Short answer: "do things that don't scale" means doing slow, personal work by hand to win and delight your first customers, instead of waiting for marketing or automation to bring them. Paul Graham made the phrase famous in a 2013 essay. Recruit people one by one, help them personally, and learn from every conversation. Automate later, once you know what works.
Every owner dreams of a business that runs itself. Customers find you through search, buy without help and never need support. That can happen, eventually. But almost no business starts that way.
Paul Graham, a co-founder of the startup programme Y Combinator, wrote about this in his essay "Do Things that Don't Scale." He had watched hundreds of young companies. The ones that took off rarely waited for customers to arrive. Their founders went out and got them, one at a time, doing work that would never work at a large size.
This guide explains the idea and turns it into steps for a small business.
What does "do things that don't scale" mean?
It means doing things by hand at the start that you could never do for thousands of customers. Writing every first message yourself. Calling people back personally. Setting up a new customer's account for them. Sending a handwritten thank you note.
Graham's main point is direct: "You can't wait for users to come to you." Early growth almost never happens on its own. Founders who build something and then wait are usually disappointed.
The second point is about learning. When you do the work by hand, you see every problem up close. You hear the exact words customers use. You notice where they get stuck. That knowledge shapes everything you build later.
Why do early customers need to be recruited by hand?
When nobody knows you, there is nothing to find. Search engines do not rank a new site. Social posts reach few people. Ads cost money and teach you little until you know what to say.
Personal effort works because it does not depend on any of that. A personal message to the right person gets read. A conversation tells you more than a hundred page views. Our guide to finding your first 10 customers goes into the practical steps.
Graham also warns founders not to be put off by small numbers. Ten customers feels like nothing. But if you can keep adding customers steadily, small numbers grow. The first ten matter because they teach you how to get the next hundred.
What are famous examples of doing things that don't scale?
Graham's essay gives several examples from companies he worked with.
Stripe. The founders were known for what Y Combinator people called the "Collison installation." When someone agreed to try their payments product, the founders did not send a link for later. They asked for the person's laptop and set it up on the spot.
Airbnb. In the early days, the founders went door to door in New York, meeting hosts, helping them improve their listings and taking better photos. Graham describes this kind of effort as heroic, and as typical of what marketplaces need at the start.
Wufoo. The online form company sent each new user a handwritten thank you note, for as long as it could manage.
None of these founders thought the method would last. That was the point. They did it to get started and to learn.
How do you delight your first customers?
Graham argues that early customers should get an unusual amount of attention. Not because you will treat everyone that way forever, but because these first people decide whether you have a business.
For a small business, delight is usually simple:
- Reply fast. Same day, ideally within hours.
- Fix things personally. If something goes wrong, say sorry and make it right yourself.
- Check in. A week after someone starts, ask how it is going.
- Do a little extra. A useful tip, a small bonus, a personal note.
- Ask what would make it better. Then act on the answer and tell them you did.
These things cost time, not money. They also give you something no large competitor can easily copy: customers who feel known.
Should you treat early customers like consulting clients?
Graham suggests another approach: pick one customer and build exactly what they need, almost like a consultant. You learn deeply about one real case, instead of guessing about many.
This works well for service businesses and software alike. Find one customer with a clear problem. Solve it fully for them, by hand if needed. Then look at what you did and ask which parts would apply to the next ten similar customers.
The risk is that you end up building one-off custom work forever. To avoid that, choose customers who are typical of the group you want to serve, and write down every step so you can turn it into a repeatable offer later.
When should you stop doing things that don't scale?
Stop when doing it by hand becomes the bottleneck, and you know exactly what works.
Signs you are ready:
- You are answering the same questions every day.
- You know which first message gets the most replies.
- You have a clear, repeatable onboarding that works for most customers.
- You are turning down customers because you have no time.
At that point, write down the process, cut what you can, and automate the rest. Our guide on first principles thinking explains why to simplify before you automate. Our guide on what to automate first helps you pick.
What does this look like in a small business?
Doing things that don't scale looks different in every trade. A few examples:
A local bakery. The owner delivers the first office orders personally, chats with the person who ordered, and asks what they would like next time. Those chats can turn into a regular weekly order.
A consultant or freelancer. Instead of waiting for referrals, they write a short, personal note to ten past colleagues each week, offering one specific kind of help.
An online shop. The owner writes a short thank you note to every buyer in the first months, and emails each one a week later to ask how the product is working. The replies show which products to stock more of.
A software product. The founder sets up each new customer on a video call, watches where they get confused, and fixes those spots before the next call.
In each case the owner is doing something slow and personal on purpose. The aim is to win customers who would otherwise never notice you, and to learn things you could not learn any other way.
How to use this in your business this week
- Make a list of 20 people. Each should fit your customer closely. Use contacts, groups you belong to and local businesses.
- Write to five a day. Write each message yourself. Start with why you are writing to that person. Ask about their problem, not your product.
- Give one customer the full treatment. Pick your best current customer or prospect. Set them up personally, check in after a few days, and ask what would make it better.
- Keep a notes file. After each conversation, write down the exact words they used about their problem. These become your marketing copy later.
- Send one thank you note. A real one, written for that person.
If you use Extra Arm AI, your team can do the slow typing while you keep the personal touch. Ask your chief of staff: "Find 20 people or businesses who fit our customer, with links, and draft a personal first message for each." Every message waits as a draft for you to read, edit and approve. You stay the person they talk to. You can start free and try it with your own list.
Questions people ask
Who said "do things that don't scale"?
Paul Graham, co-founder of Y Combinator, in a 2013 essay of that name. It has become one of the most widely shared pieces of startup advice.
What does "doesn't scale" mean in business?
It means a method that works for a few customers but would be too slow or costly for thousands, like writing every message by hand or setting up each account personally.
Is doing things that don't scale a waste of time?
Not at the start. It is often the fastest way to get your first customers and learn what they need. It becomes a waste only when you keep doing it after you know what works.
How many customers should I win by hand?
There is no fixed number, but many owners aim for the first 10 to 20. By then you usually know what to say, who to say it to and what to automate.
Sources
- Do Things that Don't Scale, Paul Graham
- How to Get Startup Ideas, Paul Graham
- Y Combinator's essential startup advice, Y Combinator