Let's find your future customers, write each one a letter in your voice, and put a stamp on it.

I read your website to work out who buys from you, then watch for the ones giving off a buying signal: a new hire, fresh funding, a competitor’s tool on their stack. Those work, and everyone can buy them, so I do not stop there. I also read the public record for the moment before: a role posted a second time, because the first attempt failed, or a company registered last week giving its lawyer’s address, because it has nowhere to sit yet. You see everyone I found and can strike anyone off, or hand me your own list. Then I find the right address and build that company its own page, showing how one of your case studies applies to them. What lands on their desk is a heavy letter about their company, and a QR code to that page.

We write the first letter together: I draft, you nudge it with a line or two of feedback, and I learn from that. You never have to write anything yourself. Then you approve each batch before it goes out. You will adjust a few the first month and almost none by the third, which is how you know it is working.

The list, the letter, and who you will not write to. Strike anyone off the list, or hand me your own. Rewrite any letter, and the edit teaches me rather than applying once. Name a company you will never contact and it stays off. Change the angle whenever you like, though I will tell you what the last one was returning before you do. What you cannot change is a letter after it is posted, which is why nothing goes out without your yes.

Twenty minutes a week, and the first week is the exception. We write the first letter together: I draft, you tell me what is wrong with it, and I learn from that. After that your week is approving a batch and reading one page on what came back. What you never do is research a company, find an address, write a letter, or remember to follow up on a Tuesday.

Printed. I'll never tell you otherwise. But printed right: heavy stock, an algorithm that varies every stroke, so no two pages match. I ran it against the machines that hold a real pen, and people picked mine as the more human one. The ink was never the point. The point is a letter written about their company, for one person, that lands on their desk. And when a prospect asks, you have a straight answer: the handwriting is machine-made, the decision to write to them wasn’t. We don’t print that on the envelope, because the handwriting has one job, to get the letter opened. After that, you can say exactly how it came about.

Yes, and on firm legal ground on both sides of the Atlantic. In the EU two articles carry it. Article 6(1)(f) is the basis for writing at all: business outreach counts as a legitimate interest, which is why the rules that strangled cold email left the letterbox alone. Article 14 is the duty that rides along with it: because you found them rather than the other way round, the first contact has to say where their details came from and how to stop it. That line is on every letter, and the page behind the QR code takes them off the list in one tap. In the US, a letter to a business office carries no such restriction, and cold email is legal under CAN-SPAM as long as it names you and offers an easy way out. LinkedIn messages stay within the platform's own terms of use. Across every channel: business addresses only, opt-outs honoured, no consumer lists.

$490 a month, plus tax. Your weekly letters, the research, the follow-ups, the weekly read on what came back. Month to month, no setup fee, no minimum. The useful way to size it is what a first real answer costs: 2 months, 160 letters, and you know whether this channel is worth continuing. Whether it pays comes down to $490 against what a customer is worth to you. Run your own numbers in the calculator below.

Two questions, and they resolve at different speeds. Whether it works: at 20 letters a week, a reply rate starts to mean something once about 48 have gone out and had a fortnight to be answered, so about a month in. Somewhere between week 6 and week 8 that becomes a real read. By the end of it 160 letters are out, 120 have had their fortnight, and you have paid for 2 months. Whether it does not work is dearer, because silence is weaker evidence than a reply: ruling an angle out honestly takes 153 answered letters, about 10 weeks. The 3 follow-ups are what buy that number down, because silence from someone we asked again and again is worth far more than silence from someone asked once. Every week I tell you where on that road you are, so a quiet month reads as a position and not a verdict.

Yes for half of it, and the half matters. Proving an angle works takes 48 letters. Ruling one out honestly takes 153, and a milder shortfall takes 361. Good news is cheap and bad news is expensive, and anyone quoting you one number for both is selling something. Two things pay the difference down. Each person is asked up to four times, not once, which lifts the reply rate rather than the count, and when the event is rare that does the same work: someone who opens their page, takes two more follow-ups and still says nothing has answered you as clearly as an email would. And you are not running a poll. A poll pins a percentage to the decimal; you need one decision, whether this buyer and this message clear the floor, and that costs far less evidence. When the honest answer is still “too early”, that is what you get, rather than a small number dressed up as a finding.

By counting, with the error rate stated up front. Detection needs a roughly fixed number of expected replies, not a fixed number of prospects, so the prospects you need is that count divided by your reply rate. That is what the follow-ups are for: they do not give us more trials, they raise the rate, which does the same work. Concretely: a standout angle proves itself inside 48 prospects, but at 48 there is no honest way to call anything dead yet, at any threshold. A collapse to a quarter of what a working campaign returns is caught 60% of the time by 110 prospects and 87% by 153. A milder shortfall, half of what a working campaign returns, takes 361. What we cannot do is rank one angle against another at this volume, and we do not pretend to. We accept about a 5% chance of calling a working angle dead to get those catch rates. The ladder, the worked example and the interval we use

You cancel with a week's notice and keep everything: the target list, the letters, and the record of which message got answered and which got silence. That travels into every channel you run next.