【斯坦福大學公開課】《如何創業》(全20集) P17 - 17.17.17運營管理巧上手

財富導航
16粉絲
2026年11月17日
So I've watched some of the prior classes and I'm going to assume that
you've already sort of hired a bunch of relentlessly resourceful people,
that you've built a product that at least some people love,
that you've probably raised some capital, and now you're trying to build a company.
So you've been forging a product and now you've got to forge a company.
And I'd actually argue forging a company is much more difficult than forging a product.
The basic reason is people are irrational.
So you probably all know this, either your parents, your significant other,
your brother or sister, your teacher, somebody in your life is irrational.
Building a company is basically taking all the irrational people you know,
put them in one building and then living with them 12 hours a day at least.
So it's very challenging.
Now there's techniques for coping with that and some people get good at it and some people don't.
But that's really what operating is about.
So basically what you're doing when you build a company is you're building an engine.
And at first you have a drawing, literally on a whiteboard,
and you're architecting it.
And it looks very conceptually clean and beautiful and pretty.
But when you start actually translating it to practice,
it actually looks more like this and you're holding it together with duct tape.
And it takes a lot of heroic efforts of people to actually hold it together.
That's why people work 80, 100 hours a week.
It's that heroic efforts is actually necessary to keep this thing together
because you don't actually yet have polished metal in place.
Eventually you want to construct a high-performance machine.
A machine that actually almost nobody really has to worry about every hour, every minute.
And you know, we used to joke about eBay that if Martians took over eBay,
it would take like six months for the world to notice.
That's what eventually what you want to get to.
Or as Warren Buffett says, you know, build a company that idiots can run
because eventually they will.
So this is what you want.
Basically a performance machine that idiots can run.
Now as a leader, what is your real job?
What's your role?
Strictly speaking, there's only one book ever written that actually explains how to do this.
It's a little old. It's written in 1982 by Andy Grove.
It's quite famous and quite successful.
And his definition of what your job is, is to maximize the output of an organization,
your organization that you're responsible for.
Your CEO is everything and a VP would be part of the organization
and the organizations around you.
So if you're a VPE, you actually are responsible for the performance of the product team too
or the marketing team because you have influence there.
So this is how you measure people.
And you want to focus on the output, not the input.
The old adage about measuring, not measuring motion and confusing that with progress.
You're measuring only progress.
And this is going to sound like a fancy and glamorous thing to do.
Maybe people get excited about managing a whole large organization
and being responsible for the output.
But in practice, what you're going to learn today, hopefully,
is that it's more about things like ordering smoothies,
teaching your receptionist how to answer the phone properly,
and serving as a $10 an hour task rabbit for your employees.
So let's talk about that.
So at first, when you start a company, everything's going to feel like a mess.
And it really should.
If you have too much process, too much predictability,
you're probably not innovating fast enough and creatively enough.
So it should feel like every day there's a new problem.
And what you're doing is fundamentally triaging.
So some things will look like a problem, and they're actually a cold.
They're just going to go away.
So somebody's annoyed about this or that.
That may be a cold, and you shouldn't stress about it.
And you certainly shouldn't allocate a lot of your time to it.
And some things are going to present themselves as colds.
But just like in the emergency room, if they're not diagnosed properly,
they actually can become fatal.
So what I'm going to try to do is help give you frameworks
for thinking about which things are colds and which things are potentially fatal.
So one of the most important things I've learned at Square
is actually a concept of editing.
And this is, I think, the best metaphor I've ever seen
in, like, 14 years of running stuff of how to think about your job.
It's natural. It's a natural metaphor.
So it's easy to take with you every day.
And actually, it's easy to transmit to each of your employees
so that they can figure out whether they're editing or writing.
It's a natural construct.
You generally know when someone asks you to do something,
am I more writing or am I more editing?
So an editor is the best metaphor for your job.
And we're going to talk about the specific things you're doing in editing.
The first thing an editor does,
and you've probably all had this experience in school,
is you submit a paper to a TA, a draft to your friend.
The first thing an editor does is they take out a red pen,
or nowadays, you know, online,
and they start striking things, basically eliminating things.
The most important task of an editor is to simplify, simplify, simplify.
And that usually means omitting things.
So that's your job, too, is to clarify and simplify for everybody on your team.
The more you simplify, the better people will perform.
People cannot understand and keep track of a complicated set of initiatives,
so you've got to distill it down to one, two, or three things
and use a framework that they can repeat.
They can repeat without thinking about it.
They can repeat to their friends.
They can repeat at night.
Don't accept the excuse of complexity.
A lot of people will tell you that, well, this is too challenging,
this is too complicated.
Well, yeah, I know other people simplify, but that's not for me.
This is a complicated business.
They're wrong.
You can change the world in 140 characters.
You can build the most important companies in history
in a very simple-to-describe concept.
You can market products in less than about 50 characters.
There's no reason why you can't build your company the same way.
So force yourself to simplify every initiative, every product,
every marketing, everything you do,
and basically take out that red and start eliminating stuff.
Second thing is what editors do is ask you clarifying questions.
So when you present a paper to somebody, what do they usually do?
They find some ambiguity somewhere, and they say,
well, do you really mean this?
Do you mean that?
Do you have an example of this?
That's what your job is.
So you're in a meeting, and people are going to look to you,
and the real thing you do is you actually ask a lot of questions.
And they can be simple, basic questions like,
should we try this seven days a week or six days?
They can be fundamental questions like,
where's our competitive advantage here?
We try to do this actually as investors too.
Some investors will ask you a billion questions about a billion things,
and they'll have you do diligence forever.
We try to narrow down to what are the one, two, or three,
four things that actually matter for this company
and only focus on those things.
So it allows us to be more decisive,
and we can make decisions rapidly.
It allows us not to distract you from your day job,
which is actually building a company.
And yet I think we still get to the highest fidelity answer
because we don't have all these extra extraneous pieces of details and data.
Now, it's hard.
It's something you have to practice.
But when you're good at it, every step you eliminate,
Andy Grove estimated that you can improve performance by 30% to 50%.
The next thing you do is allocate resources.
So the editing construct, this is what editors do all the time.
They take editors from the Mideast and covering the Mideast,
and they move them to Silicon Valley
because Silicon Valley is now more interesting.
Or they move them to a sports section because they want to compete
on the basis of sports with other journals and other publications.
So that can be top-down where I take a bunch of allocated resources
and people and say, we're now going over here.
We're going to compete on this basis.
And then next month, next quarter, next year, I'd be like,
well, that Middle East coverage is getting boring and bland.
We don't want to do that anymore.
Let's go chase after something else.
Or it can be bottom-up.
Just like journalists mostly come up with their own stories,
the people who work with you generally should be coming up with their own initiatives.
So a reporter will generally, who covers Google,
come up with the interesting stories that they're hearing in the ether
and propose one or two to their editor for approval.
But it's not like the editor is saying, go cover Google,
and this is the angle I want.
Once in a while, they do that.
But that's not the meat and potatoes of what a journalist does every day.
Your goal over time is to actually use less red ink every day.
So one way of measuring how well you're doing in communicating to your colleagues
about what's important and what's not, why some things are important
and some things are not, is how much red ink you're pulling out every day.
It's okay if you have a bad day and the red ink is all over the place.
But it's not okay if the red ink next month is more than it was last month
and next quarter is more than this.
So measure yourself with how much red ink you're creating.
The other thing that's very important that actually isn't as intuitive
to a lot of people is the job of an editor is to ensure a consistent voice.
So if any of you read The Economist, you can tell that there's one consistent voice.
You can pick up any article, any post in The Economist,
and it feels like it was written by the same person.
Ideally, your company should feel like, on your website, in your PR release,
on your packaging, if it's a physical product, anywhere,
on your recruiting pages, should feel like it was written by one person.
That's extremely difficult to do.
And at first, you're going to be tempted to do that yourself,
which is okay for a founder to do that, him or herself initially.
Over time, you do not want to be doing all of the consistent voice editing by yourself.
You want to train people so they can recognize the differences in voice.
So if you see this website page, it looks very different than the recruiting page.
You start asking questions, why is that?
Is the reporting messed up?
Is one of the readers over here not really understanding the voice of the company?
So you have to fix that over time.
But you want to start with the objective of everything should feel exactly the same.
It's quite difficult in practice.
Almost every company has at least one piece of the organization
that isn't exactly on the same voice.
At Apple, which is notorious, even under Steve's regime,
which was notorious for getting this right,
if you ask someone who worked at Apple,
ask them about the internal tools about recruiting.
Do they really feel like Apple products?
All of them will tell you no.
So you're never 100%, but you definitely want to get as close to that as you can.
The next complicated topic is delegating.
So just like the other thing I like about the metaphor of editing is
writers do most of the work in the world.
Editors are not actually writing most of the content in any publication.
So that's actually true of your company.
You are not going to do most of the work.
You shouldn't be doing most of the work.
And the way you get out of doing most of the work is you actually delegate.
Now the problem with delegating is
that actually you're responsible for everything.
So as CEO, founder, there is no excuse.
There's no like, that's that department over there,
this person over there, screwed up.
You're always responsible for every single thing,
especially when things go wrong.
So how do you both delegate but not abdicate?
It's a pretty tricky challenge, and both are sins.
Like if you over-delegate and you abdicate,
or you micromanage, those are both sins.
So I'm going to give you a couple techniques for solving this.
First, and this actually comes from high-output management in Andy Grove,
is what's known as task-relevant maturity.
It's kind of a fancy phrase for basically,
has this person ever done this before?
It's really simple.
How mature is this person in doing something?
And the more they've done the exact same task before,
the more sort of rope you're going to give them,
and the more they're trying something new,
the more you're going to actually instruct them
and consistently, constantly, regularly monitor.
So that's kind of a basic concept,
but it's worth keeping in the back of your brain.
The interesting implication, and this is pretty radical,
is that any executive, any CEO,
should not have one management style.
Your management style actually needs to be dictated by your employee.
So with one particular person,
you may be very much a micromanager
because they're quite low on this scale.
And with another person, you may be delegating a lot
because they're actually quite mature on this scale.
So it's actually a good thing if you do reference checks on somebody,
and half the people you call say they're a micromanager,
and the other half say they actually give me a lot of responsibility.
That's a feature, not a bug.
I didn't understand that at first at all.
I used to be befuddled when people would do reference checks on me
and come back with this complicated mosaic.
And I basically finally figured out
that maybe I was actually doing my job correctly.
And so then, of course, taught other people
that this is the way to do it.
A more nuanced answer, though,
that I've sort of came up with,
is how do you make decisions
and delegating versus doing it yourself.
You don't want to do it yourself too often.
So what I've basically borrowed from Peter,
this is my first two-by-two matrix ever in my life,
but he taught me something at least,
is you basically sort your own level of conviction
about a decision on a grid.
Extremely high, extremely low.
There's times when you know something's a mistake.
There's times where you wouldn't do it that way,
but you have no real idea whether it's the right or wrong answer.
And then there's a consequence dimension.
There are things that if you make the wrong decision
are actually catastrophic to your company, and you will fail.
There are things that are pretty low impact
that really, at the end of the day,
aren't going to make a big difference, at least initially.
So what I basically believe is that
where there's low consequences
and you have very low confidence in your own opinion,
you should absolutely delegate.
And delegate completely.
Let people make mistakes and learn.
On the other side, obviously,
where the consequences are extremely dramatic
and you have extremely high conviction that you're right,
you actually can't let your junior colleague make a mistake.
You're ultimately responsible for that mistake,
and if it's really important,
you just can't allow that to happen.
Now, the best way to do that is to actually explain your thinking, the why.
It's easy to shortcut when you get busy
explaining why's of the world,
but it's very important to try.
When I was at LinkedIn,
I had a colleague who's quite, quite talented,
but occasionally would get annoyed
if I didn't exactly agree with his opinion on something.
And so I spent a lot of time trying to persuade him
why I was making a decision a certain way.
And his wild card, the card he would ultimately call out
if I didn't quite persuade him was, he's like,
okay, you're the boss.
And that meant to me like I was burning a lot of social capital.
Every time he said that, I knew I was like really creating a thin line
and that ultimately that was going to backfire if I did that too often.
So you want to track the times that you're doing that.
An example of this is Square.
It's one of my favorite people in the world
and my second hire, first marketing hire,
had this program he wanted to run called Inner Square,
which basically allowed people to give out,
Square merchants to give out 10 other squares to their,
imagine a food truck outside,
put like 10 squares on the counter
and people could just grab them.
And Kyle had this great idea,
this would be an awesome marketing program,
squares would spread squares to other people.
And to some extent, it was on brand,
so it didn't have catastrophic consequences.
Each of these 10 squares didn't cost that much money,
so financially we could afford to do it.
But at that time, my 10 years of experience said
there's just no way this is going to work
at a meaningful enough scale
to move our metrics enough
and I actually would prefer we not do this.
But Kyle was so excited about this
that I decided to just let him do it.
He learned that actually when you measure this thing,
it's not massive,
it doesn't create massive value for the company,
it did require a fair amount of operational complexity
to ship all these squares to people
and figure out how to get them, squares, etc.
But it allowed him to be excited about his job
and to learn how to filter future ideas,
so it was totally worth letting him make the quote-unquote mistake.
The next and possibly most important thing you do
is actually edit the team.
So these are the people that you work with
and nobody's going to have a perfect team
and you're certainly not going to start that way.
So what I'm going to try to do
is maximize the probability for success
in editing the team.
So I like this idea of barrels and ammunition.
So most companies, once they get into hiring mode,
as Sam pointed out, you defer that for a while,
but once you do, they just hire a lot of people.
And you expect that as you add people,
your throughput, your horsepower,
your velocity of shipping things is going to increase.
Turns out it doesn't work that way.
Usually when you hire more engineers,
you actually don't get that much more done,
you actually sometimes get less done.
You hire more designers, you definitely don't get more done,
you get less done per day.
We can talk about why.
So the reason why is that most people,
most great people even, are actually ammunition.
But what you need in your company are barrels
and you can only shoot through the number
of unique barrels you have.
So that's how the velocity of your company improves,
is by adding barrels.
And then you stock them with ammunition
and then you can do a lot.
So if you go from a one-barrel company,
which is mostly how you start,
to a two-barrel company, suddenly you get
twice as many things done per day, per week, per quarter.
If you go to three barrels, great.
If you go to four barrels, awesome.
Barrels are incredibly difficult to find.
But when you have them, give them lots of equity,
promote them, take them to dinner every week,
because they're virtually irreplaceable.
Because they're also very culturally specific.
So a barrel at one company may not be a barrel
at another company because one of the ways
and the definition of a barrel is they can take an idea
from conception all the way through shipping
and bring people with them.
And that's a very culturally specific opportunity,
specific skill set, rather.
So two questions probably are occurring to you,
is how do you figure out who's a barrel and who's not?
One is you start, actually, with a very small
set of responsibilities.
It can be fairly trivial.
It can be something like I want to reward the engineers
that are in my office at 9 o'clock every night
with a nice, cold, fresh smoothie.
This is actually a real example.
I was frustrated.
Our engineers were working really hard at Square.
And maybe 20% to 30% would stay very late into the evening.
And we already had served them dinner,
but I wanted to give them something cool to reward them.
You could think about alcohol, but that's a little complicated.
So smoothies were probably a little bit better than the pizza,
which drains you of energy.
But nobody could get smoothies in my office
to show up at 9 o'clock sharp that were cold,
that tasted good, and delivered in the right place
that engineers would find them.
You'd think this is simple, but in fact,
it took months to get this right.
So we had an intern start, and I think on the second day,
I was explaining this problem, and he said,
well, I'll do it.
And I was looking at him like, there's no way.
I've seen my office manager fail, my assistant fail,
all who were actually pretty good.
This just isn't going to happen.
And then, lo and behold, that night, they show up on time,
cold, delivered in the right place,
and my first instinct was, great,
nothing about the smoothies, but okay,
now I can actually give them something more important
and consequential and complicated to do.
And that's what you actually want to do
with every single employee every single day,
is expand the scope of their responsibilities
until it breaks.
And it will break. Everybody.
I couldn't run the world.
Everybody has some level of complexity
that they can handle, and what you want to do
is keep expanding it until you see
where it breaks, and that's the role
that they should stay in, that level of sophistication.
But some people will surprise you.
There will be people who you don't expect,
with different backgrounds, without a lot of experience,
that just can handle enormously complicated tasks.
And so keep testing that and pushing the envelope.
The other signal to look for
is once you've hired someone,
is with an open office, just watch
who goes up to other people's desks,
particularly people that they don't report to.
If people start going to your desk,
someone, individual employee's desk,
and they don't report to them, it's a sign that they believe
that that person can help them.
So if you see that consistently, those are your barrels.
Just promote them, give them more opportunity
as fast as you can.
The other question everybody asks about people is
when do you hire somebody above somebody,
and when do you sort of mentor somebody,
or when do you need to replace somebody?
And the way to think about this is every company
has its own growth rate, and every individual
has their own growth rate.
So some companies that are very successful, let's say LinkedIn,
LinkedIn was always a very linear company.
It never went like this.
So for example, I joined LinkedIn
18 months after we launched, and we had
only 1.5 million users, which in a social product
is a very small number.
And when I joined, I was the 27th employee.
When I left, two and a half years later,
we only had 57 employees.
In contrast, when I joined Square
as the 20th employee,
two and a half years later,
we had like 250, 300 employees.
So each company has its own velocity
on this curve, and if the company's going
like this, you can only keep people
in the roles that their own personal learning curve
is going like this. On the other hand, if the company's growing
like this, anybody whose learning curve
is faster than that, you can keep giving them
the same role to do. So always track
the individual slope of an employee
and the company's growth rate.
Now that you have your barrels
identified, so you're going to pick out the people that can
really take an idea that you have in the back of your head,
scope it out, run with it, make it happen,
ship it, and it's perfect.
Where do you aim these barrels?
So I'm going to argue that you really need
to spend a lot of time focusing people.
This is something I learned from Peter Thiel
actually. He used to insist at PayPal
that every single person could only do
exactly one thing.
And we all rebelled. Every single person in the company
rebelled to this idea, because it's so unnatural,
it's so different than every other company
where people want to do multiple things, and especially as you get more
senior, you definitely want to do multiple
things, and you feel like it's insulting
to be asked to do just one thing.
And Peter would enforce this pretty strictly. He'd basically
say, I will not talk to you about
anything else except this one thing I've assigned to you.
I don't want to hear about how great you're doing over here.
Like, just shut up.
And Peter would run away. And then
focus until you conquer this one
problem. And the
insight behind this is that
most people will solve problems
that they understand how to solve.
Roughly speaking, they will solve B-plus
problems instead of A-plus problems.
A-plus problems are high-impact problems for your company,
but they're difficult. You don't wake up in the morning
with a solution, so you tend to procrastinate
them. So imagine you wake up in the morning,
you create a list of things to do today.
There's usually the A-plus one at the top of the list,
but you never get around to it. And so you solve
the second and third, and over an entire
company of hundreds of people, that just cascades.
So you have a company that's always solving B-plus
things, which does mean you grow and does mean you add value,
but you never really create that breakthrough
idea, because no one is spending
100% of their time banging their head against the
wall every day until they solve it. So I
highly recommend some version of that.
You can be less stringent. You can be like,
you can get three things to work on,
but I would still track at least
the concept of what would happen if you only
gave everybody one thing to
prioritize. Now, because you can't
make decisions, you
don't want to be making all these decisions yourself.
You have to create tools that enable
people to make decisions at the same
level, ideally, of fidelity that you would
make them yourself. So how do you create scale
and leverage? First thing
I'd recommend is building a dashboard.
This is an old square dashboard. It actually looks
pretty presentable, even today.
The construct
of a dashboard, first of all, should be drafted
by the founder. You need to
basically simplify the value proposition
in the company's metrics for success
on a whiteboard.
You can have other people build the dashboard. I don't actually
care about that. But you need to draw it out.
Like, what does business success look
to us, and what are the key inputs
to those? And then have someone create
something that is very intuitive for
every single person in the company, including customer
support, to use. And then
the key metric of whether you've
succeeded is what fraction of your
employees use that dashboard every day.
If it's actually useful, it
should be close to 100%. It's not going to be
probably 100%, but you want to measure
that. Just like you have quality scores for
all your other KPIs with users, the dashboard
needs to be as intuitive as it is
as your product is for users.
Other things...
Yeah, let's go back
one second. Another concept is
transparency. And people will
often...that's weird.
Okay. Transparency
people talk a lot about.
It's kind of a goal that everybody
ascribes to. But when push comes to shove,
very few people actually adhere to it.
So let me walk through a little bit of transparency
and different stages of transparency.
Metrics are the
first step. So everybody in your company absolutely
should have access to every single thing that's going on.
Other things that
I like to do are take your board decks
and as you get more formal, the board decks will get
more complicated. And actually review
every single slide with every single
employee after the board meeting. You can strip
out the compensation information if you really
want to, but every other slide
you should go through with the entire employee base
and explain it. And if you can remember some
of the feedback you got from your board, that's really cool to
pass on. Another thing we did at
Square, as your company scales, everybody's
not going to get invited to every meeting, but they're going to want
to go to every meeting. The way you scale
that is you create notes for every meeting and you
send it to the entire company. So we created
a notes at alias for
every single meeting involving more than two people.
Someone would write notes and send it to the
entire company. So people felt like
as the company added employees, they could continue
to monitor and track what was interesting, what was
going on, and they never felt excluded
hopefully. The other thing is even
details around conference rooms. Every conference
room at Square has glass walls.
Because as soon as you have regular walls, people
wonder what's going on. It's amazing
if they can see who's actually in the meeting
and who's meeting with who when, they don't
worry nearly as much as what's going on
behind those closed doors. Stripe,
you may have seen a blog post by, I think Patrick
wrote it, about email transparency, about
actually allowing everybody to have access to email.
That's pretty far out there, but
it's actually got interesting certain
merits to it. I would all call
the tactics that you read about and hear about as
sort of minimal viable transparency. I actually
think you could push the envelope a lot more.
Steve Jobs tried this at Next. He actually tried
transparent compensation.
I actually think, although Next didn't do
extremely well, the real
reason wasn't because of the experiment
around compensation transparency, and that
there's a lot of merit to that. And the
critique of compensation transparency
is often, well, we want people to be teammates
or work together and collaborate.
And if you look in the sports world, though,
where people are actually teammates and they do have
to collaborate, all of their compensation
is completely public. In fact, each of us can
look up anybody's compensation in the sports world
and get it exactly accurate. And somehow
it seems to work. So I'm not totally bought into
the idea that you need to keep
compensation non-transparent.
Then finally, metrics.
So you want to measure
things. You want to measure outputs, not
inputs. And again, you should
dictate this yourself. You should draft the dashboard
to tie this all together. One important
concept, or what are known as pairing metrics, or
pairing indicators, which is if you measure
one thing and only one thing, the company tends
to optimize to that. And often at the expense
of something else that's important. A classic
example in payments and financial services
is around risk. It's really easy to give
the risk team the objective and say, we want
to lower our fraud rate. It sounds great.
Until they start treating every single
user in this audience as a suspect user
because they want to lower their fraud rate. So they require
each of you to call them up on the phone
and give them more supplemental information and facts
in things. And you have the lowest fraud rate in the world.
You also have the worst customer
satisfaction score. So what you want to measure
at the same time as your fraud rate is your false positive
rate. That forces the team to actually
innovate. Similarly, you can
give recruiters metrics around hiring.
And guess what? You'll have a lot of people come in
through interviews, but if you're not tracking the quality
of hires, you may be very unhappy with the
quality of people you're interviewing or the people you're giving offers
to. So you always want to create the opposite
as an indication and measure both. And the people
who are responsible for that team need to be
measured on both. Finally,
around metrics,
one insight I've had over
my career is you
kind of want to look for the anomalies.
You don't actually want to look for the
expected behavior. So a famous
example was PayPal.
None of the top 10
markets that the company was planning to
go after included eBay.
But one day, someone noticed that
54 power sellers had actually
handwritten into their eBay listings,
please pay me with PayPal, and
brought this to the attention of the executive
team at the time. The first reaction of the
executive team was, what the hell is going on? Let's kick them
out of the system. That's not our focus.
Fortunately, I think David
Sachs came back the next day and said, I think we
found our market. Let's actually build tools
for these power sellers. Instead of forcing
them to write into their listing, pay me
with PayPal, why don't we have an HTML button that
they can just insert? Well, that started
to work. And then he said, well, actually, why don't
we make them insert it? Why should we make them insert
it all the time? Why don't we just auto-magically insert it
for them so they can insert it once, and then every
single listing they have forever,
it'll just auto-magically appear there.
That became the success for PayPal.
Similarly, I was at LinkedIn,
and I saw this stat that made no sense to me.
The UI of the site was a little bit different
back then, but 25%
of all clicks, maybe 30% of all clicks
from the homepage were people going
to their own profile. That made no sense
whatsoever. It was the settings.
You had to literally go to the right margin,
find a link, and it was 25%
to 30% of every single click at scale.
This is statistically valid stuff.
It made no sense whatsoever.
Never seen a UI perform that way.
I kind of
went around for weeks trying to figure this out,
and then someone very smart, actually
Max Levchin, said something to me.
He's like, it's vanity.
I'm like, aha.
People are looking at themselves in the mirror.
That's a pretty good answer.
They weren't editing their profile. Nobody has
something to edit every day in their profile,
but they actually were just looking at themselves
in the mirror every day because it made them feel good.
Then you could actually test that hypothesis and say,
well, if I have more content, do I look at myself in the mirror more often?
It turns out you did. If you had more endorsements,
did you look at yourself in the mirror more often? It turns out you did.
We actually figured out what was actually
underneath the utilitarian product
that the product team thought they were building
was actually a lot of emotional vanity.
It didn't exactly translate it to the best
possible feature like the PayPal example,
which you couldn't easily put a button that said,
be more vain today on the homepage.
That would probably not work perfectly.
It never really took off the way the PayPal example did,
but it really clarified
what users of the product really wanted.
We wouldn't have found that
without looking for anomalous data.
The final topic I want to talk about
is details.
In my assigned reading, there's a great book
I really like by Bill Walsh
called The Score Takes Care of Itself.
The basic point
of the book is that if you get
all the details right,
you don't worry about how to build
a billion-dollar business, how to have $100 million
in revenue, how to have a billion users.
That's a byproduct
of getting all the details of what you do every day
to be excellent. The example he
talks about in the book that really resonated
with me was he took over the 49ers
in 1979. They were the worst team
in football. I believe they were 2-14
the year before, which is really bad for those of you
who don't know football. In the next 10 years,
he managed to transform the team into the NFL's
best, won three Super Bowls.
What's the first thing he did
to start going
from a terrible team to the
best team ever in many ways?
He actually taught the receptionists how to answer
the phone properly. He wrote a three-page
memo about how to actually answer the phone.
That may sound absurd,
but his point was
if the organization as a whole does everything
exactly the right way, then receivers,
for example, will start running their
routes at 7.5 yards,
not 7 yards, not 8 yards, and that actually
will matter. If everybody on the team
executes exactly up to the same standard of
performance, you will have an organization
that is performing at the highest possible
level, and then with enough random variation,
the highest possible performance team will do
the best.
The way you translate this to a company are
to a lot of details that may not matter. It may not
seem that they matter superficially. Most people
would agree that details matter
when it faces the user,
but where the real debate is on things that don't
face the user. Steve Jobs famously
in the Mac insisted upon an immaculate
circuit design board. You can read about this in various
books. The Mac, for those of you who
don't remember the Mac, probably everybody here, but some of you
may have seen it, actually couldn't be opened.
So the circuit board design couldn't be seen
by any single person in the world.
There was no way to open the Mac except the people who
worked at Apple, and Steve insisted that it be
absolutely perfect and beautiful.
That's the kind of detail obsession that
this sort of philosophy of building a company requires.
Examples that may be
more practical for you instead of circuit boards
are things like, what food do you serve
people? This actually matters
more than you might guess. When people don't
like the food you serve them, what do they do?
They go gossip. They go complain
to their friends. They go walk over to someone's desk,
and all of a sudden at lunch, what they're complaining
about is they're mostly spending time
gossiping and complaining instead of
brainstorming. So you don't have this serendipitous
idea matching another
serendipitous idea that creates a spark.
Instead, they're all wandering and willowing around.
So the best thing you can do is actually give people
the food they want or food that's good for them that makes
them more productive. So it may seem
a lot like this glorious job
you thought you had is actually more like running
around being a task rabbit for people, but
it's to take the things off their plate
that are a distraction so that they
can be high-performance machines.
If you take enough things away from
people that distract them and give them the tools
to be successful, all of a sudden your organization
produces a lot more. Similarly, another
example that's often got wrong is office space.
So one
natural instinct is when you need an office
to have an office manager or someone on your team
go out and find offices and they'll go on tours with agents
and they'll come back with photos and
ideas. You need to do that yourself.
The office environment that people live in
and work in every day dictates
your culture and how people make decisions.
It dictates how hard people work. There's
almost no important more decision other than
what company you're going to be. Then what's the
office environment you're actually in
and most people don't do that. And then the final
thing and then I'll take some questions
is around
effort. Ultimately, I don't
believe that you can build a company without a lot
of effort and that you need to lead by example.
So Bill Walsh, the first chapter of
his book, gets
asked this question of how do you know whether you're doing your
job? And this is the answer he gives
coaches that used to ask him that question.
So if this is what you feel like every day
you're probably on the right track. And if that doesn't
sound appetizing, you probably shouldn't start a company
truthfully. Alright, with that, I'm done
with the prepared part. Let me see if
anybody has questions that I can try to be helpful with.
Any questions?
Yes. So
you talked about making compensation
transparent. How would you do that
especially when people
equate themselves to the value
of how much the salary is?
I would do it in probably
bands. You could do it either just everybody in the
company gets paid the same or you could have like
all discipline, all engineers or you could do it by
experience like
the way Steve did it actually at Next was
there's a high band and a low band. And you either had
a lot of experience or low experience and that was it.
So low band back then, you know, now
it would probably be like $85,000
kind of everybody just flatly gets paid
$85,000 and if you're super experienced everybody gets paid like
$130,000 and that's just it. Like it's
sort of the next translated for inflation.
I mean, like
instead of food, what's the like top
three details that people care about?
You'd be amazed.
Well, yeah, so the question was besides
food, what other kind of details do people care about?
The laptops they use. I mean
this is now the default that everybody
has, but five years
ago, it was a benefit
to give people high quality machines as
opposed to optimizing our costs and having Dell
machines and ugly monitors just as an example.
So if you think about all these people
that are relentlessly resourceful, incredibly talented
in a massively competitive
ecosystem competing for talent, you want
to give people the best possible tools to do
the best possible job. And so right
rigorously thinking through how do I
make people more successful? What things
do they not need to be working on that
are distracting? And what things can I
actually give them that make them more valuable per
day? And then just break that down every
day and solve that stuff yourself.
It's a good question. I actually think that you should
start, first of all, you must have your own office.
I don't believe ever in shared office spaces.
Peter talks a little bit about this,
that every startup, every good startup is a cult.
And it's really hard to create a cult if you're sharing
space with people, because a cult means that
you think you're better than everybody else in the world, and you have a special
way of doing things that's different than everybody else in the world.
And if you're sharing physical space with people,
it's very hard to inculcate that. So I
would start there. But it is a
prioritization question. Every
company has a scarce resource. It's just a question of
magnitude, like how many zeros
are you paying attention to? Probably
not $10 expenditures, but $100,
then $1,000, then $10,000, then $1,000,000,
then $10,000,000 starts being your rounding error.
So what I would do is
figure out what's most important in a high
quality office that creates a good vibe
that allows you to recruit people, because recruits are very
savvy about this. They walk into your office,
and they can tell a lot about the culture instantly.
I walk into a company
office, and I can tell often whether I'm going to
invest as soon as I walk in.
I could absolutely rule things out that
I just don't want to invest in as soon as I walk in.
And there's times walking into the office, it's like, wow,
this is really impressive. You can tell how
people work together, how hard they're working, how
distracted they are. Roloff Botet
at Sequoia made a point to me about YouTube.
So when I invested in YouTube
at the very, very beginning,
it wasn't obviously going to be successful.
And then Roloff
led the Series A investment for
Sequoia and YouTube, and we were at a board
meeting together, and he said, you know, I think
YouTube's really going to work. And I said, why?
And he said, well, every time I go to one of my portfolio companies,
half the office is watching YouTube at lunch.
And I was like, pretty good sign.
So you pick up on these little
things, and you can predict a lot.
What do you think is the best
way to gain street cred for a new
manager? Oh, boy.
Yeah, so the question
is, how's the best way to gain
street cred for a new manager?
Almost all good managers in
Silicon Valley are promoted because of their individual
performance. In cultures
that are meritocratic, the percentage
is even higher. So we tried at
PayPal to only promote people
that were basically kicking ass at their discipline.
So Peter didn't believe in
general managers.
In fact, I remember going for a jog around campus
within my first week at PayPal, and he's asking me
how things are going, the usual kind of CEO
questions. And then we got into this debate
about whether the company needed any managers.
And he's like, nope, no managers.
We're only going to promote people. So the VP
of engineering is going to be the single best engineer.
The VP of design is going to be the single best designer.
The VP of product is going to be the single best product person, and they're going to learn
to manage later. And the advantage
of that is you don't
demoralize people.
Because everybody knows that their boss actually
is better at their job than they are,
and they can learn stuff. And you can learn a little
bit of the management techniques later
as opposed to promoting people who are just
good people managers that
don't actually have the discipline and skill, and that
does demoralize people. So I think just
being excellent at something, and
then getting excellent at
getting a bunch of people to do something is the next task.
But people,
you just have to learn. Some things you have to learn by doing.
You can't learn to play the guitar by reading a book.
You've actually got to try to manage a bit,
and you won't do it well. I have another sort
of set of tactics and classes
on what you actually do when you transition
from individual
contributor to manager, and it's hard. One of the
first things people don't get right is their time
allocation. And so actually I
would recommend doing what
I call a calendar audit, and
tracking for a month what you spend your time on
and how much is managing and editing, how much is writing,
et cetera, and then optimize
it over time. You can get a mentor,
find somebody who's been a manager before
that will work with you, not your boss,
because your boss has a set of complicated
objectives, including
how much are we shipping. A mentor
can just focus on you and making you more
successful.
Yeah, so
I would look at every
piece of copy in every department.
So like another area that almost never
is done, so look at your recruiting website,
almost never done to the same quality as your
conversion funnel. Look at customer
support, another classic area
that isn't up to the same quality, and treat
customer support like a product
so that you actually have an engineering team and a design
team over time that actually focuses on
making that world class.
Usually where you have different executives at a
scaled company, most executives
were trained differently at different companies, and they bring
some of that with them, so you've got to cross-train that.
So if you've hired a VP of engineering from Google,
it's very different than a design leader who came from
Apple. They don't actually learn how to do anything
the same way, so you're going to have
to stitch that together somehow.
Either one or the other is going to have to learn in the other style,
or you're going to have to create your own style and really
teach that to your executives.
It shows up all the time, but all you
need to do is just pick up the company's products
and look for things that have a different
voice, and you can see it. Visual voice,
word choice, all over
the map.
So the canonical
advice, and this now sounds obvious, but it actually
was pretty radical in 1982 when Andy
wrote his book, is you should have a one-on-one
roughly every two weeks.
Some people say every week, and
I don't think you want to go longer than two weeks.
One week can be ideal, actually, in many
companies. The reason why, in fact, there's
another adage, which is you should only have
five to seven direct reports, that actually
derives from the concept of
a one-on-one every week, so that
the direct reports so that you can fit enough
one-on-ones in your calendar and still get other things done.
So I think one-on-ones per week
are a good idea. The agenda
should be crafted by the employee
who reports to the manager, not the manager.
The one-on-one is mostly for
the benefit of the employee, so they should
walk in with, here's the three or four things I
want to talk about. Ideally, they circulate
that, it can be even bullet points in advance by
email, but in advance so you have time to chew
on it and you're not on the fly
winging your responses, but that's probably
the best structure. Now, if someone's really
good and really talented and has been doing something for
a long time with a lot of internal credibility,
you might push out the one every week
and relax that to one every two weeks, possibly once a
month. I don't know that would go beyond
once a month ever.
When is it acceptable to compromise and hire someone
who's ammunition rather than a barrel?
Yeah, so the question is
when do you compromise and hire more ammunition
instead of a barrel? Truthfully, you're going to
hire more ammunitions by definition
than you are barrels. So there's a ratio
between the two. The question is the ratio.
So at some point, the ratio
is going to get out of whack. So if you're the only barrel
in the company and you have 50 engineers,
you might as well only have 10 engineers because you're not going to get
any more done. So you're just wasting resources.
You're going to frustrate engineers because everybody's going to
need your approval, your sign off, your
editing. It's just going to stack and frustrate
people. So it's really,
different disciplines, engineering,
I actually think roughly
1 to 10
to 20 is probably about the right
range.
You don't need more engineers
until you have more barrels.
Designers are a little different, but
it's always going to be hiring more
ammunitions. And a good leader
or a good barrel will kind of have a
feel for that. So one way to
correct for this. The actual tendency is to increase
your headcount on your team.
That's sort of an empire building tendency.
So I manage 20 people and Sam only manages
10 and you manage 3. So I'm more important than
Sam and you're more important than me. One way to do
that is you put an X here of the number of people that
the output, sorry, specify
how many things they've done successfully
and then divide by the number of people on their
team and tell them that this is going to be their grade
and their performance review. Shockingly
this Y doesn't start increasing on that team.
It's amazing how this works.
And be really explicit about it.
How often do you
intervene for your company
and what's your philosophy?
So how often do I meet with them?
So the question is as a venture capitalist
how often do I interact with my companies, meet
with them? Generally when
we invest, we do
do some seed investments where we invest
less money, but when we invest a fair amount of money
and lead a round, like a series A or series B
round, we join the board
and roughly I meet with
the founder CEO every two weeks. That's
the default. Now there can be
obviously there's inflection moments and things go
right, go well or wrong
and that's on an ad hoc basis.
These days actually surprisingly
I do a lot by text message.
I even have one CEO who Snapchats me all the time.
I'd actually
rather not, but
the world
has changed a lot, but I try
in-person meetings every two weeks.
But you don't meet them in a separate location?
No. I mean really it's like
being a venture capitalist is more to me like being
a psychologist. So if you come to my
office you'll see actually we have two chairs kind of arrayed
like this with a little table in the middle and
we sit down and basically like, so tell me
your problems. It's like a question.
And then
my response is usually, well have you thought about this?
Have you talked to this person? Have you tried this?
Etc. And it's just asking a lot of questions
and going back that way, but that's 90% of what I do.
So a CEO
when you've got these parallels, right?
Let's say if you're a startup company, you've got two
really good engineers and
you're working on this product.
I've heard that you should always be recruiting.
Yes. How do you balance that?
Well it depends on where
your prioritizations are. So
Sam talked a little bit about this in his lecture, but
every company will move recruiting
first, second, or third somewhere
in that sort of spectrum.
If it's your number one priority, then about
25% is probably a pretty good allocation.
Actually I like the calendar
audit for CEOs even more so
than for new managers. So when
I work with CEOs, sometimes who aren't
thriving in that role for the first time,
I actually force them to show me
their calendar. And before I do that, and now
I'm going to ruin this trick because I'm going to tell it
to you, I actually ask them to rank their priorities.
Write them down on a piece of paper and
just specify whatever they are. Then we go pull
up their calendar and see if it matches. And it never
matches. Like never. Like recruiting
is the one that's most often
awry. So let's say half the CEOs
you meet with will say recruiting is their number one priority.
It's almost never the biggest block of time
on anybody's calendar. And so
that's what you're trying to do is match resources
and inputs against priorities
and a calendar audit. Unfortunately there's no software that
does this really well. It would be great.
Like actually we literally pull up someone's Google
calendar and kind of manually add up the hours
which is somewhat insane. But
that's the best way. It's just what are your priorities?
If your priorities are raising money,
you don't want to allocate 100% of your time to recruiting.
You want to allocate a fair amount of time to
raising money for that block of time until you're
successful at it.
One more question. Anybody?
Okay. Go ahead. On the surface
some of this advice seems contradictory
because you emphasized
focusing on A-plus tasks
and delegating less important things
but then you said it's good to
write a three-page script for your receptionist
and focus on all the details.
How do you harmonize those for us?
That's a good question. I think the way to harmonize, the question is
really how do you harmonize things like where details
can really matter but you only get
one thing to do and you've got
to allocate to the top one, two, or three
things. How do you put those
two things together? And actually there is some tension
and even in a healthy organization
there's some tension of like why are we actually
focused on writing the script
as opposed to something a user may see.
I think the underlying
philosophy of getting the details right
is pretty important to install in the very
very very beginning of a company
because people will start acting that way
and making decisions that way themselves.
So you won't have to actually literally do that
and if you have to do that it actually shows
that the foundation isn't actually that
solid. So when you first start
the company it's about
getting the details right, that everybody
is precise, everybody on every task
is always thinking that way
and then that scales and then the people you bring
in, it's a little self-fulfilling, people who can
think that way will tend to get hired, people who can't
won't get hired. Each team and each
leader will tend to enforce that themselves.
The CEO is almost never doing it.
So it's partially like how do you start.
Cultures like that, I mean the key to culture
is it's a rule, it's a framework for making
decisions and if you have a culture baked
people learn how to make decisions across that culture
and you're never saying anything. You never have to really
do anything except just
watch and
promote and move people around.
Cool, well I guess that's it. Thank you.
