Create products that do not scale.

One of the most universal tips we give at Y Combinator is to take on the hard work. Many novice founders believe that start-ups either “take off” or not. You create something, make it affordable, and if you come up with the best mousetrap, people, as promised, will come to you. Or do not come, in which case you do not have a market. [one]

In fact, startups take off, because the founders make them take off. There are only a small number of startups that have grown themselves, since usually their buildup requires some effort. Here you can make a comparison with the crankshaft, which was completed cars before the appearance of electric starters. If the engine started, it worked, but the launch was a separate time-consuming process.

Attracting customers

The most common complex work for which the founders should undertake at the start is the independent involvement of users. This should deal with almost all startups. You can not wait for users to come to you. You have to go and bring them yourself.
')
Stripe is one of the most successful startups we have funded, and the problem they solved was urgent. If anyone could sit and wait for users, it is Stripe. However, within the framework of Y Combinator, this project is known for its aggressive involvement of users at an early stage.

Startups that create something for other startups have a large pool of potential clients of other companies that we finance, and no one has taken advantage of it better than Stripe. In Y Combinator, we call the method they invented “Collison installation”. More timid founders ask: “Would you like to try our beta version?” And in case of a positive answer they say: “Great, we will send you a link”. But the Collison brothers were not going to wait. As soon as someone agreed to try Stripe, they said: “Great, and now give me your laptop,” and immediately made the installation.

There are two reasons why the founders refuse to go out to the people and attract users individually. The first is a combination of shyness and laziness. It is better to sit at home and write code than to go out and talk to a bunch of strangers and most likely get rejected by most of them. However, for a startup to succeed, at least one founder (usually the executive director) must spend a lot of time on sales and marketing. [2]

The second reason why the founders ignore this path is that the concrete numbers seem at first too small. Major well-known startups could not start like this, they think. And they make the mistake of underestimating the power of aggregate growth. We ask each startup to measure progress by weekly growth rates . If you have 100 users, you need to get another 10 over the next week for growth to be 10% per week. And even if 110 does not seem to be much better than 100, maintaining growth rates at around 10% per week, you will be surprised what big numbers you will end up with. In a year, you will have 14,000 users, and in two years, 2 million.

You will act completely differently when you receive users in the thousands, and the growth rate will gradually slow down. But if there is a market, then you can usually start attracting users manually and then gradually switch to automated methods. [3]

Airbnb is a classic example of this method. The market is so difficult to get off the ground that you must first be prepared for heroic deeds. In the case of Airbnb, it was home visits in New York, attracting new users and helping existing ones. I remember Airbnb at the time of the Y Combinator: forever with bags on wheels, because on Tuesdays dinners they always flew from somewhere.

Understanding startup fragility

Now Airbnb seems to be a power that cannot be stopped, but at the very beginning the project was so fragile that its success or failure depended on those approximately 30 days of apartment-building in order to attract users individually.

The initial fragility was not only Airbnb. Almost all startups are fragile at first. And this is one of the main things that inexperienced founders and investors (as well as journalists and know-alls on forums) misinterpret, unconsciously evaluating start-ups in the embryonic stage according to the standards of already formed start-ups. It is like looking at a newborn and saying: "This tiny creature can never achieve anything."

Nestrachno if journalists and know-alls do not believe in your startup. They always understand everything wrong. It is also normal if investors do not believe in your startup, they will change their mind when they see your growth. The most dangerous thing if you do not believe in your startup. I often have to cheer on founders who do not see the full potential of what they create. Even Bill Gates made such a mistake. He returned to Harvard for the fall semester after the founding of Microsoft. And although he remained there for a while, he could not return at all if he understood that Microsoft would become even a fraction of what it had become. [four]

At an early stage of a startup, you should not ask “Will the company conquer the world?”, But “How large can a company become if its founders do everything right?”. And the right steps can seem time consuming and illogical at the same time. Microsoft was not very impressed when it was just a couple of guys in Albuquerque who wrote Basic interpreters for the market of several thousand hobbyists (as they were then called), however, looking back, it was the best way that allowed the company to take over position in the software market for personal computers. And I know that Brian Chesky and Joe Gobbia (the founders of Airbnb) did not think that they were on the road to great success when they were doing “professional” pictures of the first apartments that were surrendered. They were just trying to survive. But, as time has shown, this was also the right way, which allowed us to take a dominant position in the huge market.

How to find users to attract manually? If you create something that solves your own problems , then you only need to look for others like yourself, which is usually quite simple. In other cases, you need to make a more thoughtful effort and find the most promising flow of users. You can usually get the first group of users by making a relatively untargeted launch, and then observe who will show more enthusiasm and look for people like them. For example, Ben Zilberman noticed that many of the first Pinterest users were interested in design, so he attended a conference of design bloggers to attract users, and it worked perfectly. [five]

User pleasure

It is necessary to do everything possible not only to get users, but also to make them feel happy. Wufoo, for example, sent each handwritten letter a letter of thanks for as much time as they could (it was surprisingly long). Your first users should feel that they have made the right choice by registering with you. And you, in turn, must find new ways to please them.

Why do we have to teach this start-ups? Why don't the founders understand this themselves? I think for three reasons.

First, many startup founders have been trained as developers, and work with clients does not include such training. You are expected to develop something reliable and beautiful, and not slavishly attention to each user, as from a sales person. Oddly enough, the development is opposed to mentoring, and this tradition arose at a time when developers were less influential and stood only at the head of their narrow field of development, and not commanded by a parade. You can be stubborn if you are Scotty, but not Kirk.

The second reason why the founders do not pay enough attention to individual customers is the fear that this tactic will not work. When the founders of start-ups in the embryonic stage express such anxiety, I explain that in the current state they have nothing to lose. Maybe if they make a special effort to make existing users super-happy, one day there will be so many users that they can no longer continue to do so. It would be great to face such a problem. And when this happens suddenly, you will realize that it is easier to please users than it seemed. Partly because you can usually find easier ways than originally intended, and partly because the desire to please users by then will firmly enter your culture.

I have never seen a startup who would be at a dead end due to the fact that he was trying too hard to make his first users happy.

But the main reason why the founders do not understand how attentive they could be to their users is probably that they themselves have never experienced such attention. Their standards for working with clients have been shaped by the companies whose clients they are, and these are mostly large companies. Tim Cook does not send you a handwritten thank you for buying a laptop. He can not. And can you! This is one of the advantages of a small company: you can provide a level of service that no large company can provide. [6]

Once you understand that user interaction is not limited to existing traditions, you should think about how far you can go, trying to please your users.

User experience

I tried to come up with a phrase that conveyed how excessive attention should be to the user, and I realized that Steve Jobs had already done it: it must be insanely great. Steve did not use the word "insane" as a synonym for "very." He literally meant that it was necessary to concentrate on the quality of performance to a degree that would be considered pathological in everyday life.

With all the most successful startups that we have funded, this has been the case, and this probably does not surprise potential founders. What novice founders do not understand is what the desire for a “insanely great” rudimentary startup turns into. When Steve Jobs began using this expression, Apple was already a successful company. Steve had a Mac in mind (including his documentation and even packaging - this is the nature of obsession), which was supposed to be insanely great in terms of design and execution. Developers understand this easily. This is just a more uncompromising version of the development of a reliable and beautiful product.

The founders are hard to understand (maybe Steve himself was hard to understand), which gives “incredibly great” in the first few months of a startup’s life. Not the product should be insanely great, but the user experience of your users. For a large company, the product is the dominant component. But you can and should provide users with an incredibly great experience with an early, incomplete and imperfect product, compensating for the flaws with attention.

Perhaps this can be done, but is it necessary? Yes. Excessive attention to the first users is not just an acceptable method of stimulating growth. For most successful startups, this is a necessary interconnection element to get a good product. Creating a better mousetrap (traps for customers) is not an indivisible operation. Even if you start like the most successful startups and create something you need, your first product will never be correct enough. In addition to areas where mistakes cost too much, it is often better not to tune in to perfection from the very beginning. This is especially true of software: it is usually better to present the product to users as soon as it has some utility, and then look at what they will do with it. Perfectionism is often an excuse for delaying, in any case, the first user model is always inaccurate, even if you are one of them. [7]

The feedback you get by communicating directly with your first users will be most helpful. Becoming a large company, you will be forced to use focus groups and regret that you cannot come to your users home or work and see how they use your product, as you did when there were very few of these users.

Flame retardation

Sometimes it will be right to concentrate on a very narrow market. It’s like making a campfire: first you make a small fire and let it burn, then add more wood.

That was the case with Facebook. At first it was intended only for Harvard students. In that form, its potential market was only a few thousand people, but since the resource did satisfy their needs, a critical mass was registered. Further, for some time certain colleges were added to Harvard. When I spoke with Mark Zuckerberg at the Startup School, he noted that compiling a list of courses for each college was a laborious task, but it was thanks to him that students felt at home on the site.

Any startup that can be described as a market usually starts in any market segment, but it can work for other startups. You should always think about whether there is a market segment where you can quickly reach a critical mass of users. [eight]

Most startups that use flame control strategies do so unconsciously. They create something for themselves and their friends, who become the first customers, and only later realize that they can offer their product to a wider market. The strategy also works well with its unconscious application. However, the greatest danger of the unconscious application of this model lies in the naive abandonment of its part. For example, if you do not create something for yourself and your friends, or even if you create, but come from the corporate world, and your friends are not your first customers, you no longer get the perfect initial market on a silver platter.

Among companies, the best first customers are usually other startups. They are more open to new in nature and, having just begun work, have not yet had time to make a choice on all points. Plus, if they prove successful and grow quickly, you will grow with them. One of the many unforeseen advantages of the Y Combinator model (and especially the Y Combinator extensions) was that B2B startups now have an instant market on hand from hundreds of other startups.

Path meraki

Hardware startups can do something like Meraki. And although we did not finance Meraki, the founders were graduates of Robert Morris, so we know their history. They really started from the difficult: they themselves collected their routers.

Hardware startups face an obstacle that is not in the way of software projects. The cost of the minimum order to be sent to the factory is usually several hundred thousand dollars. This can drive you into a trap-22: without a product, it is impossible to provide the growth necessary to raise funds for the production of a product. Previously, when hardware startups had to rely on investors' money, it was necessary to be very convincing in order to overcome this obstacle. The emergence of crowdfunding (or rather pre-orders) helped a lot. But still, I advise startups to act like Meraki whenever possible. This made Pebble. They collected the first few hundred hours on their own. If it were not for this stage, they probably would not have sold hours for 10 million dollars, finding themselves on Kickstarter.

As well as excessive attention to the first customers, independent production turns out to be valuable for hardware startups. You can quickly change the design, and you will find out what you could not otherwise know. Erik Migikowski of Pebble said that among what he learned was “how important it is to have good screws.” Who would have thought?

Counseling

Sometimes we advise the founders of B2B startups to bring attention to customers to the extreme: choose one user and act as if they are consultants who create something for him alone. The first user serves as a mold for reflux, continue to adjust it until you perfectly match his needs, so usually you can get what other users want to use. Even if there are not many of them, there are probably related markets where more can be found. If you can find one user who really needs something, and will work to meet this need, you will work on what people need, which is what a startup should initially do. [9]

Counseling is a typical example of complex work. It (like other ways of displaying attention) is safe until you get paid for it. This is where companies cross the line. While you are a product company that is simply too attentive to the customer, they are grateful to you, even if you do not solve all their problems. But as soon as they begin to pay for this attention, you are expected to do everything.

Another advisory-like method of attracting initially reluctant users is using their own software on their behalf. We did this in Viaweb. When we communicated with sellers and asked if they would like to use our software to create online stores, some answered that they would not, but allowed us to create such online stores for them. Since we were ready for anything to get users, we did it. We felt pretty lousy at the time. Instead of organizing major strategic e-commerce partnerships, we tried to sell suitcases, pens, and men's shirts. Now we understand that then it was the right thing to do, because thanks to him we understood how sellers would feel when using our software. Sometimes the feedback was almost instant: creating the site of a store,I understood that a function is needed, which we do not have, so I spent several hours implementing it and continued to build the site.

Manual work

There is a more extreme option, when you do not just use your software, but you yourself work in software quality. With only a small number of users, sometimes you can manually do what you plan to automate later. This allows you to run faster, and when you eventually automate the process, you will know exactly what to do, because doing these actions on your own will provide you with muscle memory.

When manual components appear to the software user, it begins to resemble a cruel joke. For example, when Stripe distributed “instant” accounts to its first users, the founders simply manually secretly registered regular accounts.

Some startups at the initial stage can be completely manual. You can find someone who has a problem to solve, solve this problem manually and continue to do it as long as possible, gradually automating the bottlenecks. You may be a little intimidated by the fact that user problems are not yet solved in an automated way, but this is not as scary as the more common case of having an automatic function that does not solve anybody’s problems.

Waiver of loud launches

It should be noted one initial tactic that usually does not work: a loud launch. I sometimes meet founders who believe that start-ups are unmanaged rockets, not manned aircraft, and that success can be achieved only if the launch will be done with a sufficient initial speed. They want to receive publications immediately in 8 different editions. And to start, of course, on Tuesday, because they read somewhere that this is the best day to start.

Understanding how little starts are made is very simple. Think of any successful startups. How many of their launches do you remember? All that needs a launch is some initial user base. How well things will go in a few months depends on how happy you make these users and how many were there. [10]

So why do the founders think that launch matters? This is a combination of selfishness and laziness. They think that what they create is so great that everyone who hears about it will register immediately. In addition, the work will be significantly less if it will be possible to receive users, simply declaring their existence, and not attracting each separately. Even if you are really creating something great, getting users will always be a gradual process, partly because the great is usually new, but mainly because there is something for the user to think about besides your product.

Partnerships also usually do not work. They do not work for startups in general, but especially do not work if they are used to stimulate growth. A common mistake of inexperienced founders is the belief that a partnership with a large company will provide a major breakthrough. After six months, they all say the same thing: the work turned out to be much more than we expected, but in the end we didn’t get much from it. [eleven]

It is not enough just to create something extraordinary from the very beginning. From the very beginning extraordinary efforts must be made. Any strategy that does not involve effort - waiting for a loud launch to get users or a large partner - is suspicious due to the fact.

Vector approach

The need to do something frighteningly time-consuming at the very beginning is almost universal, so it may be worth stopping to consider the ideas of startups as scalars. Instead, you should try to treat them as a combination of what you are going to create and the hard work you are going to do to make the company work.

Such an approach to start-up ideas can be interesting, because now, when there are two components, you can try to creatively approach both the second and the first. However, in most cases, the second component will be what it usually is: manual involvement of users and providing them with an unusually pleasant experience. The main benefit of a vector approach to startups is to remind the founders that they must work hard in two directions. [12]

At best, both components of the vector will be a valuable contribution to the DNA of your company: the hard work you have to do at the start is not an inevitable evil, but irreversible changes for the better. If you have to aggressively attract users while you are still small, probably becoming large, you will not cease to be aggressive. If you have to produce your own hardware or use your software on behalf of users, you will learn something that you could not learn in another situation. And most importantly, if you have to work hard to please users, when you have very few of them, you will continue to do it when their number becomes large.

Notes

[1] In fact, Emerson never specifically mentioned a mousetrap. He wrote: “If a person has good grain, wood, boards or pigs for sale, or he can make chairs, knives, boilers, or church organs better than anyone, a broad, well-trodden road will lead to his house, even if the house stands in the forest. ” Back

[2] Thanks to Sam Altman for the advice, I will say straight. No, you can’t escape sales by hiring someone to do it for you. First, you yourself must be engaged in sales. Subsequently, you can hire a real sales person who will replace you. Back

[3] This works because, as you grow, your size helps you grow. Patrick Collison wrote: “At some point there was a very noticeable change in Stripe’s self-perception. He ceased to be the clump we were supposed to push, and became a carriage that moves by inertia. ” Back

[4] One of the more tricky ways that Y Combinator can help the founders is to adjust ambitions, because we know exactly how many successful startups looked when they started. Back

[5] If you are creating something for which you cannot put together a small group of users that you can monitor (for example, corporate software), and in an area where you have no connections, you will have to rely on cold calls and recommendations. But is it worth taking on such an idea? Back

[6] Harry Tan has noticed interesting traps into which the founders fall at the start. They want to appear so large that they even imitate the shortcomings of large companies, such as indifference to individual users. It seems to them more "professional." In fact, it’s better to come to terms with the fact that you are small and to take full advantage of this situation. Back

[7] Your user model is almost never perfectly perfect, because user needs often change in response to what you create for them. Create a computer for them, and they suddenly want to work on it with spreadsheets, because thanks to the appearance of your computer, someone invented spreadsheets. Back

[8] If you need to choose between the group that will register the fastest and the group that will pay more, it is usually better to choose the first one, because these people are likely to become your first customers. They will better influence your product, and you do not have to spend so much effort on sales. They have less money, but you don’t need a lot to maintain targeted growth at an early stage. Back

[9] Yes, I can imagine cases where you can end up working on something that only one user actually needed. But this is usually obvious even to inexperienced founders. Therefore, if it is not obvious, creating something for the market from one user, do not worry about it. Back

[10] There may even be an inverse relationship between the grandeur of launch and success. The only launches I remember are such famous failures as the Segway and Google Wave. Wave is a particularly disturbing example, as I believe that initially the idea was great, but partly it was killed by too bloated launch. Back

[11] Google grew up with Yahoo, but it was not a partnership. Yahoo was their customer. Back

[12] It will also remind the founders that an idea, the second component of which is empty (an idea where you can’t do anything to make it work because you don’t know how to find users that you can draw manually) is probably a bad idea, at least least for these founders. Back

Thanks to Sam Altman, Paul Buchgate, Patrick Collison, Kevin Hale, Stephen Levy, Jessica Livingston, Geoff Ralston and Harry Tan for reading the drafts of this article.

*** The title of the original article sounds like - Do Things that Don't Scale. There is a definite play on this: on the one hand, scale - scaling (of the project); on the other hand, unscalable is impregnable, difficult. The first value is the most familiar and generally accepted, but in essence is a provocation. The second - reflects the essence of the article. In general, do not succumb to provocation)) The

translation was made as part of the summer school of start-ups Tolstoy Summer Camp and MetaBeta .