Narrow your target until you know where these people gather and what they call the problem. Then work three channels in descending order of trust — your personal network, public communities where the problem gets discussed, then direct outreach using the words the first two taught you. Ask for a small, real commitment early: interest costs nothing and proves nothing. Run it two hours a week, log every conversation, and go deeper on whichever source produced the people who paid.
Most advice on how to find customers is a list of channels: SEO, ads, content, cold email, partnerships, communities. The list is not wrong, it is just useless before you have customers, because you cannot evaluate a channel until you know who you are looking for and what a real buyer sounds like.
The first ten customers are found by hand, one at a time, from three sources: people who already know you, people talking in public about the problem, and people you contact directly. Everything else — the ads, the funnel, the content engine — is a way to do more of whatever turned out to work here. This guide is the order to work in.
Decide who you are looking for
“Anyone who needs this” is not a target, it is a reason to never start. If your customer could be any of a million people, no message fits, no community is the right one, and every rejection is ambiguous: you never learn whether the offer was wrong or the person was.
Narrow until the group is findable. A useful description tells you where these people gather, what they call the problem, and how you would recognise one on a Tuesday afternoon. “Operations managers at three-to-ten site dental practices who still schedule staff in a spreadsheet” is findable. “Small businesses” is not.
- Vague
- “Freelancers who want to save time on admin.” Where do you look? What do you say? Every answer is a guess.
- Findable
- “Freelance video editors billing more than five clients a month, who post in editing subreddits about chasing late invoices.” The description names the place, the trigger, and the words.
Write it down before you contact anyone, and treat it as a hypothesis rather than a decision. The first twenty conversations will move it — usually narrower, and usually toward a smaller problem than the one you set out to solve. That is the process working. Our guide on how to define your ideal customer goes through this in detail, including the intent signals that separate someone who has the problem from someone who is doing something about it today.
Work three channels, in this order
The three sources of early customers differ in one respect that matters more than reach: how much trust you start with. Work them in descending order of trust, because each one teaches you what to say in the next.
People who already trust you. Highest reply rate, smallest pool. Start here on day one.
Strangers describing your problem in their own words, unprompted. Slower, but it hands you the language.
Cold, and the only one that scales past the first two. It works once you know what to say — which the first two taught you.
Your personal network is the fastest route to a first conversation and the one most founders skip, because asking people you know feels like an imposition. Do it anyway, and ask for the wrong thing on purpose: not “would you buy this”, which invites a polite yes, but “how do you handle this today”, which produces facts. Former colleagues, clients, and second-degree LinkedIn connections are the pool; a specific, forwardable introduction request is the ask. See how to find customers through your personal network.
Online communities are where people describe the problem before they go looking for a solution. Reddit threads, Discord servers, Slack groups, and industry forums are full of recommendation requests, complaints about a competitor, and “we still do this manually” posts — each one a person who has already admitted the problem in public. Search these places rather than scrolling them, answer completely before mentioning what you sell, and disclose that it is yours. See how to find customers in online communities.
Manual outreach is the one channel that works when you have exhausted the other two. Fifty short, researched messages sent by hand beat five thousand automated ones, because the recipient can tell within a sentence whether a person looked at their situation. Keep it under seventy-five words, lead with a specific observation about them, and ask a question they can answer in one line. See how to do manual outreach that gets replies.
Do not run all three at once in week one. Run the network for a fortnight, then add communities, then start outreach with the phrases the first two gave you. Sequencing is what turns three mediocre channels into one that works.
Get paid customers, not interested ones
Interest is free, and people give it away to be kind. “This looks great, send me a link” costs the person nothing and tells you nothing. A pipeline full of encouragement is the most common way to spend six months finding no customers at all while feeling like progress is being made.
The fix is to ask for a commitment early and to make it small enough that a yes is plausible and a no is informative. Money is the clearest commitment, but not the only one: a scheduled call, a data export, an introduction to their manager, or an hour of their team’s time all cost something real. Anything that costs nothing proves nothing.
Before quoting a price, establish four things in their words:
- What they do today. The current workaround is your real competitor, and it is usually a spreadsheet and a person’s patience.
- What it costs them. Hours per week, errors per month, customers lost. This is your price justification, and they have to say it, not you.
- Why now. A new hire, a lost client, a system being retired, a rule changing. No trigger means no urgency, and no urgency means a deal that stalls forever.
- Who signs. If the person cannot approve the spend, you are not in a sale yet — you are being helped by someone who will need help selling it internally.
Then ask plainly, with a number and a date. A small paid pilot with a defined scope and a decision point beats an open-ended free trial, because a free trial with no end date is how a promising conversation quietly dies. If the price is wrong, they will tell you, and that is a better outcome than a polite drift into silence.
Expect to be told no, and treat each no as output rather than failure. Ask what would have made it an immediate yes, and what they are doing instead. Five specific noes will reshape your offer faster than fifty vague expressions of interest.
Run it weekly and read the evidence
Finding customers fails as a project and works as a habit. The founders who get stuck do it in bursts: a week of outreach, a bad response rate, three months of building, another burst. Nothing accumulates, and every attempt starts from zero.
Book two hours a week and give the block a definition of done: ten new people contacted, three conversations held, one thing changed as a result. Ten a week is a hundred and twenty in a quarter — enough to see a pattern, which one burst of forty never gives you.
Keep one row per person somewhere you will actually reopen:
Who: [name, role, company]
Source: [network, community, or direct outreach]
Signal: [what suggested they have the problem now]
Said: [their exact words about the problem]
Asked for: [the commitment you requested]
Result: [paid, call booked, no and why, or check back on a date]
Read it monthly and ask two questions. Which source produced the conversations that went somewhere? And which phrase keeps appearing in the “said” column? The first tells you where to spend the next quarter; the second is your landing page copy, written by your customers, which is the only reliable way to get it right.
When one channel starts producing repeatably, stop spreading effort and go deeper on that one. Scaling comes later, and it is far easier once you can point at ten people who paid and say exactly where each of them came from.