← Back to blog

A Summary of Paul Graham's Advice for Startups

Originally published Translated

This article was machine-translated from Japanese by AI (Claude). It may contain errors. Read the original Japanese article.

These are my own notes from reading Paul Graham's essays. I still find them insightful when I come back to them, so I decided to write them up on the blog as well to make them easier to read. A list of Japanese translations of Paul Graham's essays is available here. Please do read the originals too (the sources are listed at the end).

This is a reprint of slides I made a while ago.

https://www.slideshare.net/takaumada/paul-graham-advicesonstartups

Advice for Startups from Y Combinator Founder Paul Graham (Paul Graham's Advice to Read When Your Startup Is Lost) from Takaaki Umada

Principles

Make Something People Want

The hardest thing for a startup is making something people want (the second hardest is raising money). Keep launching, persistently and quickly, and don't give up until you make something people want.

Know Your Users

If you know your users, you can keep your morale up even when things are at their worst, and having just ten users who love your product is enough to keep you going.

Do Things That Don't Scale

To get to know your users, Do Things That Don't Scale. Then you can give customers surprisingly good service and get to know them better. It is far better to be loved by a small number of users than to be liked by many.

Startup = Growth

A startup is not simply a newly founded company, nor is it starting a company in the technology sector. An IPO or an M&A is not what makes a startup. What a startup absolutely needs is "growth." Everything else is merely a factor in achieving growth.

If you head in the wrong direction, you will never succeed. So never forget to pursue growth. If you chase growth, everything else will fall into place. When in doubt, use whether it leads to growth as your compass. Make whether it leads to growth the criterion for every decision.

Terrible Things Will Happen

Terrible things will happen. That is the nature of startups. No startup escapes disaster. But don't let that discourage you. If you lose motivation, the startup dies. If it doesn't die, it will eventually succeed.

Ideas

Six Principles for Making New Things

  1. Look for simple solutions,
  2. to problems that are overlooked,
  3. and that are actually worth solving,
  4. present the solution in the most informal way possible,
  5. start with a rough version 1,
  6. and iterate quickly.

From Six Principles for Making New Things.

Good Ideas Look Stupid at First (Knowing the Secret, Black Swans)

Successful founders know about "a problem other people haven't noticed." And the best idea is one that "you want yourself," "you can build yourself," and "most people don't recognize as valuable."

People who are strong in technology can be said to have a lead in ideas for that reason alone. Technology advances quickly, and ideas that were infeasible in the past sometimes become solvable without anyone noticing. Young people have the same advantage.

However, make something that somebody wants. Don't make something nobody wants. But there is no point in asking someone whether they want your product before you release it, so launch early. When you do, start with a small number of people who will love your product. Make something that people absolutely want to use, even if it is a crude version 1.0. They will give far better feedback than a large number of users who merely sort of like it.

Be Flexible

Talk with your users and get their feedback. And if you want to benefit from talking with users, you need to be willing to change your idea. Stay flexible enough to iterate quickly and change your idea.

If You Want to Tackle a Big Problem, Avoid a Frontal Attack

Start with a small market. If you want to build a website used by hundreds of millions of people, you can build a stalking site for Harvard students. If you want to dominate the computer OS market, make a Basic interpreter for machines used by a few thousand people.

However, don't pick a small niche market simply because you want to avoid competition. If you try to make something good, competitors will appear, and it is important to face them rather than avoid them. Avoiding competition ends up meaning avoiding good ideas.

If you are competing with superior technology, small customers are better. You can't beat large companies on sales power, and in the world of technology, the low end is always eating the high end. Make cheap products powerful. For example, take a medium-sized non-technology company and observe for about two weeks what they do with computers.

Problems

The Best Problem to Solve Is One You Have Yourself

Google came into being because its founders wanted information online. Apple came into being because its founders wanted computers. Avoid derivative ideas. That is somebody else's problem. It is far better to start from your own direct experience. So rather than asking "is this idea worth starting a company over?", just fix what you personally found to be a problem. By doing so, you end up making something many people want.

Live in the Future and Build What's Missing

Just live in the future and build what is missing.

You Don't Think of Problems, You Notice Them

Startup ideas are not "thought of" but "noticed." At Y Combinator, ideas that come out of the founders' experience are valued highly. If you notice an existing inefficiency, it can become a problem to solve. To notice it, live in the future and build what is missing. If, in the course of doing so, you notice an unknown need you can address, it is like a gold mine. You do have to work hard to dig out the gold, but what matters most in a gold mine is knowing where to dig.

For example, if you work at a large company now and, in the course of your work, come up with a problem like "I would pay if someone built X," that could become a good product idea.

Observe Customers

YC partner Blackwell's recipe for starting a company sums it up: "Watch people who have money, see how they spend their time, figure out a solution, and sell it to them. Profitable solutions are surprisingly small and still untouched."

Growth

Are You Growing Properly?

To grow big, you need to make something a large number of people want and reach everyone who wants it. Software excels at reaching people, but the constraint remains whether you can make something a large number of people want. For a software company to grow, make something a large number of people want.

In the early stage, a startup may not grow. But once you find a way to make and deliver a product people want, rapid growth begins. The phase called a startup is the phase of rapid growth, and it is measured not by the absolute number of customers acquired but by the ratio of new customers to existing customers.

Launching a startup is an optimization problem. Focus makes you efficient. If you focus on the growth rate, you no longer need to think about the various problems a startup would otherwise have to deal with. So optimize your startup for the growth rate.

To check whether you are on the right track, ask not "what growth rate makes something a startup?" but "at what growth rate do successful startups grow?" At Y Combinator, weekly growth rates are measured, and a good growth rate is 5–7% per week. Achieving 10% can be called extremely good.

The best way to measure the growth rate is to look at revenue growth. If you don't charge, measure by the number of active users, because the number of users is an indicator of later revenue.

Pursue Growth and You Will Succeed

If you make growth your goal, you will gradually become optimized and will also come up with new ideas. You might think that pursuing growth alone could cause problems. But achieving a weekly growth rate is exactly what drives founders to act. And taking action leads to success. So when in doubt, use growth as your measure. Use the growth rate as a compass for every judgment.

The roles of the CEO and CTO change completely every 6–12 months. You will spend half as much time writing code, and you will need to deal with the various motivations of employees. So be prepared. And once you are on a growth trajectory, the stress may be far lower.

Product Development

Launch Quickly, Talk to Users, Improve, and Launch Again

Real learning begins after you release and start talking to real users.

Start with the first version simple and minimal. Pride makes people reluctant to release something that has room for improvement, but serious learning begins after you release the first version. Keep in mind that if you are not embarrassed by what you released, you released too late. That said, you must not launch too early either. Find a core that is useful in itself and can gradually be expanded to the whole project, and release just that part.

Engage with Users

Product development is the conversation with users that truly begins after release. And if you want to benefit from that conversation, you will need to change your idea. The idea at the time of founding is a hypothesis, not a blueprint. Do whatever seems best at the time.

Acquiring users is hard, because it is difficult to tell whether your product simply isn't reaching users or is just bad. Even good products can lose users. If you want to know how to collect data on what users want, or how to get closer to users, consult the advisors around you. They often have wisdom.

Shipping Constantly Keeps Engineers Engineers

Steve Jobs's maxim "Real artists ship" means not only that artists ship but also that artists want to ship. So if you don't let them ship, the artists will be gone. It is important to let engineers release code right away and keep them engineers at all times.

There are many places to expand around existing technology. The question of what to build first is the most important, and in choosing which product to build, you choose based on "how long will it take to build" and "how excited will users be about the quality of the product." Once the release is done, talk about the basic line for acquiring users and selling.

Cofounders

You Need Cofounders, but Choose Carefully

For a startup, a cofounder is like location in real estate. Like location, a cofounder is hard to change. And a startup's success depends on the work of the founders. Founding alone is usually too hard to work. In particular, you need a colleague who cheers you up at the worst of times. You need cofounders, but you can't swap them out. Choose carefully.

Choose cofounders for character and dedication, not ability. Failures in particular come from a lack of character and enthusiasm. If possible, found a company with someone with whom you have had a relationship for more than a year. Even so, maintaining the relationship after founding is a struggle. Co-founding is closer to marriage than to friendship. A startup is the child of the cofounders.

Whether a technical startup needs a business-minded cofounder depends. But the team needs at least one person who understands customers and is glad to focus on what customers want. That person doesn't have to be on the business side, of course; they can be a programmer.

A Founding Team That Doesn't Need Our Help

At Y Combinator, teams are divided into "promising teams," "unpromising teams," and "promising teams stuck on an unpromising idea." Y Combinator accepts the first and third cases.

A successful startup gives the impression that "they'll be fine without our help." They are not described as "really smart" or "having a great idea."

In other words, the qualities that underlie success include mental strength, adaptability, and determination. Choose cofounders who, as a team, have such qualities.

Fights Will Happen

Even the best startup teams fight. There are more fights than people think. When a fight breaks out, consult an advisor. They have surely seen similar situations many times and should be able to mediate well.

Hiring

Hire Only After Careful Consideration

Hiring creates extra work and can worsen the growth rate in the short term. Also, if there is a way to keep achieving your growth rate without adding people, do that.

The general advice on hiring is to avoid hiring as long as you can, pay in equity rather than salary (which attracts dedicated types), and hire only people who write code or go out and get users. Those are the only people you need at first.

The Animal Test for Deciding Whom to Hire: "Can You Compare This Person to an Animal?"

Great people for a startup work compulsively, like animals. They are excessively serious and, once they start, don't stop working. A salesperson doesn't come back until the deal is closed; a hacker would stay up until morning rather than go to bed with a bug left in the code. Those are the animals. As for designers, the preferred type is someone who feels physical pain when something is off by two millimeters.

Ask whether the person you are hiring is animal-like. No matter how smart someone is, if they are not animal-like they are not suited to a startup.

Don't Hire Bad Programmers

For programmers, in addition to the animal test, test "are they genuinely smart," and if so, "do they get things done," and finally "can you stand being around them." People with big attitudes are often not smart, so they mostly fail the first test (smart people don't try to appear smart).

And keep in mind that good programmers don't want to do the job of implementing exactly what business people tell them.

Cash Flow

Making Ramen Money Changes How You Raise Funds

Earn your day-to-day living expenses (enough revenue to eat instant ramen, the cheapest food, every day). Then you no longer need to raise money to survive, and the relationship between founders and investors changes completely. You can take a stronger stance with investors, and investors like startups that don't need their investment anyway. If you are making ramen money, investors can tell that the founders have customers who pay them, are building what people want, and are cost-conscious. That shows investors that this startup is trying to make money.

Also, once you can make ramen money, morale rises. Gradually you feel as though you are earning your own living and begin to feel like a company. Being able to cover your own living expenses is the important turning point, and from that point on, survival rather than going under becomes the default.

Above all, if you don't need to raise money, you don't have to interrupt your work to do so. Interrupting work for fundraising can be fatal for a startup, so always think of "this fundraising as the last one."

Cut Spending

When you run short of money, you usually have no choice but to cut people. Firing people is hard, but you have to accept that now is the time. Cutting everyone's salary is not a good idea.

(If It's Appropriate) Be Able to Do Consulting

Consulting can earn you ramen money. But consulting doesn't scale. It is easy to become a $3,000-a-month consulting shop, and it will get you through a tight spot, but it amounts to becoming a programmer under contract who sells cheap. Contract work and consulting can also make you do things you're not naturally good at.

Consulting does have good sides, though. You can understand users more deeply. And with consulting, you may be able to win customers you couldn't catch by selling software.

Fundraising

Fundraising Is a Distraction for Startups: Even So, You Need to Raise Money

You need to raise money either to stay alive or to grow rapidly. A startup doesn't immediately produce enough customers or revenue; it takes time. And how long it takes you can't know until you try. Without raising money a startup won't become a huge success, but for a startup fundraising is a sideshow.

During the fundraising period, surprisingly little else gets done. At worst, work stops for months. One person should be assigned to fundraising full-time while everyone else stays focused on the product. The company's resources shrink, but this is still the best move.

You should raise money only when you need it to grow rapidly and have the potential to grow rapidly.

If you are not in fundraising mode, avoid meetings with investors. Even if they ask, you should politely decline.

If you do enter fundraising mode, try to finish in as short a time as possible. Get it over with and go back to work. A long slowdown is the worst thing for a startup, and since startups are short of money, the longer negotiations are delayed, the more often the startup ends up conceding. If an investor is interested, don't let them sit back; have them write a check while they're enthusiastic.

Decide whether to take an investment using the equation 1/(1 – n). What you give up (n) should make the company's value larger.

Don't Spend the Money You Raised

If you aim for rapid growth with a large sum and try to get many customers, you end up hiring many people quickly. The biggest use of money is hiring people, but that is generally not a good move. The best use of raised funds is not to spend them. Lower your burn rate.

Also, don't get so busy spending a large sum. You should focus on building what customers really want.

Don't Expect Too Much from Fundraising

Fundraising is far harder than you expect, and it breeds disappointment in startups. Disappointment will kill a startup.

Investors

Understand How Investors Work: It Explains Why They Are Timid and Unscrupulous

Investors are usually incompetent. They are forced to make decisions about things they don't understand, and they are often wrong. Because they have to decide about things they don't understand, investors become timid and may even jerk founders around.

VCs are timid mainly because of their compensation structure. VCs typically take 2% of the fund per year as operating expenses (base salary) and receive 20% of the profits as an incentive. One person can't manage many deals, so each deal has to be large (even though a huge sum of money can sometimes be bad for a startup). That is why they have to be timid, and they have to behave unscrupulously. Founders should also know that for investors, virtually all returns come from a few companies, and the other startups are a cost.

Avoid new VCs and second-tier VCs. New ones don't look scary but may be nervous and dangerous. If you want money from a new investor, either keep control yourself or use them for part of a round in which another investor holds the lead. Second-tier VCs usually only get deals that top-tier VCs turned down, and you often end up working with founders better than yourself. If you have the choice, avoid second-tier ones as much as possible.

Investors are most influenced by the fact that "other investors are investing." While negotiating a financing, don't answer when a VC asks about other VCs you're negotiating with. Angels coordinate, and VCs sometimes band together to lowball a company's valuation. And because of this "other investors are investing" quality, persuading the first investor is painful, but the second and later ones are easy.

Founders Need to Manage Investors

Just as startups need investors, investors need startups. To generate fund returns, they need to invest in startups that are especially likely to succeed.

Investors may have useful insights, so you shouldn't ignore them. But the startup needs to manage them so as not to be dominated. These days it is normal for the startup to keep control of the company even after Series A.

Pitching

Focus on Achieving Two Things: "Explain What You Are Doing" and "Explain Why Users Will Want It"

Choose a limited description over a vague one. A description that says you can do anything has close to zero content.

Don't spend time on side topics beyond those two points. It is enough to list about three competitors and explain the difference in one line.

Bring a Demo Early and Talk About Users

There is no need to go on at length about the business model. In a short pitch, investors don't want to hear about the business model, and early-stage business models are often wrong. And a demo speaks louder than any words. Do the demo early.

Talk about users. What investors fear is founders building something nobody wants based on a theory of their own. If people are already paying a lot of money, that is a sign you have found the users' needs.

Don't forget to create a catchphrase that investors will remember. Investors hear many pitches and forget over time. To avoid being forgotten, build a phrase that sticks in memory.

Get an Enthusiastic Investor to Say Yes First

Focus on what the investor is most likely to say "Yes" to. An investor's value can be expressed as amount × probability of Yes. You should focus on the latter. Once one investor says "Yes," other investors rapidly start to take an interest in you. So persuading an enthusiastic investor is the best way to persuade a less enthusiastic one.

Decision-Making

Don't Trust Your Intuition About Startups (But Do Trust Your Intuition About People)

Startups make many mistakes and later regret not having "listened properly" to the advice of Y Combinator partners. This is because decisions about startups are counterintuitive. So don't follow only your own intuition; try listening to what your mentors say. If you aren't convinced, push back and ask why, as Sam Altman did.

But you can trust your intuition about people. Many of the mistakes young founders make tend to come from not trusting their intuition about people enough.

Whether You Succeed Depends on Determination: Don't Get Distracted by Irrelevant Things

Whether a startup succeeds depends not on the founders' intelligence or talent but on their determination. It might seem that intelligence matters more in technical startups, but even in the ultimate high-tech industries, success depends more on determination than on intelligence.

Determination is stubbornness, and you need not only to be stubborn but also to be hard on yourself. You must balance your stubbornness with self-discipline, and you must have ambition. Once you resolve to do big work, temptations around you increase, so you need to strengthen your self-discipline accordingly. Without self-discipline, stubbornness gradually prevails, and what you can achieve will be mediocre.

Above all, nothing ruins a startup like distraction.

Procrastinate Well

Neglect small tasks for the sake of big ones. Everyone who gets things done is a master of procrastination. There is an infinite amount to do, and while you are doing one thing you can't do anything else. So don't avoid procrastinating; procrastinate well.

Real work needs a block of time and a good atmosphere. Sometimes, put off all your obligatory work. A startup should always ask what the best work it can do is and why it isn't doing it. It may not be fun, but do it.

Work

Tedious, Boring, Unpleasant Work (Schlep) Is Unavoidable by Definition

If you want to do great work, tedious and boring work is unavoidable. Writing programs alone does not make a startup succeed. You could say a company is defined by the tedious work it takes on. The bigger the work, the more tedious work comes with it, and you must not run away from it. If you don't run away, it becomes a bold idea, and its value doubles. By tackling the bold and tedious idea of payments head-on, Stripe got, in effect, a cheap stock in ideas.

The way to deal with slow-moving organizations is to separate dealing with them from your main business. And accept being rejected by prospective partners. Investors invest in only about 1% of the plans brought to them, and most customers will decline too.

Get Advice

The advice from many of the people who speak at Y Combinator is consistent. Listen to the advice. To get advice, put yourself in an environment with people who help founders. Even just having someone take a strong interest helps a startup.

Community is also highly valuable. The extent to which the volunteer-spirited exchanges among startups contribute to startups is immeasurable. And for startups, getting rich is not a zero-sum game. Help each other.

A Startup Is Not a Job

Everyone expects a startup to be like a job. But in reality it isn't. In a startup you need to be careful in choosing cofounders and pay attention to maintaining relationships, whereas in a job you don't have to choose your colleagues. Discard the mental model of a job. If you can't believe terrible things happen at a startup, it is because you are being dragged along by your image of past jobs. A startup is not a job. When you are going through something awful, think of it that way.

Failure

The Only Failure That Kills a Startup Is One: "Making Nothing Users Want"

Startups die when they run out of money, but a lack of money is not the real cause of a startup's death. The real cause is the single reason that "they failed to make anything users want." Stop looking for tricks to pass the test, as with exam prep or university entrance exams. We have been too poisoned by schooling, and we believe there is a trick somewhere to making a startup succeed, but in reality there almost isn't one. The market is harsher than school. A startup only has to make what users want, but that is the hardest part.

Many Failures Threaten You, but Many Failures Are Not Threats

The conspicuous failures for a startup are not as threatening as they look. Failure is routine for startups, and if handled properly it is not fatal. And know that even famous startups, the kind that speak at YC, fail. Whether a competing patent turns up, the servers go down, or you run into technical problems, deal with it properly.

Counterintuitively, startups almost never fail because competitors appear. One reason people overreact to competitors is that they overestimate their idea. What matters is execution, and in the end building your own product and acquiring users is the hardest part, so don't be misled by press announcements. You shouldn't worry even if a big-company competitor appears. Even if Google makes a similar product, they have no fear of going under, and they have bureaucratic procedures that slow them down, so you can outmaneuver them.

But watch out for competitors that are not yet known. Looking only at competitors that already exist tends to make you complacent, but you should compete with whatever others might make happen. If you let your guard down, they will catch up at tremendous speed.

Stay Flexible and Stay Focused

The worst thing for losing focus is side businesses and consulting that bring in money. Being interrupted to answer calls from current customers, or holding fundraising meetings, all cause a startup to lose focus.

Failed founders were often hard to have a conversation with. They were not flexible. Failed founders had the ability to do everything they were told, but when they reached an unwelcome conclusion beyond the words they were told, they did not do what they needed to do even though they disliked it. In other words, they couldn't get serious. Even if you are smart, don't lose your drive. Knowing this will surely help.

Mindset

A Startup Dominates Your Life and Shakes Your Emotions Violently

With any startup, the startup dominates your life more than you expected. And you go through many emotionally jarring experiences. You may enjoy a moment you thought was wonderful, only to have it turn into despair a few hours later. Expect that. And expect that something worse than even your worst expectation will happen.

You Need Resilience

Most deals fall through. Try not to get your hopes up. And even if you get down, don't give up. Just not giving up can take you surprisingly far. Never stop improving your product; keep improving it. Don't get discouraged. The direct cause of a startup going under is often a lack of money, but the real cause is when you fail to stay focused and lose your drive. Don't give up.

Be Relentlessly Resourceful

A great founder can be summed up in two words: relentlessly resourceful. The opposite word is hapless. Making what people want is the goal, but staying relentlessly resourceful is how you reach the goal.

Luck Matters

The success rate of selling a product can be expressed as technology × persistence × luck. Failed founders tend to blame themselves, but if luck is zero, everything fails. Usually, successful founders don't realize they were lucky. Luck matters a great deal for ordinary people. Hackers place the highest value on technology and probably get what they expect, but business is not like that; luck matters. So stay relentless and resourceful until luck comes.

Table of Contents

References

Principles
Ideas
Problems
Growth
Product Development
Cofounders
Hiring
Cash Flow
Fundraising
Investors
Pitching
Decision-Making
Work
Failure
Mindset

Basically, everything here is Paul Graham's insight and not that of the compiler of this summary, but any misunderstandings are the compiler's responsibility.