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

Source: Lecture 1 - How to Start a Startup (Sam Altman, Dustin Moskovitz), YC Root Access, 43:53, uploaded 2014-09-23, Watch Later position 427.

Sam Altman opens the first CS183B lecture with a microphone that may or may not be working. He has taught startup founders at Y Combinator for nine years, and estimates that around 30% of that teaching applies beyond the individual companies. The course tries to put that part on the record. Its advice targets technology startups built for hyper-growth and a very large eventual company, so Altman warns that much of it will fail in a large company, a normal business, or another setting with different aims.

The four areas and the reason to begin

Altman reduces startup success to four areas that founders can influence: a great idea, a great product, a great team, and great execution. He writes the outcome as idea × product × execution × team × luck, with luck represented as a random number from zero to 10,000. The formula is a reminder that competence cannot remove chance. It can improve the part of the result that the founder can touch.

He also treats startups as an unusually open field. Young and inexperienced founders can succeed, as can people with years of experience. Being poor and unknown can help because a startup gives those conditions a different value than an established workplace does.

Before explaining how to start, Altman asks why someone should do it at all. Starting a company is hard and painful, and people who have done it tend to underestimate that cost in advance. The reason should begin with a particular problem that feels compelling, followed by the judgement that a new company offers the best way to solve it. The startup comes after the passion for the problem.

An idea worth a decade

Altman begins with a popular piece of startup advice: ideas do not matter, so founders should launch quickly, throw experiments against the market, and celebrate pivots. He accepts part of this. Reality changes once users touch a product, and execution matters far more than the first idea. He still argues that the pendulum has swung too far. A bad idea stays bad, and great execution aimed at a terrible market goes nowhere.

The successful pivots he trusts tend to move towards something the founders themselves wanted. Airbnb began with Brian Chesky’s inability to pay his rent and the spare space he could offer. Altman uses the example to distinguish a change of direction that reveals a real need from a random search for anything that might work.

The idea also includes the market’s size and growth, the company’s growth strategy, and the way the business might defend itself from imitation. A founder may work on the company for ten years, so early thought about long-term value and defensibility earns its cost. Plans become obsolete, yet the act of planning teaches the founder to see the shape of the business. Altman wants a small kernel that can grow into something more ambitious without requiring a complete map to world domination.

That kernel needs a business that other people will find difficult to copy. Altman repeats the ordering several times: the idea comes first, and the startup follows when the founder feels compelled to explore it. When several ideas seem plausible, he suggests choosing the one that occupies the mind during time away from work. Founders often regret starting before they found an idea they loved.

Mission gives the idea force inside and outside the company. A large group can sustain the focus a startup needs when the work feels like an important mission. A founder who expects to build for two or three years before moving on to a more personal passion is usually misjudging the time involved. Good startups often take a decade. A mission also attracts help from people who want to support a hard, important project, whilst a derivative copy gives the team little reason to endure the strain. Altman points to Paul Graham’s upcoming lecture on finding ideas and treats the skill as something that improves with practice.

Ideas that look poor at first

The best startup ideas can look absurd at the start. Altman names the thirteenth search engine without the features of a web portal, the tenth social network limited to college students without money, and a service for staying on strangers’ couches. Search seemed finished, MySpace appeared to have won, and couch-surfing sounded unsafe. Google, Facebook, and Airbnb made those initial descriptions look different in hindsight.

The advantage lies in taking a small market where the company can become dominant and then expanding. The first version does not need to sound large. It needs a specific group of users that the company can serve completely, together with a credible path towards a much larger future market. The useful idea may sound bad to people who cannot see the path.

That calls for conviction and a willingness to withstand other people’s doubts. Altman describes a narrow line between being right and being crazy. A good idea may be safe to discuss because it sounds unappealing to competitors. The founder should be able to explain why the apparently poor idea works, with a reason that holds up under inspection. Unpopular and right is the target.

Market growth matters more than present market size. Altman prefers a small market growing quickly to a large market moving slowly because desperate customers will tolerate an imperfect product that improves fast. Students may have better intuition about new technology and emerging markets than older investors. They still cannot create a market that does not want to exist. A founder can change nearly everything about a startup except the underlying demand.

He collects several phrases for this condition: surfing somebody else’s wave, stepping into an up elevator, and joining a movement. They all describe a market with a strong tailwind. Marc Andreessen’s claim that software is eating the world supplies the broad example, whilst Sequoia’s question makes the test concrete: why now? A good answer explains why the company could not have been built two years earlier and why waiting two years would be too late.

The first product is easier to build when the founder needs it personally. If the founder is building for someone else, the distance creates a serious disadvantage. Altman advises working close to those customers, even in their office if possible, and speaking with them several times a day. He also wants the idea to fit into a sentence. Google separated itself from portals by making search work very well. SpaceX represented a new category. A clone with a small cosmetic difference usually has no such force.

The lecture returns to students as a source of two advantages: a close view of new technology and an unusually good environment for meeting future co-founders. Altman says that learning to notice ideas matters less than getting to know the people with whom a company might later be built. A quotation from 50 Cent about Vitaminwater appears on the lecture slide to make the same point about customer demand. Altman treats thinking about the market first as one of the clearest gaps in Y Combinator applications.

A product that users love

Altman uses a broad definition of product. It includes the service itself, customer support, and the copy that explains what the company has built. The founder’s first job is to turn a good idea into something users love. Until that happens, raising money, seeking press, hiring, and partnerships receive attention that the company has not earned.

Early founders of successful startups tend to spend their days building the product and talking to users. Y Combinator’s rough prescription is work on the product, talk to users, eat, exercise, sleep, and remove most other activity. Something that a large group likes a little gives a company weak evidence. A small group that loves the product supplies a stronger base from which to expand.

Altman explains this with a diagram where the total amount of early user enthusiasm stays roughly fixed. A startup can distribute that enthusiasm across many people who feel mild interest or concentrate it in a small group who care deeply. The two shapes look equal on the slide, yet the concentrated form expands more easily. Weak enthusiasm rarely becomes strong enthusiasm through scale alone.

Word of mouth is the practical test. People who love a consumer or enterprise product tell their friends, which creates early organic growth. A partnership or marketing plan that promises to rescue a product with no organic pull is usually evidence that the product needs more work. Altman says that most startups die from failing to make something users love, with competition receiving more blame than it deserves.

Simplicity raises the chance of getting there. The first Facebook was almost comically simple. The first Google was an ugly page with a text box and two buttons, yet it returned the best results. The iPhone made a smartphone easier to use than its predecessors. Each example gives a founder a smaller surface on which to do one thing very well.

The required care reaches into small details. Successful founders describe themselves as fanatical about product quality, explanatory copy, and customer support. Some early Y Combinator companies connected PagerDuty to their support tickets so that a user who wrote in during the night still received a response within an hour. The founders feel the failure in the product and fix it quickly.

Initial users should arrive by hand. Altman advises against buying Google ads before the product has found its people. Ben Silbermann recruited early Pinterest users by approaching strangers in Palo Alto coffee shops and asking them to use the product. He also set browsers in an Apple Store to Pinterest before employees removed him. The story illustrates Paul Graham’s essay about doing things that do not scale: a founder can learn more from a few carefully chosen users than from a large anonymous audience.

The founder then builds a tight loop between user feedback and product decisions. Ask what users like, what they would pay for, whether they would be upset if the company disappeared, and whether they have recommended it to a friend. Watch them use the product rather than relying on their description. A product that improves by 10% each week compounds its advantage. The founders should keep this loop in the company for as long as it remains possible.

Altman is especially wary of delegating the first contact with users. Founders should handle sales and support themselves because that work embeds the feedback loop in the culture. He sees Stanford startups hire sales and customer-support staff too soon, before the founders understand the customer’s problem.

Metrics keep the founder honest. An internet service should ignore total registrations and track active users, activity levels, cohort retention, revenue, and net promoter scores. The company builds around what its chief executive measures, so the measures must describe behaviour that matters. Growth is the sign that the product is beginning to work. Altman ends his part of the lecture by saying that everything else in the course can wait until this is true.

The stories people tell about entrepreneurship

Moskovitz takes over with four common reasons for starting a company: glamour, being the boss, control over one’s schedule, and the hope of greater impact or money. He wants each listener to identify the reason they actually hold because several of these motives lead people into a poor decision.

The press and Hollywood tend to show entrepreneurship as a chain of brilliant insights, parties, and sudden success. Moskovitz’s own time at Facebook involved long hours at a desk, customer support, sales, and difficult engineering work. He shows images from The Social Network that suggest Mark Zuckerberg moved between parties and flashes of insight. Moskovitz says the real work kept him at the same table, focused on the company. The lecture’s image of startup life is quieter and more repetitive.

Stress comes from responsibility for the people who follow the founder. Employees may depend on the company for their livelihood, and even when they do not, they have committed some of the best years of their lives to the project. The founder carries the opportunity cost of that decision and stays available when an important problem appears. Fundraising adds a special form of pressure, as does unwanted media attention. Moskovitz also describes his own early Facebook years, when anxiety, poor health, and repeated back injuries showed him that managing the founder’s psychology belongs to the job. He borrows Ben Horowitz’s formulation that a CEO’s first responsibility is managing their own psychology.

Being the boss carries a similar misunderstanding. A founder may imagine setting the rules from the top of a pyramid. In practice, employees, customers, partners, users, and the press all make demands. Moskovitz describes arriving on a Monday with a plan for improving the company, then abandoning it when an important employee threatens to leave. The CEO often spends the day resolving conflicts between people and priorities, choosing the action that disappoints the fewest people.

Flexibility has the same limit. The founder is always on call, serves as the team’s example, and sets the pace through their own energy. A passionate founder keeps working because the problem draws attention back to itself. Investors and partners also expect a serious commitment. Moskovitz allows that a small business or niche company can suit a shorter week. Once a company grows beyond two or three people, he expects the founder to work full time.

Money, impact, and the company that already exists

Candidates often tell Moskovitz that a smaller company or their own startup would give them a larger share of the outcome. He uses a slide comparing the possible value of joining a large company with starting two theoretical businesses. An experienced engineer joining Dropbox around employee 100 might have received about 10 basis points, worth roughly $10 million in the slide's example. A person joining Facebook a few years into its life could have made around$200 million, and employee 1,000 in 2009 could still have made around $20 million.

The theoretical startup on the slide is “Uber for Pet Sitting”, a $100 million company in which the founder retains around 10$2 billion after four years, should stop taking the class and build it. The figures are Moskovitz’s illustrations of how to compare risk and ownership, rather than current valuations or a forecast.

He then separates ownership from impact. A late-stage company may offer a large multiplier through an existing user base, infrastructure, proprietary technology, and a strong team. Bret Taylor joined Google around employee 1,500 and invented Google Maps, which reached hundreds of millions of users. Justin Rosenstein prototyped browser chat at Google before it became part of Gmail. At Facebook, he joined around employee 250 and led a hackathon project with Andrew Bosworth and Leah Pearlman that produced the Like button.

These examples show why the company context matters. Rosenstein needed Facebook’s distribution for the Like button, and the same idea would likely have failed as a separate startup. Starting a company is one route to impact. Joining the institution with the users, infrastructure, and team needed to make the work real can be another.

The reason that survives the costs

Moskovitz’s preferred reason is that you cannot avoid doing it. The phrase has two meanings. The founder feels such passion for the idea that the work continues even when nobody has asked for a company. The world also needs the idea, and the founder is unusually well suited to bring it into existence.

Passion helps a founder endure the hard parts and recruit people who can sense whether the conviction is real. The second test concerns the use of time. If an existing company or another team can build the idea better, starting a company may waste the founder’s effort and produce a worse result for the world. Moskovitz wants the founder to ask whether the problem matters and whether this particular person has a reason to be the one solving it.

He returns to Asana with a story from before the company existed. Moskovitz and Justin Rosenstein worked at Facebook during the day and kept working at night on an internal task manager. They could see its value inside Facebook and believed it could help people outside the company. The problem had existed for a long time, yet the available solutions remained incremental. They became convinced that nobody else would build the version they thought the world needed. The work kept pressing towards a separate company until they had to face the cost of leaving Facebook.

That pressure is Moskovitz’s practical test. A founder should feel compelled by a valuable idea and should have a concrete reason that their company is the best path for it. The lecture ends with recommended books on the slide and a return to the next class.

Limits

This is a 2014 lecture aimed at founders pursuing hyper-growth technology companies. Its market advice, employee-equity examples, company valuations, and claims about startup outcomes belong to that context and date. The speakers use personal experience, anecdotes, named companies, and slide figures rather than a method that would let a reader treat each number as a verified general law. The captions occasionally mishear proper names and phrases, so this note follows the complete caption track and the accessible annotated transcript where wording or names needed checking.

The description links to the course slides and readings at startupclass.samaltman.com/courses/lec01/ and to a discussion page at startupclass.co/courses/how-to-start-a-startup/lectures/64030. The first was DNS-unavailable and the second returned HTTP 525 during source acquisition. Their contents are therefore not used here. The description states that the video is released under CC BY-NC-ND 2.5, whose terms require attribution and restrict commercial use and derivative distribution.

Further reading / references

• Annotated lecture transcript, linked in the video description and read alongside the captions.
• Advice for Ambitious 19 Year Olds, Sam Altman’s recommended reading.
• Good and Bad Reasons to Become an Entrepreneur, Dustin Moskovitz’s recommended reading.
• Course slides and readings, linked in the video description and unavailable during source acquisition.
• Course discussion for Lecture 1, linked in the video description and unavailable during source acquisition.