Transcript: Lecture 1 - How to Start a Startup

Sam Altman

Welcome to CS183B. I’m Sam Altman, President of Y Combinator. Nine years ago I was a Standard student, and then I dropped out to start a company, and then I’ve been an investor for the last few.

So at YC we’ve been teaching people how to start startups for 9 years. Most of it’s very hands on and specific to the startups, but 30% of it is pretty generally applicable, and so we think that we can teach that 30% in this class and even though that’s only 30% of the way there, hopefully it’ll still be really helpful.

We’ve taught a lot of this at YC already but it’s all been off the record, and this is the first time that a lot of what we teach at YC is going to be on the record. So we’ve invited some of our best speakers to come and give the same talks they give at YC.

We’ve now funded 720 companies, and so we’re pretty sure that a lot of this advice is pretty good. We can’t fund every startup yet, but we can hopefully make this advice very generally available. Guest speakers are going to teach 17 of the 20 classes, I’m only teaching 3.

Counting YC itself, every guest speaker has been involved in the creation of a $1B+ company, so the advice shouldn’t be that theoretical. It’s all from people who have done it.

All of the advice in this class is geared towards people starting a business where the goal is hyper-growth and eventually building a very large company. Much of it doesn’t apply in other cases and I want to warn people up front that if you try and do these things in a lot of big companies or non-startups, it won’t work. It should still be interesting; I really do think that startups are the way of the future and it’s worth trying to understand them, but startups are very different than normal companies.

So over the course of today and Thursday I’m going to try and give an overview of the four areas that you need to excel at in order to maximize your chances of success at a startup. And then throughout the course the guest speakers are going to drill in to all of these in more detail.

So the four areas. You need a great idea, a great product, a great team, and great execution. These overlap somewhat but I’m going to have to talk about them somewhat individually to make it make sense. You may still fail. The outcome is something like idea × product × execution × team × luck, where luck is a random number between 0 and 10,000. Literally that much. But if you do really well in the four areas, you can control. You have a good chance of at least some amount of success.

One of the exciting things about startups is that there’s a surprisingly even playing field. Young and inexperienced? You can do this. Old and very experienced? You can do this too. And one of the things that I particularly like about startups is that some of the things that are bad in other work situations, like being poor and unknown, are actually huge assets when it comes to starting a startup.

Before we jump in on the how, I want to talk about why you should start a startup. I’m somewhat hesitant to be doing this class at all, because you should never start a startup just for the sake of doing so. There are much easier ways to get rich, and everyone who starts a startup always says, always, that they couldn’t have imagined how hard and painful it was going to be. You should only start a startup if you can feel compelled by a particular problem, and that you think that starting a company is the best way to solve it. The specific passion should come first and the startup second. In fact all of the big successes we have at YC followed this.

So for the second half of today’s lecture, Dustin Moskovitz, the co-founder of Facebook and Asana, is going to take over and talk about why to start a startup. We were so surprised by the amount of attention that this class got that we want to make sure we spend a lot of time on the why.

A great idea

OK, so the first of the four areas: a great idea.

It’s become popular in recent years to say that the idea doesn’t matter. In fact it’s almost uncool to spend a lot of time thinking about the idea for a startup. You’re just supposed to start, throw stuff at the wall, see what sticks and not even spend any time thinking about if it will be valuable if it works. And pivots are supposed to be great, the more pivots the better.

So this isn’t totally wrong, things do evolve in ways that are difficult to predict, and there’s a limit to how much you can figure out without actually getting a product in the hands of users. And great execution is at least 10 times more important and 100 times harder than a good idea, but the pendulum has swung way out of wack here. A bad idea is still bad, and the pivot-happy world that we’re in today feels really sub-optimal. Great execution toward a terrible idea will get you nowhere.

There are exceptions of course but most great companies start with a great idea, not a pivot. If you look at successful pivots, they almost always are a pivot into something the founders themselves wanted, not a random made-up idea. Airbnb happened because Brian Chesky couldn’t pay his rent, but he did have some extra space. In general though if you look at the track record of pivots, they don’t become big companies.

I myself used to believe ideas didn’t matter that much, but I am very sure that’s wrong now. The definition of the idea as we talk about it is very broad. It includes the size and the growth of the market, the growth strategy of the company, the defensibility strategy and so on.

When you’re evaluating an idea you need to think through all of these things, not just the product. If it works out, you’re going to be working on this for 10 years so it’s worth some real upfront time to think through the longterm value and the defensibility of the business.

Even though plans themselves are worthless, the exercise of planning is really valuable and totally missing in most startups today. Longterm thinking is [unintelligible] anywhere but especially in startups, that it’s a huge advantage if you do it.

Remember that the idea will expand and become more ambitious as you go. You certainly don’t need to have everything figured out and a path from here to world domination but you really want a nice kernel to start with. You want something that can develop in interesting ways.

As you’re thinking through ideas, another thing that we see young founders get wrong all the time is that someday you need to build a business that’s difficult to replicate. This is an important part of a good idea.

I want to make this point again because it’s so important. The idea should come first, and the startup should come second. Wait to start a startup until you come up with an idea you feel compelled to explore. This is also the way to choose between multiple ideas. If you have several ideas that all seem pretty good, work on the one that you think about most often when you’re not trying to think about work. But we hear again and again from founders that they wished they had waited to start a startup until they came up with an idea they really loved.

Another way of looking at this is that the best companies are almost always mission oriented. It’s difficult to get large groups of people to the extreme levels of focus and productivity that you need for a startup to be successful, unless the company feels like an important mission. And it’s usually really hard to get that without a great founding idea.

A related advantage of mission oriented ideas is that you yourself will be dedicated to them. It takes years and years, usually a decade, to create a great startup. If you don’t love and believe in what you’re building, you’re likely to give up at some point along the way. There’s no way I know of to get through the pain of a startup without belief that the mission really matters.

A lot of founders especially students believe that their startup’s only going to take 2 or 3 years and then after that they’ll work on what they’re really passionate about. That almost never works. Good startups usually take ten years.

A third advantage of mission oriented companies is that people outside the company are more willing to help you. You’ll get more support on a hard important project than a derivative one.

When it comes to starting startups, in many ways it’s easier to start a hard startup than an easy startup. This is one of those counter-intuitive things, it takes people a long time to understand.

It’s difficut to overstate how important being mission driven is, so I want to emphasize it one last time.

Derivitive companies, companies that copy an existing idea with very few new insights, don’t excite people and they don’t compell the teams to work hard enough to be successful.

Paul Graham is going to talk about how to get startup ideas next week. It’s something that a lot of founders struggle with but it’s something I believe you can get better at with practice, and it’s definitely worth trying to get better at.

The hardest part about coming up with great ideas is that the best ideas often look terrible at the beginning. The 13th search engine and without all the features of a web portal: most people thought that was pointless. Search was done and anyway it didn’t matter that much. Portals were where the value was at. The 10th social network and limited only to college students with no money, also terrible. MySpace had won and who wants college students as customers? Or a way to stay on strangers’ couches, that just sounds terrible all around. These all sounded really bad but they turned out to be good. If they had sounded really good, there would have been too many people working on them.

As Peter Thiel’s going to discuss in the 5th class, you want an idea that turns into a monopoly, but you can’t get a monopoly in a big market right away, [there’s] too much competition for that. You have to find a small market in which you can get a monopoly and then quickly expand. This is why some great startup ideas look really bad at the beginning. It’s good if you can say something like, “Today only this small subset of users are going to use my product, but I’m going to get all of them and in the future almost everyone will use my product.”

Here’s the theme that’s going to come up a lot: you need conviction in your own beliefs, and a willingness to ignore others’ naysaying. The hard part is that this is a very fine line. There’s “right” on one side of it, and “crazy” on the other. But keep in mind that if you do come up with a great idea, most people are going to think it’s bad. You should be happy about that. It means they won’t compete with you. This is also a reason why it’s not usually dangerous to tell people about your idea. The truly good ideas don’t sound like they’re worth stealing.

You want an idea about which you can say, “I know it sounds like a bad idea, but here’s specifically why it’s actually a great one.” You want to sound crazy, but you want to actually be right. And you want an idea that not many other people are working on, and it’s OK if it doesn’t sound big at first. A common mistake among founders, especially first-time founders, is they think that the first version of their product, the first version of their idea, needs to sound really big. But it doesn’t, it needs to take over a small specific market and expand from there. That’s how most great companies started.

Unpopular but right, is what you’re going for. You want something that sounds like a bad idea but is a good idea.

You also really want to take the time to think about how the market’s going to evolve. You need a market that’s going to be big in ten years. Most investors are obsessed with the market size today, and they don’t think at all about how the market’s going to evolve. In fact I think this is one of the biggest systemic mistakes that investors make. They think about the growth of the startup itself, they don’t think about the growth of the market.

I care much more about the growth rate of the market than it’s current size, and I also care if there’s any reason that it’s going to top out. You should think about this. I’d prefer to invest in a company that’s going after a small but rapidly growing market, than a big but slow growing one.

One of the big advantages of these sorts of markets, these small but rapidly growing markets, is that customers are usually pretty desperate for a solution, and they’ll put up with an imperfect but rapidly improving product.

And a big advantage of being a student, one of the two biggest advantages, is that you probably have better intuition about which markets are likely to start growing rapidly than older people do.

Another thing that students usually don’t understand or at least it takes a while, you cannot create a market that doesn’t want to exist. You can basically change everything in a startup but the market. So you should actually do some thinking, to be sure or at least as sure as you can be, that the market you’re going after is going to grow and be there.

There are a lot of different ways to talk about the right kind of market. For example surfing someone else’s wave, or stepping into an up elevator, or being part of a movement. But all of this is just a way of saying, you want a market that’s going to grow really quickly. It may seem small today, it may be small today, but you know and other people don’t, that it’s going to grow really fast.

So think about where this is happening in the world. You need this sort of a tailwind to make a startup successful. The exciting thing is there are probably more of these tailwinds now than ever before. As Marc Andreessen says, “software’s eating the world.” It’s just everywhere. There are so many great ideas out there, and you have to pick one and find one that you really care about.

Another version of this that gets down to the same idea is Sequoia’s famous question, “Why now?” Why is this the perfect time for this particular idea and to start this particular company? Why couldn’t it have been done 2 years ago, and why will 2 years in the future be too late? For the most successful startups we’ve been involved with, they all had a great idea, a great answer to this question, and if you don’t you should be at least somewhat suspicious about it.

In general, it’s best if you’re building something that you yourself need. You’ll understand it much better than if you have to understand it by talking to a customer to build the very first version. If you don’t need it yourself and you’re building something that someone else needs, realize that you’re at a big disadvantage and get very, very close to your customers. Try to work in their office if you can and if not, talk to them multiple times a day.

Another somewhat counter-intuitive thing about good startup ideas is that they’re almost always very easy to explain and very easy to understand. If it takes more than a sentence to explain what you’re doing, it’s almost always a sign that it’s too complicated. It should be a clearly articulated vision with a small number of words.

And the best ideas are usually either very different from existing companies in one important way, like Google being a search engine that worked just really well and none of the other stuff of the portals, or totally new like SpaceX. Any company that’s a clone of something else that already exists, with some small or made up differentiator, like we’re going to be X beautiful design, or we’re going to be Y for people that like red wine instead; that usually fails.

So as I mentioned, one of the great things about being a student is that you have a very good perspective on new technology. And learning to get good at having new ideas takes a while. So start working on that right now. That’s one thing we hear from people all the time, that they wish they had done more when they were a student. The other is meeting potential co-founders. You have no idea how good of an environment you’re in right now for meeting people that you can start a company with down the road. And the one thing that we always tell college students is more important than starting any particular startup, is getting to know a lot of potential co-founders.

So I want to finish this section of my talk with a quote from 50 Cent. This is from when he was asked about Vitamin Water.

“Most people think first of what they want to express or make, then find the audience for their idea. You must work the opposite angle, thinking first of the public. You need to keep your focus on their changing needs, the trends that are washing through them. Beginning with their demand, you create the appropriate supply.” - 50 Cent

I won’t read it, it’s up [on the screen], but it’s about the importance of thinking about what customers want, and thinking about the demands of the market. Most people don’t do this. Most students especially don’t do this. If you can just do this one thing, if you can just learn to think about the market first, you will have a big leg up on most people starting startups. And this is probably the thing that we see wrong with Y Combinator apps most frequently, is that people have not thought about the market first, and what people want first.

A great product

So for the next section I’m going to talk about building a great product. And here again, I’m going to use a very broad definition of product. It includes customer support and copyright explaining the product. Anything involved in your customer’s interaction with what you built for them.

To build a really great company you first have to turn a great idea into a great product. This is really hard but it’s crucially important and fortunately it’s pretty fun. Although great products are always new to the world, and it’s hard to give you advice about what to build, there are enough commonalities that we can give you a lot of advice about how to build it.

One of the most important tasks for a founder is to make sure that the company builds a great product. Until you’ve built a great product, almost nothing else matters. When really successful startup founders tell the story of their early days, it’s almost always sitting in front of the computer working on their product or talking to their customers. That’s pretty much all the time, they do very little else. And you should be very skeptical if your time allocation is much different.

Most other problems that founders are trying to solve–raising money, getting more press, hiring, business development, etc.–these are significantly easier when you have a great product. It’s really important to take care of that first. Step 1 is to build something that users love.

At YC we tell founders to work on their product, talk to users, exercise, eat, and sleep. And very little else. All the other stuff I just mentioned–PR, conferences, recruiting advisors, doing partnerships–you should ignore all of that and just build the product and then get it as good as possible by talking to your users.

Your job is to build something that users love. Very few companies that go on to be super successful get there without first doing this. A lot of good-on-paper startups fail because they merely make something that people like. Making something that people want, but only a medium amount, is a great way to fail and not understand why you’re failing. So these are the two jobs.

Something that we say at YC a lot is that it’s better to build something that a small number of users love, than a large number of users like. Of course it’d be best to build something that a [large] number of users love, but opportunities to do that for version 1 are rare and they’re usually not available to startups.

So in practice you end up choosing either the gray or the orange [see slides]. You make something that a lot of users like a little bit, or something that a small number of users love a lot.

And this is a very important piece of advice. Build something that a small number of users love. It’s much easier to expand from something that a small number of people love to something that a lot of people love, than from something that a lot of people like to a lot of people love.

If you get this right, you can get a lot of other things wrong. If you don’t get this right, you can get everything else right and you’ll probably still fail. So when you start on a startup, this is the only thing you need to care about until it’s working.

[student asks Sam to explain again]

Sure, so you have a choice in a startup. The best thing of all worlds would be to build a product that a lot of people really love. In practice you can’t usually do that, because if there’s an opportunity like that, Google or Facebook will do it. So there’s a limit to the area under the curve of what you can build. And you can either build something that a lot of users like a little bit, or you can build something that a small number of users love a lot.

And the total amount of love is the same, it’s just a question of how it’s distributed. And there’s like this law of conservation of how much happinesss you can put into the world with the first product of a startup. And so startups always struggle with which of those two they should go, and they seem equal right? Because the area under the curve is the same.

But we’ve seen this time and again that they’re not, and it’s so much easier to expand. Once you’ve got something that some people love, you can expand that to something a lot of other people love. But if you only get ambivalence or sort of like weak enthusiasm, and then try to expand that, you’ll never get up to a lot of people loving it.

So the advice is, find a small group of users and make them really love what you’re doing. Does that make sense?

One way that you know when this is working is that you’ll get growth by word of mouth. If you make something that people love, people will tell their friends about it. This works for consumer products and enterprise products as well. When people really love something, they tell their friends about it and you’ll see organic growth.

If you find yourself talking about how it’s OK that you’re not growing, because there’s a big partnership that’s going to come save you or something like that, it’s almost always a sign of real trouble. Sales and marketing are really important and we’re going to have two classes on them later, but if you don’t have some early organic growth then you’re product probably isn’t good enough yet.

A great product is the secret to longterm growth hacking. You should get that right before you worry about anything else. It doesn’t get easier to put off making a great product.

If you try to build a growth machine before you have a product that some people really love, you’re almost certainly going to waste your time. Break-out companies almost always have a product that’s so good that it grows by word of mouth. Over the long run, great products win.

Don’t worry about your competitors raising a lot of money and what they may do in the future. They probably aren’t very good anyway. Very few startups die from competition. Most die because they themselves fail to make something users love. They spend their time on other things. So worry about this above all else.

Another piece of advice to make something that users love: start with something simple. It’s much, much easier to make a great product if you have something simple. Even if your eventual plans are super complex (and hopefully they are), you can almost always start with a smaller subset of the problem than whatever you think is the smallest. And it’s hard to build a great product, so you want to start with as little surface area as possible.

Think about the really successful companies and what they started with. Think about products that you really love, they’re generally incredibly simple to use and especially to get started using. The first verison of Facebook was almost comically simple. The first version of Google was just an ugly webpage with a textbox and two buttons, but it returned the best results and that’s why users loved it. The iPhone is far simpler to use than any smartphone that ever came before it, and it was the first one that people really loved.

Another reason that simple is good is because it forces you to do one thing extremely well, and you have to do that to make something that people love.

The word fanatical comes up again and again when you listen to successful founders talk about how they think about their product. Founders talk about being fanatical in the way they care about the quality of the small details. Fanatical in getting the copy that they used to explain their product just right. And fanatical in the way they think about customer support.

In fact one thing that correlates with success among the YC companies is the founders that hook up PagerDuty to their ticketing system so that even if the user emails in the middle of the night when the founders are asleep, [the user] still gets a response within an hour. Companies actually do this in the early days. These founders feel physical pain when the product sucks, and they want to wake up and fix it. They don’t ship crap, and if they do they fix it very, very quickly. And it definitely takes some level of fanaticism to build a great product.

You need some users to help with the feedback cycle, but the way to get those users is manually. You should go recruit them by hand. Don’t do things like buy Google ads in the early days to get initial users. You don’t need very many, you just need ones that will give you feedback everyday and eventually love your product. So instead of trying to get them on Google AdWords, just find a few people in the world that will be good users, recruit them by hand.

Ben Silbermann, when everyone thought Pinterest was a joke, recruited the initial Pinterest users by chatting up strangers in coffee shops. He really did. He just walked around Palo Alto and said, “Will you please use my product?” He also used to run around the Apple Store in Palo Alto and he would set all the browsers to the Pinterst homepage real quick before they caught him and kicked him out. So then when people walked in there, they were like “Oh what is this?” This is an important example of doing things that don’t scale. If you haven’t read Paul Graham’s essay on that topic, you definitely should.

So get users manually and remember that the goal is to get a small group of them to love you. Understand that group extremely well, get extremely close to them. Listen to them and you’ll almost always find out that they’re very willing to give you feedback. Even if you’re building the product for yourself, listen to outside users and they’ll tell you how to make a product they’ll pay for. Do whatever you need to make them love you, make [unintelligable] you’re doing, because they’re also going to be the advocates that help you get your next users.

You want to build an engine in the company that transforms feedback from users into product decisions. Then get it back in front of the users and repeat. Ask them what they like and what they don’t like and watch them use it. Ask them what they’d pay for. Ask them if they’d be really bummed if your company went away. Ask them what would make them recommend the product to their friend. And ask them if they’ve recommended it to any yet.

You should make this feedback loop as tight as possible. If your product gets 10% better every week, that compounds really, really quickly. One of the great advantages of software startups is just how short you can make the feedback loop. It can be measured in hours, and the best companies usually have the tightest feedback loops. You should try to keep this going for all of your company’s life, but it’s really important in the early days.

The good news is that all of this is doable. It’s hard, it takes a lot of effort, but theres’s no magic. The plan is at least straightforward and you will eventually get to a great product.

Great founders don’t put anyone between themselves and their users. The founders of these companies do things like sales and customer support themselves in the early days. It’s critical to get this loop embedded in the culture. In fact a specific problem that we always see with Stanford startups for some reason is that the students try and hire sales and customer support people right away. You gotta do this yourself. It’s the only way.

You really need to use metrics to keep yourself honest on this. It really is true that the company will build whatever the CEO decides to measure. If you’re building an Internet service, ignore things like total registrations. Don’t talk about them, don’t let anyone in the company talk about them, and look at growth in active users, activity levels, cohort retention, revenue, net promoter scores, these things that matter. And then be brutally honest if they aren’t going in the right direction. Startups live on growth. It’s the indicator of a great product.

So this about wraps up the overview on building a great product. I want to emphasize again, that if you don’t get this right, nothing else we talk about in the class will matter. You can basically ignore everything else that we talk about until this is working well. On the positive side, this is one of the most fun parts of building a startup.

So I’m going to pause here, I’ll pick back up with the rest of this on Thursday. And now we’re going to have Dustin [Moskovitz] talk about why you should start a startup. Thank you for coming Dustin.

comments powered by Disqus


Ben Sullivan