cable what cable had done to broadcast television.”
Tailwinds accelerate everything you do. They also cover up a lot of mistakes.
2. Purpose Matters
Drive, by Daniel Pink. People are motivated by mastery, autonomy, and purpose.
Shishir and Sal Khan went to college together, and would get together regularly. In 2008, they were having dinner, and Sal mentioned, “You know, I use YouTube a lot.” Sal used YouTube for his videos because it was easier than sending a video file.
“I realized that Sal was getting more viewership than Stanford and MIT combined. So we invited him into the partner program.”
A few months after that, he and his wife were over for dinner again. Sal said, “I’m thinking about quitting my job and working on this full time.” I told him, “I’ve bet my career on this, and I think you should do the same.”
A few years earlier, Sal would have had to go to a cable network, and they would have asked, “Are you a teacher? Are you an entertainer?” He would have gotten laughed out.
In 2010, I was thinking about leaving YouTube. I called Dean Gilbert for advice, and he told me to make three lists.
- What do you want to accomplish in the next year? (Objectives)
- Which of those things will you not be able to do if you leave? (Marginal Utility)
- Which of those things would you really regret not doing? (Purpose)
There will always be times when you’re not sure whether you can go on; what will get you through is purpose.
3. Thesis Matters
Peter Thiel: “What important truth do very few people agree with you on?”
2008: Decided to leave Microsoft and start a company. Started to meet up with people in Silicon Valley. Jonathan Rosenberg asked Shishir to come by and meet the folks at Google. “Google is an advertising company, and all the money seems to go to TV ads that nobody even watches.”
I started thinking about the Superbowl. When people come over for the Superbowl, everyone watches all the ads. So I asked, how can we make TV so that it’s always like the Superbowl? I told them to Jonathan, and he convinced me to give up on starting a company and work on YouTube.
The ideas became “True View.” Put a button on the ads so that people can skip the ones they don’t like, then don’t charge advertisers. The way to change advertising is to change the incentives. YouTube had no desire to show ads that would be skipped; advertisers now had an incentive to make great ads.
Today, True View is a multi-billion dollar business. By the time I left, revenue per viewer hour was roughly even with cable TV.
Peter’s contrarian principle applies when you’re scaling a company, not just when you’re starting one.
Student Q: What did Jonathan do to convince you to join YouTube?
A: Mostly, he fed my interest in solving the problem. I didn’t know anything about advertising, and he showed me all the people who had succeeded because they lacked traditional experience.
4. Metrics Matter
2011/2012 was the “wandering period” for YouTube. From 2008–2010, we were just trying to avoid failing. It was very motivating. By 2011, we were sustainable, and the pace slowed. We needed to find a way to get that back.
We had no real competitors. Everyone thinks it’s better to be the clear #1 in a market. There are downsides. It’s incredibly motivating to be #2.
Two stories:
One, Coca-Cola. At Coke, people were saying, why do we care about market share relative to Pepsi. We should be thinking of stomach share.
Two, the British rowing team. They hadn’t won a title in many years. They decided to make every decision based on one thing: “Will it make the boat go faster?”
So, we decided that our single clarifying metric was watch time, and our goal was to get to one billion hours per day of watch time. At the time, we were doing 100 million hours per day. Facebook had about double that. Television as a whole was 5.5 billion hours per day.
The goal was crazy. Our head of data science said I was crazy and it would never happen. Every team would have to do things differently. At the time, we were using about 20% of the Internet’s bandwidth; at a billion hours per day, we would be streaming double the capacity of the Internet…so the networking team had to get to work.
A couple of things happened. One, it greatly clarified decision-making. We have tons and tons of data, which makes it easy to argue different points and hard to make decisions. Why was YouTube oriented towards short-form content? Even though there was no technical limitation, all our algorithms were optimized for views, not watch time. Two, pace increased. People started proposing bolder initiatives.
Picking a single clarifying metric is very hard, but it clarifies decision-making and what constitutes success.
5. Decisions Matter
What makes a good decision? Typical answers are “quick,” “made with the right people’s input.” I believe the key is whether or not the decision sticks.
In 2008/2009, a product manager came to me and said we should link out to other video websites. “Our job is to give people answers; we should point them to the right website.” That’s not what we decided to do.
Someone brought up Google Product Search and Amazon. By any logic, Google Product Search should have kicked Amazon’s ass. It is a superset. Same for Google Video Search and YouTube. In both cases, the narrower property one. In certain markets (like online video), consistency beats comprehensiveness.
So we didn’t add external links to YouTube. This led to a number of other decisions. We stopped allowing external embeds. We removed the choice as to whether or not videos would play on mobile. We even sacrificed distribution, such as when we pulled Apple’s YouTube player, which meant that we weren’t default installed on iPhones.
A decision that sticks helps you make the next 10 decisions, or better yet, obviates the need to even make them.
6. Ecosystems Matter
YouTube is very important to the music industry and vice versa. Young people use YouTube for music more than terrestrial radio.
During one marathon negotiating session with the record labels, I asked, “When is your relationship with YouTube going to help you win clients?” In the old days, they capitalized on their relationships with record stores or DJs. The record label guy told me, it wasn’t about the money. Most of the money is made on touring. The one thing they are obsessed with is where they stand on the Billboard Charts. If the artist drops from 2 to 4, they’re on the phone, “What are you going to do to fix this?”
So I asked my team, “What will it take for Billboard to take YouTube data into account for their rankings?” Nobody had ever called them. Billboard said, “Great idea, how much are you going to charge us for the data?” We said, “Oh, we thought we’d have to pay you.”
A few weeks from launch, Billboard asked, “Can we launch tomorrow?” The reason why was to have an unknown, unsigned label debut at #1. “Harlem Shake is No. 1 After Billboard Adds YouTube to Chart Formula.”
YouTube is a now a primary channel for artists, partially because of the decision we made to share our data with Billboard.
Don’t presume; get the data, ask the questions, and you might learn something completely unintuitive.
7. Values Matter
On September 11, 2012, we get a call. There was an attack on the American Embassy in Libya; the ambassador was killed. People were saying that the cause was the video, “The Innocence of Muslims.”
Some people wanted to take the video down. “It’s just a video. If it saves one life, we should take it down.” On the other hand, our policy team couldn’t find anything wrong with the video. Every couple of hours, there would be another protest.
The video stayed up. As these properties and businesses get bigger and bigger, the impact we have is incredible. We were the last decider; no judge or government could tell us what to do. You have to think about your values in advance so you have some frame for making urgent tough decisions.
Student Q: How did you make the decision to leave China on the table?
A: There was no real choice. I tried many different tactics, but there wasn’t a practical way to do it. In my opinion, I would have done it. We launched YouTube in all sorts of countries where we didn’t believe in their form of government. We felt that it was better for people to express themselves.
8. Talent Matters
Here’s a framework from my old boss on how to compare people:
X-axis: Scope
Y-axis: PSHE (Problem/Solution/How/Execution)
You start as a junior employee, and you work on a very limited scope. You’re given the Problem/Solution/How, and you go Execute. As you get more senior, your scope widens, and you get less and less (e.g. you just get the Problem, and you figure out the Solution/How/Execute).
The career path on this graph is an S-Curve; in the middle, it’s more about taking off the “training wheels” than increasing scope.
The “Dream Team” Approach: “Who would you hire first if you were starting this team from scratch?” We’d end up with an ordered list.
We’d draw three lines: The Awesome Bar, The Hiring Bar, The Wouldn’t Hire Again Bar. You want to invest your time in the people who clear the Awesome Bar. A lot of organizations set the Hiring Bar at the same place as the Wouldn’t Hire Again Bar. We’d set the Hiring Bar higher. We’d use actual people to define the bar, rather than abstract qualities.
We did this every quarter; more frequently would be even better.
Student Q: Did you fire people based on this?
A: Once people were told that they were below the “Wouldn’t Hire Again Bar,” they usually did the right thing. But in some cases, we absolutely did [fire people].
9. Your Role Matters
There’s more than one way to be a bad manager.
A) Micromanager (someone comes to you with a problem, you take that problem away and solve it)
B) Dictator (someone comes to you with a problem, and you tell them, that’s your problem, not mine)
C) Over-empathizer (someone comes to you with a problem, and you spend all the time helping them debug the problem, which means they no longer take responsibility).
Sometimes you need to be all of these things; you need to understand your natural tendency and how to manage it.
10. Cathedrals Matter
http://www.anecdote.com/2008/07/the-stonecutters-cathedral-builder/
At an offsite, we asked people to describe their job. After people talked about their jobs, I asked, “Did that sound like laying bricks, or did it sound like building a cathedral?”
Q&A
Q: Describe what things were like when you first got to YouTube.
A: There were four phases. The founding phase was before YouTube was purchased and before I joined. After the purchase came the wandering phase. And today, it’s in the phase of acceleration.
YouTube was an enormous success; it had been purchased for $1.6 billion, and it was a very popular property. But our backs were against the wall. On October 18th or 19th, three days after my daughter was born, I went into work and met with the CFO (who was and is one of the most awesome guys you’ll ever meet). He had a deck waiting for me with three charts. The first was how much money YouTube was losing per year. (It was a lot.). The second was how much we were losing per view (It was less, but enough). And the third was how many views we were delivering (Which were skyrocketing). “This could be the worst business in the world,” he told me. With his help, we got to work on fixing the economics.
Everyone was focused on this problem. Networking was trying to shave costs. Content was trying to negotiate better deals. It was intense, but you made decisions fast and tried to execute.
But then the wandering period was difficult in a different way. It felt like we’d become a big company all of a sudden. In a healthy dynamic, if you’re working on A and B, and someone brings up C, everyone’s too busy to do C unless they drop A or B. Temporarily, that changed.
Q: What were the key org changes you made to get YouTube not losing money?
A: The org wasn’t that important, but the people were. We moved people out of positions fast. We didn’t have time to see if people would work out. We pulled in the best help from anywhere we could.
TrueView was shelved. I couldn’t get it through the Sales team. There was a guy on the Sales team who really wanted to be a project manager. So I told him he could work on TrueView in his 20% time.
Q: What company communications process did you put in place to communicate with the team?
A: A lot of forums are standard — TGIFs, Q&As. I was a big believer in writing a weekly email. Leaders that write things down tend to deal with less communications issues. You have to clarify your thought processes in a completely different way. If you just have a meeting and say, “Okay, so we’ve all decided,” and then people play telephone. I would send a weekly YouTube-All mail.
Student Q: How do you think about timing in terms of product and business space?
A: Timing was actually one of the stories I ended up cutting! YouTube was far from the first to try to make online video work. I’ve had many arguments with Mark Cuban whether he was the first. Timing is everything, and you don’t know if the timing is right until afterwards.
Why was YouTube at the right time? Networks were finally big enough to stream video. Cellphone cameras allowed everyone to record videos. The investment environment allowed a very capital-intensive bet. And luck comes into it as well.
Student Q: How do you think VR will change the industry?
A: I think VR is a fantastic promise; I think the timing of it is unclear. Anyone who has put on an Oculus can’t help but be wowed. But one of the lessons from YouTube is that a lot of things have to come together at the same time — networks, devices, content. I don’t the time is right yet, but I think it will be great when it does happen.
Student Q: Why did you finally decide to start a company?
A: When I was at YouTube, and people came to me and said they wanted to start a company, I’d ask two questions: One, do you have an idea you can’t imagine not working on, and two, do you have a team that you can’t imagine not working with. Partly, I was trying to get them not to leave.
I found myself answering “Yes” to both questions. It’s a feeling like, damn, I’m trapped, I have to do it.
