Paul Graham: Do things that do not scale

One of the most popular pieces of advice we give to Y Combinator - doing things that do not scale. Many of those who dream of becoming the founders believe that start-ups or fly or not. They say that you are building something, make it accessible to users, and if you have failed right mousetrap, people will line up to your door. If not - then the market just does not exist. [1]

In fact, startups take off because the founders make them fly. There are a handful of projects, which have grown by themselves, but for the most part, needed a serious push to cultivate them. A good metaphor can serve as a lever to crank the engine, which existed before the advent of the electric starter. When the engine is started, the machine continues to work, but this is a consequence of a separate and time-consuming process.


The most common non-scalable thing that should do the founders - is a set of user manual. Almost all startups should do it. You can not wait for the people to come themselves. We need to come to them and get them.

Stripe - one of the most successful startups in which we have invested, and they solved very urgent and pressing problems. If someone is able to sit on the couch and wait for the user - so it's Stripe. But they are known among the YC experience aggressive set of early users.

Startups that create products for other start-ups have a large pool of potential customers among the companies in which we have invested, and no one took advantage of better than Stripe. At YC we even use the term "Installing Kollizon" (Kollizon brothers - the founders of the Stripe - approx.) For the technology they invented. Most of the founders shy asked: "Do you want to try our beta (beta - approx.)?", And if they say, "Yes", - they say, "Cool, we'll send you a link." But the brothers were not going to wait Kollizon. When someone agrees to try Stripe, they said, "Well, then give me your notebook" - and installs the application.

There are two reasons for the founders to resist to go and start typing user individually. The first - a combination of shyness and laziness. They would rather zasyadut home for programming, what will communicate with a bunch of strangers is likely that most of them fail. But in order to start excelled, at least one of the founders (usually the CEO) will be forced to spend a lot of time on sales and marketing. [2]

The second reason why startups ignore this path lies in the fact that the absolute numbers at first seem to be too small. They think that a large and well-known start-ups can not start with the little things. Their mistake is to underestimate the cumulative growth. We support the assessment of progress on the weekly "growth rating" for each of our startup. If you have 100 users, you need to score 10 more in the next week to stay at the level of growth of 10% per week. And if you think that 110 is not essentially greater than 100, you will continue to grow by 10% every week, you will be surprised figures that get in the end. In a year you will have 14, 000 users, and after 2 years - 2 million.

We'll have to use other methods, when a set of users will be in the thousands, and over time the growth is still slow. But if your market does exist, you can always start with a manual set of users, and then switch to a more automated methods. [3]

A classic example of this technique - AirBnB. Marketplace so difficult to gain momentum that initially require heroic efforts. In the case of AirBnB I had to go door to door in New York, gaining new users, and helping improve their current listing. When I think AirBnB program YC, I imagine with the founders of hiking backpacks, because when they came for lunch on Tuesdays, they're always just flew in from somewhere.


Now AirBnB looks like a juggernaut that can not be stopped, but in the beginning it was so fragile that the 30 days of the start-up and communication with the first user decides the fate of the project.

Initial fragility was characterized not only AirBnB. Almost all of these start-ups first. And this is what inexperienced founders and investors (and journalists, and the know-all forums) do not understand. They unconsciously judge the larvae start-ups following the example already held. It is as if they were looking at a newborn baby, and stated: "This small creatures have no chance to achieve something in life».

It does not matter when journalists know-all and hammer on your startup. They always appreciate all wrong. Normally, even when investors clog your start - they will change their views when they see growth. The red light should light up the danger when you do start to hammer on his project. I've seen this. I often inspire entrepreneurs who do not see the full potential of the project, which they create. Even Bill Gates has admitted the mistake. He returned to Harvard in the fall semester after the launch of Microsoft. He did not stay there for long, but if he was aware of at least some of those scales will grow to what Microsoft, he would not come back at all. [4]

The right question at an early stage startup is not "whether the company is capable to conquer the world?" And "How big can be the company if the founders are doing the right thing?" And the right things often seem to be too labor-intensive and insignificant. Microsoft did not look very promising, when a few guys in Albuquerque wrote Basic interpreter for the market of a couple of thousand "fans" (as they were called), but in retrospect it is clear that it was the best way to dominance in the software market for microcomputers. < br />
I know that Brian and Joe Gebbia Cesky (founders AirBnB - approx.) Did not feel that they are on the path to great success when doing "professional" photos first apartment hosts. They were just trying to survive. But in retrospect it turned out, too, the best way to achieve dominance in the large market.

Right - Brian Cesky - one of the three founders of AirBnB. Photo: Airbnb Community

How to find users who can draw by hand? If you create something that resolves your own problem, you just need to find your "colleagues in misfortune," which is usually quite simple. Otherwise, it is necessary to put a concerted effort to search for the target group of users. The standard way in this case - to get through the initial selection of members is relatively netargetirovanny launch and follow-up of those who have shown great enthusiasm for the product and the like. For example, Ben Silberman (co-founder of Pinterest - approx.) Noted that many of the first users of Pinterest interested in design. He went on to gain first users conference dedicated to design bloggers, and it worked well. [5]


It is necessary to take measures not only the supernatural order to recruit members, but also to make them happy. Wufoo as long as they could (and it lasted a surprisingly long time), is sent to each new user to a thank you letter written by hand. Your first users must feel that check on your service - this is one of the best things that happen in their lives. And you, in turn, must be refined by searching for new ways to delight them.

Why should we teach the technology start-ups? Why did the founders do not feel this level of intuition? I think there are three reasons.

First. Many founders of start-ups learn from engineers and customer service is not included in the curriculum of engineers. You must create a strong and elegant things, not provided in the slave depending on the desires of users, like some kind of prodazhnik. It's funny that one of the reasons why engineers are so afraid of manual controls - a tradition that appeared when power engineers was much less then they could create only part of the mechanism rather than an integrated product.

The second reason that the founders did not focus on individual users - they are afraid that the startup will have nowhere to scale. But when the founders of start-ups in the early stages of going through this, I tell them that they have nothing to lose. Perhaps if they will try to make happy users already superschastlivymi, then at some point they are faced with the fact that they would not know what to suggest next. But God forbid anyone startup such problems. Let's see if you can get to this stage. And by the way, if that does happen, you will realize that the promotion of users scales better than you expected. Partly because you can always find ways to scale something more active than you think at first, and partly due to the fact that encouraging users to that moment will be part of the culture of your company.

I have never seen a startup that would come to a standstill due to the fact that too zealously trying to make users happy first.

But perhaps the most serious reason, which separates the founders of knowing what they should be considerate towards its users is that they have never felt this attention to themselves. Their standards of customer service laid the companies clients which they were themselves - and it is, as a rule, the big players. Tim Cook did not send you a handwritten letter, when you buy a laptop. He can not. And you can. This is one of the advantages to be small: you can provide a level of service that can not be a big company. [6]

Once you understand that the existing standards - this is not the upper limit of service users, you will be very pleased to think about how far you can go in awe of its users.


I have long pondered what the phrase is best to choose, to make sure the importance of attention to the users and realized that Steve Jobs has done it before me, "Insanely great." Steve used the word "crazy" not merely as a synonym for the word "very." He was referring to his literal meaning - you have to focus on the quality of performance at such a level that in normal conditions it would seem pathology.

Most of the most successful start-ups in which we have invested, and do, and it is perhaps not too surprising those who wish to become founders. But beginners do not understand how this is translated into the language of starting a startup. When Steve Jobs began to use the phrase, Apple was already held by the company. He meant that the Mac (and its documentation, and even the packaging - this is the nature of obsession) should be insanely good in terms of design and production. Engineers understand this is not so difficult. This is only a more extreme version of strong and elegant product.

What is difficult to understand the founders (Steve himself has spent a lot of time on this realization) - maintenance must be insanely good from the first day of the existence of a startup. No product should be insanely good, and people are feeling. Product - only one component. For a large company - always dominant. But you can and should provide users with incredibly good feeling, having an early, unfinished and bugs product if you can compete at the expense of care.

You can, but should you? Yes. Very strong bunch with early users - it is not just one of the tactics to accelerate growth in the early stages. For the most successful startups - is a necessary part of the loop receiving feedbacks, which improves the product. Creating a better mousetrap - it's not a one-time operation. Even if you start as many of the most successful start-ups, with the creation of something that you need, and you yourself, the first version of the product is never completely true. Except in areas where the cost of failure is too high, it is better not to aim at perfect performance. In the best of all at once to provide users with a product that already has at least the lowest utility, and then watch what they will do with it. Perfectionism often - justification for procrastination, and in any case, the original model is not perfect, even if you yourself are one of the users of your product. [7]

Fidbek that you get to communicate with your first users will be the best of all time project. When you become so large that they have to use focus groups, you longingly remember a time when you could just come home user and see how he uses your product, since these users was practically nothing.


Sometimes the best step would be unscaled deliberately focus on narrow market. It seems to contain flames that it is sufficiently inflamed by the time zakidyvaniya firewood.

That's what made Facebook. At first, it was just a student at Harvard. In this form of the potential market did not exceed several thousand people, but due to the fact that they feel that this is the product for them, a critical mass of people to register. Once Facebook is no longer just for Harvard students, he for some time was only for students of certain universities. When I interviewed Mark Zuckerberg at the Startup School, he said that since the establishment of the list of courses for each university takes a lot of time and work that students have done this, they began to perceive Facebook, as their natural home.

Any startup that can describe the phrase "Marketplace" is usually forced to start from some part of the market, but it also works well for other startups typecasting. It's always good to know there is a part of the market, where you can quickly gain a critical mass of users. [8]

Most startups that use the strategy of containment of flame, do it unconsciously. They create something for themselves and their friends, who thus become early users (early adopters), and only then realize that this product can offer a large market. Unconsciously or not, the strategy still works. The greatest danger of the conscious use of this template is for those who naively ignores its important component. For example, if you do not create something for themselves and their friends, or even if you are doing it, but come from the corporate segment, and your friends - not early adopters (early adopters), then you will not have perfect primary market submitted on a silver platter.

Most early adopters for companies - it is usually the other startups. They are more open to everything new, and by its nature and because they just run and have not yet taken all the decisions. Plus, if they are successful, they begin to grow rapidly, and you with them. What B2B start-ups now have instant access to a market of hundreds of other startups, it was one of the unexpected benefits of the model YC (especially in terms of what made such a big YC).


For "iron» (hardware) there is the option to create startups nonscalable things, which we call the "enable Meraki». Although we have not invested in Meraki, the founders were graduates of Robert Morris, so we know their history. They started with what really does not scale: autographic create your own routers.

"Iron" startups face the fact that is not in the software world. Minimum order for production at the factory is usually several hundred thousand dollars. So you can get trapped-22: without a product you will not be able to generate the growth needed to raise funding for the production of the product. Earlier, when the "iron" startups were forced to rely on investors' money, you had to be very persuasive to get around this problem. The arrival of crowdfunding (and especially pre-orders) really helped. But even in this case, we advise startups initially "include Meraki», if possible. That's what made Pebble. The first few hundred copies Pebble were collected manually. If they did not, they would not likely have collected $ 10 million on pre-order, when we went on KickStarter.

Pebble creator Eric Midzhikovski first prototypes Pebble

Making things with his own extremely valuable for "iron" start-ups, as well as special attention to the early users of the software producers.


See also

New and interesting