How to get your first 10 customers

Most founders get this wrong in the same way: build, launch on a directory or two, post on X, and wait. Weeks pass. The product is good and nothing happens — because nobody is searching for something they don't know exists. Your first 10 customers will not find you. You will find them, one at a time, by hand. Here's the plan.

Why waiting doesn't work

Early on you have no SEO, no brand searches, no word of mouth — the three channels that bring customers to you all require customers you don't have yet. That's the cold-start loop, and the only reliable way out is outbound: going directly to people who have the problem and telling them your product exists.

Every durable company you admire started this way. Stripe's founders installed the API on developers' laptops mid-conversation. Airbnb went door to door in New York. The pattern is so universal Paul Graham wrote the canonical essay on it — we unpack the outreach half of it here.

Week 1: define the customer narrowly enough to find them

"Startups" is not an ICP. "Seed-stage B2B founders who just hired their first salesperson" is — you can build a list of those. The test of a good ICP definition: could a stranger use it to decide, in ten seconds, whether a given LinkedIn profile is in or out? If not, narrow it.

• Pick one segment where the pain is sharpest — not the biggest market, the most desperate one.
• Write the pain in their words, not yours. You'll reuse this sentence in every email.
• List 100 real people. Not companies — people with names and inboxes. (Our guide to finding B2B leads without buying a list covers the where; finding their email covers the how.)

Weeks 2–3: reach out like a person, not a campaign

At this stage you are not "doing cold email" — you're starting 100 individual conversations. The rules:

• One specific, true sentence about them first. Their launch, their hiring post, their podcast quote. Ten seconds of research beats any template.
• One sentence on the problem (their words from week 1), one on what you built.
• Ask a question, not a meeting. "Is this how you handle X today?" outperforms "got 15 minutes?" — the meeting comes after the reply.
• Follow up. Most replies come from touches two through four, not the first email. Silence is not a no — it's a busy inbox. (The follow-up sequences that work.)

Send from your real address, at human volume — there are hard limits on how many you can safely send per day, and at 100 prospects you don't need to go anywhere near them.

The math that keeps you sane

If 100 sends produce zero replies, the problem is almost never volume — it's the list or the first line. Fix targeting before sending more.

Talk to every single one

The first 10 customers aren't a revenue milestone, they're a research base. Get on calls. Ask what they tried before you, what almost stopped them from replying, what words they'd use to describe the product. Those words become your landing page, your next 100 emails, and eventually the ads. Then ask each happy one: "who else do you know with this problem?" — referrals from customer #3 are how you get customers #11–30.

Where tools fit (and where they don't)

The judgment — who to target, what's true about them, how to describe the problem — stays yours. The labor — building the list, verifying emails, sending each personalized message, remembering the day-4 follow-up, watching for replies — is exactly what software should absorb. That division is the entire design of persist: you describe the customer in a sentence, review what the agent drafts, and spend your time on the replies instead of the spreadsheet.