CONCEPT
How new companies actually acquire their first customers.
New companies do not get their first customers from a channel. They get them one at a time, manually, from people who already have a reason to take the call, on the strength of a specific claim about an outcome. A channel is a machine for repeating a sale you already know how to make. Building the machine before you know the sale is the most common and most expensive early mistake, because it produces activity that looks exactly like progress.
Two variables do nearly all the work early.
Warm surface area is the number of people who will respond to you at all: former colleagues, former customers, people who watched you do the work, and the people one hop from them. It is a stock, and it is bigger than most founders think and smaller than they need.
The claim is what you say you will produce, stated as an outcome and specific enough to be wrong. "We help companies grow" gets polite replies. "We get your quote approvals to same-day" gets either a yes or a fast no. Both are useful. Polite is not — a warm reply with no next step has consumed the contact and taught you nothing.
Paul Graham's "do things that don't scale" is the accurate description of this phase. The manual work is not a stopgap until marketing arrives. It is how you learn what the sale actually is: what event triggers the purchase, who blocks it internally, and which sentence survives being repeated to a CFO by someone who does not work for you.
Early growth almost always gets attributed to referral, and early referral is usually just warm network depleting. It behaves like a flywheel for about a year and then stops, at which point the company discovers it never had an acquisition mechanism.
The distinction is testable: are new customers producing the next ones, or are you drawing down a fixed list? You can only answer it by recording the source of every customer from the very first sale. It feels premature at five customers. It is the only way to know, at fifty, whether you have a business or a rolodex.
The first error is reaching for marketing when there are no customers. Marketing amplifies a sale that already works. Pointed at a sale that does not work, it produces traffic, dashboards, and no revenue, and it takes about two quarters to be sure.
The second is optimising the pitch. Buyers do not buy the mechanism, they buy the state they end up in, and rewriting the mechanism in better language does not move a deal that has no internal owner, no budget line, and no trigger event. Find the reason this quarter is the quarter they act. That is the repeatable part. Build the channel around it, and not before.