How to get your first 10 customers without a budget

By Meet Patel · 2026-10-03 · 6 min read

Paul Graham's 2013 essay Do Things that Don't Scale opens with a claim that most founders find uncomfortable: the way to get the first users of a startup is to go and get them one at a time, by hand. His examples are specific. Stripe's founders would set a new user up on the spot instead of sending a link, in what became known as the Collison installation. Airbnb's founders went door to door in New York, recruiting hosts and helping existing ones improve their listings. Wufoo sent each new user a hand-written thank-you note.

None of these is a marketing technique with a budget. Each one is a founder spending hours on a few people, and each produced customers who stayed and told others. This post turns that pattern into a sequence you can run in the next ten weeks, with no money, to reach your first ten paying customers.

Step 1: Choose a segment small enough to name

Graham's Facebook example makes the point. The site started at Harvard, and because they felt it was really for them, a critical mass of them signed up. He adds that any startup that could be described as a marketplace usually has to start in a subset of the market.

Write one sentence that finishes "My first ten customers are ___", and make it narrow enough that you could list 50 of them by name. "Small businesses" fails the test. "Independent physiotherapy clinics in Dubai with two to five practitioners" passes it, because you can search for them, count them and find where they gather. For more on choosing where to start, see Day-One Density.

Step 2: Build a named list of 50

Open a spreadsheet and write 50 real people or companies, with a name, a contact route and one line on why they would feel the problem. Sources that cost nothing: professional communities, the segment's own directories, LinkedIn searches, trade association member lists, and the people you already know who can introduce you.

A list of 50 gives you a base for arithmetic. Say, as an assumption to be replaced by your own data, that 1 in 5 conversations turns into a paying customer. You need 50 conversations to reach 10 customers, and at 5 conversations a week that takes ten weeks. If your first 10 conversations produce no customer, the number is telling you something about the offer or the segment, and you have learned it in two weeks instead of two quarters.

Step 3: Recruit each one by hand

Graham's observation is that founders resist this step, out of shyness or laziness and because they underestimate what compounding does with small numbers. His own illustration uses a weekly growth rate: if you have 100 users, you need to get 10 more next week to grow 10% a week. At that rate, he writes, you would have about 14,000 users after a year.

Recruiting by hand means a short, specific message to a named person, followed by a conversation. Mention the segment, the problem in the buyer's words and one concrete thing you will do for them. Keep it under 100 words, send five a day, and log who replied. Here is an illustrative version for the physiotherapy example: "Hi Dr Rao, I am researching how small physiotherapy clinics handle last-minute cancellations. Could I ask you three questions on a 15-minute call this week? If the answers are useful, I will show you what I am building and set it up for your clinic myself." The goal is a conversation, and the pitch comes later.

Step 4: Do the work for them

The first customers should get an unscalable level of service, which is the main thing you can offer that a larger competitor cannot. Graham describes treating early users like consulting clients: pick a single user and act as if they were consultants building something just for that one user. He also describes the manual variant, where Stripe handled early merchant-account setup behind the scenes so that users saw a working product before the automation existed.

For a small team this translates into three practices:

• Onboard in person or on a call. Sit with the customer, set the product up for them, and watch where they hesitate. Budget two hours per customer, which is 20 hours for ten, a small cost against what you will learn.
• Perform the unbuilt part manually. If the product will eventually automate a report, a reconciliation or an approval, do it yourself for the first customers. You learn the real workflow before you automate a guess about it.
• Delight on purpose. Graham writes that he has never seen a startup lured down a blind alley by trying too hard to make its initial users happy. A hand-written note, a same-day fix and a direct phone number cost almost nothing at this scale.

Step 5: Ask for a commitment in every conversation

A pleasant conversation is easy to mistake for demand. The Mom Test, Rob Fitzpatrick's book on customer conversations, offers a cleaner measure: a meeting should end with a commitment, where the other person gives up something of value. The forms it describes are time (a scheduled follow-up or a trial), reputation (an introduction to a colleague) and money (a deposit, a pre-order or a letter of intent). The more they give up, the more seriously you can take what they said, according to one detailed reader's summary of the book.

For your first ten, the commitment you want is money. Offer a paid pilot with a fixed price and a short term, such as a month. A free pilot tells you whether someone will accept a gift, and a paid one tells you whether the problem is worth money to them. Questions that help you reach that point are in customer discovery interview questions.

Step 6: Track it weekly and read the pattern

Keep one table with a row per week and four columns: conversations held, commitments received, paying customers added, and the channel each customer came from. After four weeks, look for the channel with the highest ratio of customers to effort, and put the next week's hours there.

Graham's weekly framing is the useful part. A founder who measures growth weekly sees a flat week within seven days, and a founder who measures it quarterly sees it after the quarter is gone. Ten customers in ten weeks is a pace of one a week, which is a rate you can change by deciding how many conversations to hold.

Four mistakes that stall the first ten

• Waiting for inbound. A landing page and a launch post can produce a handful of visitors. Graham's whole argument is that you cannot wait for users to come to you, so schedule outreach hours the way you schedule product hours.
• Counting free users as customers. Free users show interest and say little about willingness to pay. Keep two columns in your table, and report only the paid one as customers.
• Widening the segment after a slow week. Two bad weeks with 15 conversations is a signal about the offer. Adding a new segment resets the learning to zero, so change the offer first.
• Automating after three customers. Three customers give you three versions of the workflow. Wait until the same manual steps repeat across five or six before you write code for them.

What the sequence is for

Ten customers give you ten sets of specifics: who buys, who signs, what they were using before, what they said at the moment they paid. That falls short of proving a business, and it still decides what you build next, because the information is hard to get any other way. The earliest stage rewards the activity that cannot be delegated or automated. Related reading on why starting is hard: The Zero-to-One Problem.

My rule is to keep recruiting by hand until the work itself produces a pattern you could write down as a process. Automating before that point turns a guess into software.

Written by Meet Patel — startup operator and growth strategist in Dubai.

Read on themeetpatel.com