Scaling YouTube with Shishir Mehrotra — Class 13 Notes of Stanford University’s CS183C

Here is an essay version of my class notes from Class 13 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 a talk with Shishir Mehrotra (Shishir)— former head of Product, Engineering, and UX at YouTube — on lessons learned from scaling YouTube.

Video of the class, notes are below:

I. Background and context

Shishir joined YouTube in 2008 and was there for 6 years in total. YouTube in 2008 was in a very different state than the YouTube we know today. In this time period, Google had just bought YouTube and while it was a financial success, YouTube was not looked at as a successful company.

All of the press thought Google had made a mistake, and there were days even Google itself thought YouTube was not going to work. 6 years later YouTube looks very different.

Secondly Shishir’s role was a non-traditional role. Officially he was in charge of product, engineering, and UX. In addition to Shishir — Salar Kamangar was the “CEO” of YouTube and Robert Kyncl was the head of business (sales, business development, partnerships, etc).

Shishir’s talk is a collection of 10 counterintuitive lessons he learned from scaling YouTube, broken up into three categories:

  • Choose wisely —Similar to Poker, it’s about knowing what table to sit in on, as much as it is about learning the game.
  • Charting the course — The core issues of running a business.
  • Your crew — Team related items.

II. Tailwinds Matter

Tailwinds are large market forces which benefit the business and market you are in. If you work in (or start) a company with good tailwinds, it accelerates and allows you to be able to make mistakes and still weather through the storm.

In YouTube’s case, there was a trend of the #1 TV shows in each generation getting a lower and lower household audience share (1988 Cosby show, 25 point rating — 1998 Seinfeld, 22 point rating — 2008 American Idol, 12 point rating).

The reason this was happening wasn’t because the total number of hours people watched TV were going down over that time period (it went up), but the choices of channels and programs people could watch went way up.

Over the history of TV there was historically 3 major networks, which became 100’s of cable networks. The tailwind of YouTube was the thesis that the same transition would happen in the online space — online would give people much more choices vs. the cable networks — what cable did to broadcast, YouTube would do to cable.

III. Purpose Matters

Daniel Pink’s book "Drive" states that what motivates people is not money but rather — mastery, autonomy, and purpose. Based on Shishir’s experience he says the thing that matters the most about YouTube was purpose.

Shishir shared that he is good friends with Salman "Sal" Khan whom he went to college with. When Shishir joined YouTube, Sal Khan mentioned he was an active uploader to YouTube. (More about the story of how Sal Khan started Khan Academy is below).

When Shishir went back to the office, he looked at the stats of Sal Khan’s videos and noticed he was getting more viewership than Stanford and MIT combined. He helped him get onto the YouTube partner program which helped content creators make money through ads played on his videos.

Over time, Sal saw that this income stream could support his living expenses and eventually could make him a full-time income, and decide to leave his hedge fund job to be a full time YouTube educator.

The interesting part of this story is, imagine if Sal tried to do this a few years earlier. He would have had to pitch a cable network, on a new educational show, where he teaches basic math concepts, never shows his face, no plot, and no storyline. There would have never been any way Sal’s show would have ever got funded in the cable world.

YouTube’s mission was to have a gatekeeper-less world where people like Sal Khan could now become one of the biggest educators on the planet.

One exercise Shishir learned from his mentor and recommends others to do is to make three lists:

  1. What do you want to accomplish in the next year? (what are your goals, what do you want to focus on).
  2. Take the first list and figure out which wouldn’t happen if you stopped doing what you are doing (marginal utility).
  3. Take the second list and pick out the one thing you would feel terrible about, if you didn’t do it (purpose test).

This test has helped Shishir through some tough times at YouTube.

IV. Thesis Matters

When Shishir joined YouTube it was far from obvious that YouTube was going to be a success.

When Shishir contemplated joining YouTube —the XLII Super Bowl was going on and one thing he noticed was people loved the commercials during this time. He was thinking how he could make TV ads feel like Super Bowl ads every day.

He decided not to join YouTube but shared the list back with the person recruiting him and said, “I’m sure you guys are working on these ideas already, but happy to share and walk through them with you.” Turns out no one was working on these ideas and working on these ideas was the turning point to joining YouTube.

These set of ideas became YouTube TrueView — putting a skip button on all of the ads — YouTube would only charge for ads that were viewed, not ones that were skipped.

The insight was, the way to change advertising was to change the incentives. By not showing ads that people didn’t want — advertisers would have to work harder to create better ads.

At the time, this was a very counter-intuitive idea. Not many people within Google/YouTube believed in it, especially the ad sales team. Over time they learned this was the right move because the “revenue per hour” of watch-time eventually became equivalent to the amount cable TV was making per hour.

V. Metrics Matter

In 2008–2010 YouTube had a “fire under its ass” because the world thought they were going to fail. In 2010–2012 YouTube was sustainable, things were good, and the pace slowed down.

The question was — how to get the momentum back? This is tough because for YouTube they didn’t have any real direct competitors. Many people think it’s better to be #1 but the downside is you lose yourself. It’s more motivating to be #2 than to be #1.

Two stories which apply to this:

  • Coca-Cola was in a strategy meeting where they were debating going back and forth between 49% and 51% market share against Pepsi. Instead of just going back and forth, they decide to look at the “% of stomach” instead — which led to Coca-Cola investing in all the other food businesses they are in now.
  • Ben Hunt-Davis was head of a rowing team which had not won a title since 1912. They came together as a team and made the decision to ask a single question for all decisions — “Will it make the boat go faster?” They applied this question to every single decision and won the 2000 Summer Olympics.

Based on these two stories, the YouTube team got together and made the decision to make one universal metrics for YouTube — Watch Time. Secondly, they also set a goal to get to 1B hours of watch time per day within a time period (couldn’t say time period publicly).

Context:

  • The closest comparable to 1B hours of watch time was — Google which had 100M hours of watch time per day (but this is a bad metric because people go on Google to leave Google).
  • Facebook had about 200M hours of watch time per day.
  • Television (cable programming) was at 5.5B hours of watch time per day.

The justification was, if YouTube was successful, they would still only be 20% of TV. This was their version of 1% of the stomach.

The benefit was, having one core metric greatly clarified decision making; however, the problem with having tons of data is, it’s hard to choose — what do you optimize for with competing metrics?

Secondly the entire YouTube company knew exactly what they were working on and what the goal was. Because of this, all sorts of innovation started to happen again and the company became motivated.

VI. Decisions Matter

One interview question Shishir likes to ask is “What makes a good decision?”

Typical answers are:

  • The best decisions are the ones which are made quickly.
  • The best decisions are made with all of the right people's input.

Shishir’s view is there is one answer to this question — did the decision stick?

The absolute best decisions aren’t the ones which are made quickly, but rather the ones which not only stick but where the team also understands the core principles behind that decision — which stops the need to make 10 additional decisions.

VII. Ecosystems Matter

The music industry has a very symbiotic relationship with YouTube in both directions. Music was extremely popular on YouTube and YouTube would also pay music labels for music played on YouTube.

This meant the negotiations with the music labels were tough and very adversarial. Shishir had a heart to heart with one of the music execs and asked — when did their relationship with YouTube help win clients? (vs. being adversarial — similar to the relationship labels have with record stores and DJ’s). Was it when YouTube accounts for a certain % of revenue or a certain viewership scale?

The music label executive responded — no it wasn’t any of those things. The majority of revenue is made through touring, and artists don’t think in terms of viewership. The one thing artists were obsessed with were— where they stood on the billboard charts — if a certain artist even dropped one spot on the charts, the labels would be called immediately to fix it.

Thus YouTube proposed an idea to share their data with Billboard — no one had thought of this before. Billboard asked how much YouTube was going to charge for their data — YouTube thought Billboard was going to charge them — they were happy to do it for free. YouTube has since become the primary vehicle for musicians, partially because of this decision they made to share data with Billboard.

The lesson here is, every business operates within an ecosystem. If you ask the right questions to the right people within your ecosystem — you might find out something counterintuitive.

VIII. Values Matter

On September 11th, 2012, there was an attack on Benghazi Libya. At the time, the attack was thought to stem from the YouTube video “The Innocence of the Muslims.”

YouTube was under extreme pressure to take down the video from multiple government agencies and groups. There were compelling arguments on both sides: to both keep it up and take it down. In the end, YouTube decided to keep it up and over time it was determined that the video had nothing to do with the attacks.

This was important because, as these platforms get bigger and bigger — the level of impact of these platforms becomes unprecedented. YouTube was the final decider in this case — no one could force them to take action — and this power comes with great responsibility. Therefore it’s important to think about your values in advance.

IX. Talent Matters

Shishir shared a framework for judging talent which also applied at Google — namely in determining the levels of individuals (senior vs. junior executives — what is the difference between a level 3 product manager and a level 6 product manager?).

  • On the X axis is scope. What scope can the person capably work on — a feature, an area of a product, a full product, or multiple product lines?
  • On the Y axis is the autonomy a person can deal with.
  • PSHE (Junior) — A manager needs to spell out the problem, a rough solution, send you a list of instructions, and your job is to go and execute based on those instructions.
  • PSH — You are given the product and rough solution — then you figure out how to accomplish the solution.
  • PS — You are given a problem and you have to go figure out the solution.
  • P — You are given a space and it’s your job to figure out what the problems are.

Some people call this leadership, but Shishir likes to call this "training wheels." The question for employees is — what is the biggest scope and responsibility you can give someone, without any "training wheels"?

A second thing Shishir noticed is, generally the path of learning was not linear.

Typically people would initially grow by adding scope — then there is this big trough in the middle. This part isn’t about scope but rather more about how they did the job. Then a person would grow at the end by adding more scope again.
A second tool they used at YouTube was called — Dream Teams.

They would take a team of people and ask — who would you hire first if you were hiring this team from scratch? They would compile all of this information across all of the teams and divide everyone by three lines:

  1. The first line was the "awesome bar" — these are people you would build a team around, people you invest your time into (make them feel valued, rotate them through teams, give incentives to).
  2. The second line was the "hire bar."
  3. The third line was the “don’t hire bar” — these people you probably wouldn’t hire again.

While they were interviewing people, they would literally use a list of 4–5 people and use them as the hiring bar for future candidates. It’s easy to just hire people to do the job, it’s much harder to maintain a high bar.

X. Your Role Matters

One way to talk about the leader's role is to talk about the inverse — what makes a bad manager?

  • Micromanaging — taking problems and solving it for employees.
  • Dictator — making every decision go through you.
  • Over emphasizer — spends all their energy helping people with their problems (this is bad because it takes the responsibility off of the person being helped).

Everyone has a natural tendency to be one of these types of managers. The key to being a good manager is not to not be any one of these things — but rather be the one you need to be at that time.

Depending on your natural tendency, you have to stretch yourself to be the other two types of managers.

XI. Cathedrals Matter

This is best told through a story:

As the story goes, three bricklayers are working together, side by side in the hot summer sun.
A young child, passing by on his way home from school, stops and looks at the bricklayers. “What are you doing?”, he asks.
The first bricklayer replies: "I’m laying bricks.”
“And you?” says the child, “What are you doing?”
The second bricklayer answers: "Feeding my family.”
“I see,” said the child.
The third bricklayer is then asked the same question: "What are you doing?”
He responds.
“I’m building a cathedral.”
From: http://www.ryandurkin.com/blog/2012/06/laying-bricks-vs-building-a-temple/

The main lesson to think about when organizing teams is — when you look at each person — how connected do they feel to the cathedral? Do they describe their job as laying bricks? How can you help point everyone’s job (at all levels) to the cathedral?