Our no’s, annotated

We sent 212 rejections last year. Each one came with a paragraph, usually two, and a name at the bottom. In July we wrote to every founder who had received one and asked a short question. Which of our reasons was wrong?

Seventy-four answered. Most said we were right, or right enough to be useful. Eleven said we were wrong. We have read all eleven carefully, and we think they are correct.

What we got wrong

Five times, we said the market was too small. We were reading the slide. The spreadsheet, had we asked for it, would have shown a customer list that doubled the number.

Four times, we said the founder lacked industry experience. In each case the founder had spent years on a shop floor, doing a job with a different title.

Two times, we said the sales cycle was too long to fund. Both companies funded it, from other money, and are doing well.

A no is a paragraph you can actually use. Sometimes the use is proving us wrong.

The pattern in the eleven

None of the eleven were about the company. They were about our reading of it, done quickly, from a distance, with a deck. The founders who were right had, almost without exception, told us the true thing in the first call, and we had not heard it.

So we have changed what we write down. During a first call, one of us now keeps a list of anything the founder says that surprised us, and we reread the list before deciding. In seven of the last twenty calls, the list changed the answer.

What we still would not take back

One founder asked for $6M on a plan that needed three new factories. We said no in eight days. He wrote back to say we were wrong. We read his email twice, then wrote another paragraph, longer. He has since built one factory, and the paragraph is still correct.

We keep that paragraph in a folder, next to the eleven. It is the only evidence we have that we can be consistent.

What changed

We now ask for the spreadsheet before the deck. We ask what the founder’s last job was called, and then what they did. We read the first customer reference before the plan. And we send the no with a line that says: if we are wrong, tell us, and we will say so in the next letter.

One founder rebuilt her plan around our two pages, and we backed her eleven months later. This is the outcome we would like more of. It is also the reason we write them at length, and why we mean it when we say that a reply is welcome. Most of the useful ones have been short.