Is one paying customer enough to build on?

No. One yes is one person's problem. A product needs the same problem in 3 people who don't know each other, and until it has that, it's a favour you've built for a friend, however well it works.

The 4-minute phone call

The first time I asked anybody for money for something that didn't exist yet, I was sick with nerves for 2 days, and the conversation took 4 minutes. He said yes. He was a bookkeeper who did the accounts for about 12 small businesses and lost every Friday afternoon moving transactions from one system into another by hand.

I said the problem back to him in his own words, named a price and a date a fortnight away. Then he asked what would happen if I was late. I hadn't thought about it, and said on the spot that he'd get every penny back. I've made that promise on every pre-sale since.

I built it in 11 days. He sent me a photograph of his empty Friday afternoon: a cup of tea on a garden table. I still have it.

Then I launched it to everybody else

I posted in 2 places where bookkeepers talk. 3 people looked. None paid. I spent a month improving it, because "it isn't good enough yet" was the most comfortable explanation available.

My kill number was 1 paying customer by day 30. It met it. By my own rule the product was alive, so I didn't look hard at the fact that the single customer was the man I'd built it for. For 5 months it paid for itself, exactly. Then he joined a firm with its own systems, wrote a kind email thanking me for the Fridays, and that was that.

Count your yeses properly

His problem was real, but it was his: his pair of systems, his clients. Count your yeses. If they all know each other, or all know you, you have one yes, however many there are.

My rule now is 3 people out of 10 conversations, who don't know each other, paying before I build. And a kill number high enough that a favour can't meet it.

Nobody Bought It has the autopsy. The rules page has the rule.