Lecture 1 - How to Start a Startup (Sam Altman, Dustin Moskovitz)

YC Root Access43:522014-09-23English

The core advice of this introductory startup lecture is to first find a problem you feel compelled to solve, make sure starting a company is the best way to solve it, and then start with a simple product that a small number of users can truly love.

Overview

Lecture 1

Key points

  • The course is for startup teams aiming to grow fast and eventually build large companies. You can work to get the idea, product, team, and execution right, but luck still affects the outcome.
  • A startup idea is about more than product features. It also includes the size and growth rate of the market, the path to growth, and whether the business can build an advantage that's hard to copy. Plans will change, but it's still worth thinking through where the long-term value will come from.
  • Win a specific small market first, then expand. When evaluating a market, look not only at how big it is today, but also at whether it can grow quickly and why now is the right time to enter.
  • Early on, it's better to have a few users who love the product than many who think it's just okay. The former is easier to build on, and organic growth through word of mouth is a sign that the product is starting to work.
  • Founders should recruit the first users themselves, quickly turn feedback into product changes, and put the revised product back in users' hands to see if those changes work. Track progress with metrics like active user growth, cohort retention, and revenue rather than focusing on total signups.
  • Being the boss doesn't mean you control your schedule. Serious problems can interrupt your plans, and the team will take its cue from how much effort the founders put in.
  • Starting a company doesn't automatically give you more impact. An established company's user base, infrastructure, and team can also amplify your contribution. A more compelling reason to start a company is that the problem is genuinely worth solving and you're particularly well suited to build a company to solve it.

Chapter notes

Who the Course Is For and the Four Ingredients of a Startup

  • Sam Altman explains that Y Combinator has been teaching people how to build startups for nine years, mostly by tailoring its advice to each company's situation. CS183B will publicly teach the roughly 30% of that material that applies broadly. Y Combinator has funded 720 companies. Guest speakers teach 17 of the 20 classes, and every guest has helped build a company worth more than $1 billion.
  • The course is for founders pursuing rapid growth who ultimately want to build a large company. Startups are very different from ordinary companies, and much of the advice won't work if applied as-is to large companies or other businesses that aren't startups.
  • The four factors founders can control are a good idea, a good product, a good team, and excellent execution. They overlap, and later classes will cover each one in depth. Altman describes the outcome as the product of these four factors and luck, where luck is a random number between 0 and 10,000. Getting all four right improves your odds but doesn't guarantee success.
  • The starting line is more level than you might think. Young, inexperienced people can start companies, and so can older, experienced people. Having no money or name recognition may be a disadvantage in other work settings, but in a startup it can be a big advantage.
  • Don't start a company just to start one. There are easier ways to make money, and running a startup is often harder and more painful than founders expect. You should start a company only if you feel strongly about solving a specific problem and believe a company is the best way to solve it. In Y Combinator's most successful cases, the problem came first and the company came second.
  • Dustin Moskovitz, co-founder of Facebook and Asana, will discuss why to start a company in the second half of the lecture. Altman then introduces the first ingredient, a good idea. In recent years, it's become fashionable to say startup ideas don't matter, to the point that spending a lot of time thinking about an idea can seem out of step.

Why Good Ideas Are Often Underrated

  • Excellent execution is at least 10 times as important as a good idea and 100 times as hard, but even the best execution can't save a bad idea. Putting a product in users' hands can answer questions you can't settle in advance, but that doesn't mean more pivots are better. Most great companies start with good ideas. Successful pivots also often address a real need the founders have themselves. For example, Brian Chesky couldn't afford his rent but had extra space, which led to Airbnb.
  • When evaluating an idea, look beyond the product to the market's size and growth rate, the company's growth strategy, and what will make the business hard to copy later. If things go well, you may be working on the startup for 10 years, so it's worth thinking ahead about long-term value and barriers to competition. The plan itself may not be useful, but the process of planning is valuable. You don't have to work out every step at the start; a core idea that can keep growing is enough.
  • The idea should come first, and the company second. Wait until you find an idea you feel compelled to figure out before starting a company. If you have several good ideas, choose the one that comes to mind most often when you're not deliberately thinking about work. An important mission can give a team the intense focus and productivity a startup requires, and it's usually hard to create that sense of mission without a good founding idea.
  • Building a great company usually takes 10 years. It's not something you do for two or three years before moving on to what you really love. If you don't love what you're doing or believe it matters, you're likely to quit along the way. Hard but important projects attract outside help more easily than projects that copy existing ideas, so in that sense, doing a hard startup is actually easier.
  • Good ideas often look terrible at first. The 13th search engine lacked the features of a web portal, the 10th social network was only for broke college students, and staying on a stranger's couch didn't sound like a good business. Large markets are fiercely competitive, so it's hard to establish a monopoly from the start. A more practical path is to win a small market first, then expand quickly.
  • If most people don't believe in your idea, they're less likely to compete with you. That's also why sharing your idea usually isn't that risky: truly good ideas often don't sound worth copying. There's a fine line between sticking to your convictions and being blindly confident. The key is to explain exactly why the idea looks bad but is actually good, not to make it sound big from the start.

Evaluating the Market and Timing

  • The first version doesn't need to sound grand. Win a specific small market first and expand from there. Look for ideas that sound unlikely to work but are actually right. When evaluating a market, its room to grow over the next 10 years, how quickly it can grow, and how big it might ultimately get matter more than its current size. In a small, fast-growing market, customers who urgently need a solution are usually willing to accept an imperfect product that's improving quickly.
  • Students often have better instincts than older people about which markets are about to grow rapidly. You can change almost anything else about a startup, but you can't conjure up a market where there's no demand. Make as sure as you can that your target market actually exists and will grow quickly.
  • Use Sequoia's question to test your timing: Why is now the best time to pursue this idea and start this company? Why couldn't it have been done two years ago, and why would two years from now be too late? The most successful startups Sam has been involved with all had good answers. If you don't have an answer, you should take another look at the idea.
  • It's best to build something you need yourself, because you already understand the need when you build the first version. If you're solving someone else's problem, try to work alongside customers in their office. If you can't, talk to them several times a day. Good ideas can usually be explained in one sentence. If you can't explain yours clearly, it's often a sign that it's too complicated.
  • A good idea is either clearly different in one crucial way, as when Google focused on doing search well instead of offering the other features of a web portal, or does something entirely new, like SpaceX. Copying an existing company and adding only a small difference, such as a nicer design or a different user group, usually fails.
  • Coming up with good ideas takes practice. Students can take advantage of their ability to spot new technologies and start practicing now. School is also a good place to meet future startup partners. Getting to know more potential co-founders matters more than immediately starting work on any one startup idea.

Build Something a Few Users Love

  • One of the most common problems in Y Combinator applications is that founders haven't first considered the market and what users want. A product is more than the features you build. It also includes customer support, the copy that explains the product, and the user's entire experience of interacting with it.
  • Until they have a good product, founders should spend almost all their time building it and talking to users, then use that feedback to make it as good as possible. PR, conferences, finding advisers, and discussing partnerships can wait. Once the product is good, raising money, getting press coverage, hiring, and expanding the business all become much easier.
  • A first version rarely makes a large number of users fall in love with it. It's better to have a few users who love it than many who think it's pretty good. Going from a few people who love it to more people who love it is much easier than turning mild approval from many people into love.
  • For both consumer and enterprise products, organic growth from users recommending the product on their own is a sign that they genuinely like it. If there's no organic growth early on, the product probably isn't good enough yet. Counting on a big partnership to save it, or setting up a growth system before users love it, usually won't solve that problem.
  • Even if the long-term plan is complicated, the first version should solve only a small part of the problem. It can be even smaller than what you originally thought was the minimum version. That makes it easy for users to get started and forces the team to do one thing extremely well. Early Facebook was extremely simple. Early Google's page was bare-bones, with just one input box and two buttons, but it won users by giving them the best search results.

Recruit Users Yourself and Iterate Quickly

  • Successful founders are almost obsessively demanding about product details, the copy that introduces the product, and customer support. They don't rush out a rough product, and they fix problems quickly. At some YC companies, founders connect PagerDuty to their support ticket system so users get a reply within an hour even if they email in the middle of the night. This practice is correlated with success.
  • Early on, don't buy traffic with Google AdWords. Recruit a few users yourself who are willing to give feedback every day and will eventually love the product. Ben Silbermann approached strangers in Palo Alto coffee shops and set the browsers at the local Apple Store to open on Pinterest so people coming into the store would discover it.
  • Even if you originally built the product for yourself, get to know this small group of users well and listen to what they say they'd pay for. People who genuinely love the product will also help bring in the next group of users. Watch how they use it, ask what they like and dislike, whether they'd be disappointed if it disappeared, what would make them recommend it to a friend, and whether they've already done so.
  • Turn user feedback into product decisions, put the revised product back in users' hands, and repeat, keeping the loop as short as possible. Software startups can measure feedback cycles in hours. If the product improves by 10% every week, the compounding effect will show up quickly, so build this loop into the company culture early and keep it going over the long term.
  • Founders should handle early sales and customer support themselves instead of rushing to hire people to talk to users for them. What the CEO measures determines what the company builds around. An internet service should ignore total signups and focus on active user growth, engagement, cohort retention, revenue, and Net Promoter Score. If the metrics aren't moving in the right direction, be honest about it, because growth is a sign of a good product.

The Reality Behind the Startup Glamour

  • Dustin Moskovitz lists four common reasons for starting a company: thinking it's glamorous, wanting to be the boss, wanting a more flexible schedule, and hoping to have more impact and earn more than by joining a later-stage company. It's important to understand your motivation, because some reasons hold up only in certain situations, while others can be distorted by the romanticized stories Hollywood and the media tell.
  • The Social Network portrays startup life as a mix of parties and flashes of inspiration. In reality, day-to-day work means keeping your head down answering customer emails, making sales, and solving tough engineering problems. Two photos from Facebook's early days were taken on different days. Mark had changed his clothes, but he was still working at the same desk in the same position.
  • Founders don't just worry about failing themselves. They're also responsible for the people who follow them. Some people's livelihoods depend on the company, and even when that's not the case, they've invested valuable time. If something urgent comes up, you have to deal with it even on weekends or on vacation. Some startups even require founders to respond at 3 a.m. Fundraising is another common example of this kind of pressure.
  • Media attention has an unwelcome side too. Being on the cover of Time as Person of the Year might be exciting, but having your wedding photos appear in People or having a talk you gave torn apart by Valleywag might not be.
  • Unhappy employees can leave, but if founders quit early, they may not only damage their professional reputation but also cost themselves and their employees money. If things go well, the commitment can last 10 years. Even if they don't, it can still take five years: for example, three years to realize the company isn't going to work, then two years with the acquirer. Failing quickly is actually lucky.
  • Stress made Dustin very unhealthy in Facebook's early days. He didn't exercise, was often anxious, and at 21 or 22 would throw out his back about once every six months. That's why he sees managing stress as a core responsibility for founders and agrees with Ben Horowitz that a CEO's first job is to manage their own mental state.

Freedom, Financial Returns, and Impact

  • Being the boss doesn't mean you get to make the rules however you like. Management decisions that once looked stupid were often difficult trade-offs between conflicting employee views and customer needs. A CEO's time is often taken up by problems other people bring them. You might plan to spend Monday improving the company, but if a key employee is about to quit, keeping them becomes the top priority.
  • Founders can be called on to deal with problems at any time and may not get to choose when they work. The team takes its cue from how much effort the founders put in. If the founders ease up, the team often does too. A four-day workweek or a similar arrangement might suit a small business or a niche market, but Dustin thinks founders need to commit full-time once the team grows beyond two or three people.
  • A smaller equity stake doesn't necessarily mean making less money. Based on valuations at the time, an engineer with about five years of industry experience joining as the 100th employee might get about 0.1% equity, worth roughly $10 million at Dropbox or $200 million at Facebook. Even joining Facebook as its 1,000th employee around 2009 would have yielded about $20 million. These figures can serve as benchmarks for evaluating startup returns.
  • Take an 'Uber for pet sitting' as an example. Even if you build it into a $100 million company, you might be left with only about 10% equity after dilution from several funding rounds and expansions of the option pool. And a company that hasn't been started yet is far less of a sure thing than Facebook in 2009 or Dropbox in 2014. Dustin thinks that if you're very confident you can build a $100 million business, you can consider starting it. If you're well suited to build a $2 billion 'Uber for space travel,' it's worth just doing it.
  • Joining a company that's already operating at scale can also amplify your impact. Facebook or Google already has a billion users, along with infrastructure, proprietary technology, and teams that can help turn ideas into products. Bret Taylor was roughly Google's 1,500th employee, but he didn't need to start a company to build Google Maps, which is used by hundreds of millions of people worldwide.
  • When Justin Rosenstein was a product manager at Google, he built a prototype for chat inside Gmail as a side project. He demonstrated chat in a browser, which some people at the time thought couldn't be done, and the team then turned it into a product. He later joined Facebook as roughly its 250th employee and built the Like button with Andrew Bosworth, Leah Pearlman, and others through a hackathon project.

What Problems Are Worth Starting a Company For?

  • Even a popular feature that changes how people use the internet doesn't necessarily require starting a company. The feature in the previous example relied on Facebook's distribution and would almost certainly have failed as an independent startup. Before deciding whether to start a company, figure out where the idea is most likely to succeed.
  • The strongest reason to start a company is that you feel you have to do it. Even if you didn't start a company, you'd keep working on it. Passion can get you through the hard parts of running a startup and help you recruit. Candidates can tell whether you genuinely care about the problem, and they can choose to join a more passionate founding team instead.
  • Feeling you have to do it also means the world needs the solution and you really are suited to solve the problem. If the problem isn't important enough, you can spend your time on a more valuable idea or at an existing company. If you're not the right person, even winning may just mean crowding out a better-suited team and making the outcome worse.
  • Dustin and Justin weren't in a hurry to start a company. They already had worthwhile projects at Facebook. They did their regular jobs during the day, but at night they kept developing an internal task management tool. After seeing its real impact at Facebook, they had to seriously discuss what would be lost if they didn't start Asana.
  • They believed the tool could create a lot of value outside Facebook, but didn't think anyone else would build the same solution. The problem had been around for a long time, and all they'd seen were incremental improvements. Dustin believed that if they didn't release what they saw as the best solution, much of that value would never be realized. Knowing an idea matters and finding that you simply can't let it go is exactly the kind of sign that it's worth starting a company for.

What it answers

When evaluating a startup idea, what should you look at besides the product itself?
Look at the size and growth of the market, how the company will expand, and whether the business can become hard to copy. You also need to explain why the idea can work now, why it couldn't have worked two years ago, and why two years from now might be too late.
Should the first version of a product attract more people or win over a small group first?
Win over a small group first. Going from a few people who genuinely love the product to more people who love it is usually easier than turning mild approval from a large number of users into love. Organic growth from people recommending it on their own is an early sign.
How can founders use early feedback to improve the product?
Personally recruit a small number of users who are willing to give regular feedback. Watch how they use the product, ask what they'd pay for and whether they've recommended it, and quickly put the changes back in their hands to see if they work.
Do you have to start your own company to build a product that has an impact?
Not necessarily. An established company's user base, infrastructure, and team can amplify your contribution. Dustin gives the example of Bret Taylor building Google Maps at Google.
What does Dustin think is the best reason to start a company?
You care so much about the problem that you can't let it go, the problem is genuinely worth solving, and you're especially well suited to solve it. Passion can get you through difficulties and help you attract talent, but it needs to be backed by a real need.

Quotes

The idea should come first, and the startup should come second.
Unpopular but right is what you're going for.
It's better to build something that a small number of users love than a large number of users like.
Great founders don't put anyone between themselves and their users.

Mentioned

  • CS183B — This startup course aims to publicly teach the parts of Y Combinator's experience that apply more broadly.
  • Sam Altman — Then president of Y Combinator, he leads the first half of the lecture and introduces startup ideas and early-stage products.
  • Dustin Moskovitz — The co-founder of Facebook and Asana leads the second half of the lecture and discusses why to start a company.
  • Y Combinator — It provides startup cases and experience for the course. Sam says it had funded 720 companies at the time.
  • Airbnb — Sam uses the story of Brian Chesky, who couldn't afford his rent but had extra space, to show that good ideas can come from founders' own needs.
  • Sequoia — Sam cites its classic question, 'Why now?', as a way to test the timing of a startup.
  • Ben Silbermann — His approach of introducing Pinterest to strangers in coffee shops is an example of recruiting the first users by hand.
  • Facebook — It's both Dustin's firsthand example of early startup stress and an example of how existing distribution channels can amplify a product's impact.
  • Bret Taylor — Dustin uses his experience building Google Maps after joining Google to show that you can make a broad impact without starting a company.
  • Justin Rosenstein — Dustin's co-founder at Asana built a prototype for chat inside Gmail at Google and led a team that developed the Like button at Facebook.
  • Asana — It grew out of an internal task management tool Dustin and Justin built at Facebook. After seeing its value, they decided to start a company.