Sam Altman sits down with Reid Hoffman to talk about the key changes founders should think about as they transition from a startup to a scaleup.

This conversation took place at The Scaleup Offsite, a private CEO gathering hosted by Y Combinator Continuity and Greylock Partners in April 2017.

Everyone here is an important part of our joint network.

First, Sam had this brilliant idea of teaching a startup class at Stanford and getting a bunch of different founders to come and talk about kind of the key elements.

And, you know, he got this, he got me to give some talk on how to be a great founder, I think, was that 1.

And then I went, wow, that's a really good idea, and part of how innovation, I said, well, startup's 1 part, scale up's the other thing.

And so we did a similar class with teaching blitz scaling, which is the theme of a book that I'm working on with Chris Yeh, who's in the audience somewhere.

And so we did a Stanford class as well, where we got a bunch of different folks.

And then Sam and I started talking about this, and we said, look, actually, in fact, this is not just a kind of a good Stanford topic, but this is actually a good topic for, essentially, our network and for the businesses.

Because you really only build something amazing if you hit the scale problem.

It's not just a, hey, you invent, and then, hey, it's easy after that.

We hit the oil, and we just pump out the oil.

There's actually, in fact, a lot of innovation, and a lot of hard work, and a lot of skill and craft that goes into scaling and if you fail you fail And so then Sam called me and said, hey, we should do an event like this and we should start doing this as a regular basis.

So I thought I would open a little bit with some of the blitzscaling stuff.

And then we'll just kind of kick back and forth.

So, oh, and then the other thing, by the way, actually 1 other little housekeeping thing.

The stage conversation, including the Q&A, is essentially going to be published.

The hallway conversations are private, so don't tweet or otherwise publish anything in the hallway conversations unless you have permission from the person you're talking to.

So With that, 1 of the key things that I think is interesting, as you say, well, why is it that Silicon Valley produces so much of the technological impact in the world?

We tend to tell ourselves still classically this kind of startup story of, well, we're inventive and we have venture capital and we import a ton of talent here and the venture capital stuff all works.

But actually, in fact, it's the fact that we intersect technology invention with business invention where that business invention includes such things as network effects, new business models and so forth.

And then in seeing those, we realize that certain of those businesses are super important and we move really fast to establish that business model.

And that's 1 of the reasons why you'll find that most people in the Valley all say network effects as something super important.

And 1 thing that actually in fact I think is a fun hobby, in case any of you would like to do this, is occasionally when I'm feeling impish, I will actually in fact ask people, well what's network effects?

Because a lot of people will use the term and actually don't actually in fact know what it is, don't know how to measure it, don't know the different kinds of network effects, don't know like, network effects in growth, network effects in engagement, network effects in revenue, network effects in a bunch of these sorts of things.

And yet, the reason why we're obsessed with network effects is we know that's the kind of thing that causes us to scale.

Now, and so part of the whole thing is a whole set of different techniques.

And I'll go through some of the different blitzscaling areas of hiring, but maybe 1 of the things we should start with is how you've been scaling YC, actually, because of the fellowship and the class and the MOOC and everything else, because it's not just that we kind of advise the stuff and practice, you practice the stuff.

Well, you know, it gets super recursive because 1 of the things that we did was start our continuity fund because we realized this problem.

To scale ourselves, We realized that if we didn't start a growth fund that could help companies who have had their initial idea start to work, then we weren't going to be able to produce as many impact companies as we can, which is really why we started our own growth stage fund.

We sometimes invest in non-YC companies, but the idea was, can we, to scale, we need to do 2 things.

1 is fund a lot more companies, but 2 is build a practice that in the same way we work with founders to figure out their initial idea, teaches them how to scale companies.

And so we've tried to build this thing, basically to scale YC, we need to scale companies.

And that, I think, has been 1 of the most important inventions we've had in the last year.

We've also tried to go in the other direction.

So right now we're teaching a MOOC, which is kind of the third version of that Stanford class, back at Stanford, where we are advising 3, 000 startups at once.

I think this is probably a world record for number of startups concurrently advised by 1 program.

And next year we think we can scale that to

And this is teaching people a lot of the YC experience about how to start a company.

And I think this is really important, right?

That there are, well as Reid said, not everyone knows what network effects are, but our version of the network effect is the bigger we can make our community, the more we can get people feeling loyalty to our community, helping each other.

And the thing that I really try to do in terms of scaling YC is make sure we don't have to compete with other firms.

You managed to do it very well, but most venture capitalists that have to compete have a tough time of it, spend a lot of time thinking about it.

And we like to be the only people that are sort of, that really matter in our own space.

And so for us, scale is an answer to that problem.

And the more we can scale, the more founders we can help, the bigger we can make our network.

We've done a lot of other things around the edges, but you know, kind of the big strategy is very simple.

It's have a very big top of the funnel, get to the most promising new founders at the very start of their career thinking about a startup, get the best ones of those into our program, build a program with real network effects, and then have this continuity fund to help invest in these companies and teach them to scale.

And so 1 of the key things, I think, to understand about scale is kind of the what got you here won't get you there.

And so classically, Too much of the advice that's given to entrepreneurs is, like for example, when you're 20 people, hire your scale executives so that when you're at thousands and so forth, you have all those people baked and in place.

And very rarely are the right people, when you're executives at 20 or 50, are the right people for executives when you're 1, 000 or 500.

Some of them grow with it, but most often, actually, in fact, you need to trade around.

And so part of the thing about thinking about scale the right way is to think about dynamism, to think about the fact that you're refactoring your org as you're going, and that you need to be anticipating that you're refactoring your org.

You need to be anticipating that, you know, for example, you're looking at your set of folks, hopefully, if you've done it well, you've got a bunch of well-cohesive, strong A players, but A players at this stage are not necessarily the same as A players at the next stage, and you have to be 1, managing your connectivity with those folks in a way that, if you're promising this person that they're going to be head of product or head of sales or head of marketing or head of engineering forever, and that goes away.

Then when you break, you break them, or you may actually still want them in the org.

What you actually want to be promising is things like, you will actually have a seriously important role in this organization forever.

You will be a major contributor, your job will increase, you will learn things, but not necessarily unless you're pretty sure they will be, you will be in charge of this function.

And so that changed because when you change the scale of an organization, you're moving from, when you're 10, 20 people, you're all doers.

A couple of you are managing too, but you're basically all doers.

you move from doers to manager doers, to managers and doers, to managers of managers, otherwise execs, and that's changing the whole dynamic.

And so you have to anticipate in scale that you're changing around the way that your company actually works is kind of a key thing, and you need to be anticipating that that will be coming.

I had a funny example of this late last year.

Someone was writing a story about YC, and they wrote something about how we just had our second major reorg in 2 years.

And I read that as, oh, they're saying we're not moving fast enough.

And it's this, and I stand by my read of it.

But I think the amount of dynamism that the startups that really scale well have always looks a little chaotic on the outside.

It's just because there's constant shuffling of new people in and people who aren't working out out.

The very best founders that we've worked with, 1 of the things that they always evaluate their people on is how good have they been at training their replacements, how good are they at, you know, on a moment's notice, being able to move to a next role.

Because most founders are very bad at this, and Most people are very bad at this.

There was a story at Facebook at 1 point that they tied like 50% of executives bonus to how good of a job they were doing at having replacements ready to go and being ready to move into a new role.

And I think if you don't really incentivize it, you never get enough of that culture in place.

And if you're going to be blitzscaling, you're gonna just be moving people around and reorganizing constantly.

Yep, and part of the, there's 2 corollaries, which Sam was just illustrating on that dynamism.

So 1 of them is that actually the chaos isn't just external, it's internal.

And so tentatively what happens is people complain about the chaos internally, which by the way makes sense, operational efficiency and everything else.

And you have to condition the organization to say, when we're scaling really fast, there will be chaos.

We're trying to constantly manage it, but we don't manage to 0 chaos until we're actually in a relatively stable place.

And so expect that, work collaboratively with each other, and make that happen.

So it's not just external looks at chaos, it's internal chaos.

And you have to just, like, the wrong way to address that is to say, okay, my team says they don't want chaos.

I'm gonna try to manage out all the chaos.

You know, it's okay to have a little bit less, and that's a noble thing to shoot for, but the trade-off, you know, people that run these perfectly non-chaotic organizations somehow never build great companies.

So the trade-off is, we're gonna accept a little bit of chaos in exchange for a shot at 1 of these massive great companies, and founders just have to sell that to their teams.

Chaos will go up, you do some solidity, chaos will go up, you do some solidity.

The second part, and this is 1 of the really key things that I learned from my first startup.

This was kind of a classic entrepreneurial mistake.

I kind of approached this with what you, you know, actually.

I always kind of make vaguely and teasing comments of MBAs.

It is a classic MBA thing, which is like, oh, hire the people who have the experience at that job, get the CV of it, put that CV in, make that happen.

And so I literally had like job description to history and was optimizing for that.

And it's a total fail as you get to scale, because the real thing you need is to have the people who learn and who adjust.

1 part of that is learning and adjusting through what's the needs of the business, what's your product market fit, how is that working.

The other parts of the learning and adjusting is the whole organization is going to change shape.

And so if you don't optimize for people who are learners, who will go into that, you're a host.

And so this is actually part of when I got to PayPal and was, you know, kind of when Peter and Max and I were going on walks about like how to first build what was first called Fieldlink, then Confinity, then PayPal.

We were doing these walks and I said, no, no, don't look for, oh, I've got 10 years of experience QA, look for someone who learns fast.

They need to have basic skills, because learning it from scratch, too difficult.

But look for that learning curve, and look for team sports, and look for other kinds of things.

And for those of you who know Peter, you know he was particularly irked at the team sports thing but later turned out to be right.

That was the individual achievement, playing chess is what matters, not the importance.

My version of this is you want to hire for values first, aptitude second, and skills third.

And I think the problem with most executive recruiters is that they reverse that order.

As your, and so if you say I need to hire a CFO, they will go bring you people with 20 years of experience as a CFO at vaguely similar companies, but unfortunately it's usually like 20 years of the exact same year of experience over and over again and they probably can't learn and adapt if something really changes.

If you're trying to scale fast, you really need someone who is aligned with the values of the company so that when Things change or when they have to make a decision They'll make the 1 you would make if you can't be in there or they will be a good player a good team player And go do that thing you need someone who has high aptitude Because the role is going to constantly shift and the speed the ability the rate of learning the rate of improvement dominates Skills and then you know skill specific skills experience obviously matter.

Speaking of YC Continuity, Oli, who runs our continuity fund, had never had any venture job before and has done a fabulous job.

Same thing is true for a lot of the other people that we hire.

I think if you free yourself from this sort of the traditional model of how you hire executives in a non-growing, very static company and think that in a blitzscaling company you have to flip the priority of skills, you end up, or the priority of attributes, you end up with a very different team, but it's the 1 that works.

And the very best companies take this exceptionally far in how much they're willing to hire a non-traditionally qualified executive.

Yes, and actually, I mean, there's a bunch of different parts of hacks on this.

1 of the things that I actually learned from when Sean Parker and Zuckerberg hired Kohler from LinkedIn, and I was talking to Kohler about this, the thing they put them in is he's a generalist, and they put him in a recruiting role first.

I was like, look, this is the most major thing.

And so we're not hiring a traditional recruiter.

And that was actually 1 of the things that really helped set the initial talent team and culture as it scaled.

And it was that kind of thing about thinking, which are the things that you most need to solve, and then getting the generalist into it.

Because 1 of the things that you do need to eventually as you scale, hire specialists, but the theme that you look at as you move from startup to scale up is you always have generalists.

The generalist are important for learning, changing the organization, adapting, adapting product market fit, tackling new markets, going global, et cetera.

But you still very selectively add in the specialists.

When you have someone who says, look, I am just a network engineer, you're like, OK, great.

That's when we know that what we need from you is network engineering for the entire tour of duty, the entire length of time here, and that's fine.

The rest of it, you're trying to have generalists as much as possible.

And matter of fact, actually, another part of the, that was a learning from social net, another of the person who first did that role linkedin is in the audience Lee Hauer was our first like okay utility player Make sure you can tackle any problem that we that we we throw at you So I'm gonna move to culture.

We have a great talk from Jason Kylar who will be up after us and is 1 of the world-class folks on culture, So I don't think we should go overly in-depth, but I think it's worth touching quickly.

I recently did, I'm doing this podcast series on masters of scale with June Cohen, it's coming out, I think it's May 3rd, and it's kind of this heavily edited thing on these different themes of kind of what are the different theories of what it takes to scale.

And I had a conversation with Reed Hastings where I asked him, I said, look, there's 2 theories of thought in Silicon Valley.

1 theory is that culture eats strategy, and that culture is the dependent thing for how you really build great companies.

And then the other theory is that actually culture is the historical explanation for successful companies.

That when you have a successful company, you look back and say, that culture, that was really great.

And Reid looked at me and said something that I normally say to other people, so this was kind of a funny, embarrassing moment in the interview, which was, he says, well, both.

It was a clever way of asking the question, but it wasn't so clever in terms of the answer.

But I do think that the theory that's pro-culture is that, again, when you look at the traits of you're gonna be moving up at such a fast rate that you're gonna be having chaos, you're gonna be reorganizing fairly consistently, and as you reorganize, especially managerial roles, that resets a bunch of patterns and you'll have some chaos in that.

How do you keep a high-performant, very strong company?

And culture is in part the answer to that.

Because if your only culture is top-down hierarchy and messaging and communications that comes from the top, that culture won't survive as you really balloon the organization.

Instead, you want a horizontal accountability.

You want it so that everybody is keeping everyone else accountable to the culture that we're in.

And cultures are not like there is the good culture and the bad culture.

Some cultures are engineering cultures, some cultures are high IQ cultures, some cultures are collaborative teamwork cultures.

There's a different set of things you're emphasizing and that should be dependent on your organization and the problem you're solving.

The problem is is that it doesn't just end there.

You actually do have to play good strategy in everything else.

And so the thing that when people frequently sell culture too much is they say, once you have good culture, everything else follows, because you hire the right people, you play the right way together, and that's of course extremely important, but you don't get the business model and the strategy right.

What's the particular way that you guys angle on this?

You know, I think There are so many different kinds of cultures that can work and I'm certainly a believer that you have to have a great business and a great culture and they can be somewhat orthogonal, but also if you put 1 business's culture, if you take 1 business's culture and throw it into another successful business, it might be disastrous.

So they really do somehow have to fit together.

The general framework that I think about when adding people to a team, which is I think what defines culture, is what is going to be this person's net effect on the output vector of the organization.

And there are people who are brilliant and can get a lot of work done themselves and would add to the vector in that sense, but they piss everybody off so much they have a net negative effect because they make other people less productive.

There are people who are moderately intelligent and moderately productive but have a hugely positive impact on what they get everybody else to do.

And I think 1 version of culture is are you good at bringing in people that have a positive net effect on the output vector?

Which screens out just looking at people in a vacuum.

I think though that it's really easy to use in Silicon Valley in the current environment, it's really easy to use culture as an excuse for underperformance.

And so you have people say, you know, we drink a lot of green tea, we all go do yoga together, we're super nice to each other, and we have this wonderful culture, and they don't understand why people don't want to come work there, or come and leave and aren't productive.

And the answer is, the culture that matters, I think, to the best people is 1 where they can just come and be really productive and be around really other great people.

And if you have a culture which looks good on the surface, but somehow rejects super talented people or it's just covering up for constant infighting, I think that can be a real problem.

I think 1 of the things going on in the current

Silicon Valley is it's very easy to get entitled employees.

It's very easy to get entitled people at a company.

Everyone wants to work exactly how they want, they want to be really rich right now, if the company's not going to get liquid next year, they're going to go somewhere else.

We looked once at the average tenure of employees at companies in San Francisco, and I sort of couldn't believe it, it had been trending down, down, down, I couldn't believe it was going to get any lower, and it managed to subsequently get lower.

And so, you know, this idea of how important it is to have people that are going to join a company and stay there for 5 or 10 years, sounds like a crazy thing to say because no 1 does that anymore.

And yet somehow at the best companies that does still happen.

And I think if you think about what do we have to do to get the best people to stay at our company for 5 or 10 years, and then go make it at the culture, which 1 of the things that is included in that is wild success for the company and the mission that people care about.

That's really important, but getting people who are there for that and not there for the green tea, which is like an, I think an easy way people think about culture in the wrong way at the beginning is super important.

You need to create an environment where really great people will wanna come, work with each other, and not have to deal with the crop that they do at most companies.

Just in case anyone's under illusions, although I think very few people here, culture isn't benefits, culture isn't food, it isn't kombucha, right, et cetera.

Unfortunately, that's what boards tell you when they come ask you how the culture is going.

What it is, is how are you holding each other accountable to the mission and to the way that you work and to the way that you have high performance.

And 1 of the things that I think is super important for this is actually in fact I think too few organizations follow in what Hastings did in terms of creating a culture deck.

Now the thing that's interesting is that the reason this deck got there, And I actually think every company should do some version of that and some version of publishing it.

The reason Hastings did this is they first started with kind of studying how people bounced out.

They interviewed and hired really good people and then they came in and the people came in and they said, hey, we're not a family.

And we should basically say, if you have adequate performance, we give you a generous severance package and have you leave.

And people say, wait, wait, I thought I was joining a family.

And I said, OK, well, let's define what it is.

And I said, well, how do we shape the funnel of people coming in?

And then of course after publishing it, it actually really helped a whole bunch of people understand like, oh, that's that kind of culture and should I work there and that's really, I really like the fact that they're shaped that way, that that particular way of playing, that particular sports team, that's the sports team that's the right kind of thing for me.

And I think many more organizations should actually be much more explicit about that because they don't talk about like, oh, we have cafeterias, we have a volleyball court.

That's not in the culture deck, right, as a way of doing this.

The traditional problem, we see a lot of these value statements in culture decks, and the problem is they're only valuable to the degree they're different from what other companies say.

So most of the ones we see, because we suggest that every company write this out, and it's like, we value integrity.

And so I think this is a worthwhile exercise and it's good to do, but only to the extent that what you say is different from what other people would say.

And trying to figure that out is really important.

And honestly, what most companies find when they do this, or what many companies find, is that they have nothing to say that wouldn't be in sort of the macro expanded template of what a culture is, what a good culture is.

And when I talk to folks, I'm trying to get them to sharpen their mind is what filter would have A players working at other companies than yours.

Which A players do you not want working at your company?

And that, and that, that shows you that you begin to have something that's a kind of an edge in a culture and begins to approach the uniqueness.

Now another part of scaling is hiring changes.

people, founders, CEOs interview everybody, right?

Because part of the way that they try to help the culture say is they say they're doing that check.

Matter of fact, Neil Bustry, when I was talking to this, I think he and Dave Duffield interviewed everybody up to 500 because they were the final culture interview.

They presumed that everyone had done the skills and other kinds of things, and they were the, are you the right fit?

That the process, again, what got you here doesn't get you there, because now you need to start hiring a bunch more people.

You need to be trusting people in the rest of your organization to hire well.

Like 1 thing, as you begin to get from the, call it the hundreds into, you know, early hundreds into the late hundreds, you'll actually start having onboarding classes.

Like when you, someone will say, okay, we hire people and then all 15 people start at the same day so that we, we train them on our company in the same pattern in order to make that efficient and so forth.

And that's part of this whole hiring process.

So you also not only have to think about org, but also how you're hiring, what's the way you're doing it, and that itself is a scalable process, not just from sourcing and from interviewing, but also all the way into integrating.

So that's another way to look at the culture in hiring.

1, I'm gonna shift topics to communications within the company.

Classically, when you're at an early stage, you're all in the same room.

Almost like on all hands is we just happen to all be in the same room at the same time, and that's the way it works.

Then you begin to build, and part of that is that begins to shift to everyone has a complete dialogue to be on the same page.

And you have to start changing the pattern at which information is being both expected and communicated.

See all Y Combinator transcripts on Youtube

From Startup to Scaleup | Sam Altman and Reid Hoffman