Notifications disappear. A letter stays.

An envelope is judged in about two seconds. In the hand, before the eye has done anything useful. By the time the reader has focused on the first line, they have already decided whether this is a document or a piece of post to be dealt with later, which usually means never. Almost everything below is spent on those two seconds. The letter itself only has to survive them.

In the hand

Weight is the first signal and it arrives before sight. Ordinary office stock is 80gsm and feels like a form. At 120gsm and above the same envelope reads as deliberate. Somebody chose this, and choosing costs money. Nobody consciously notices the difference. Everybody acts on it. It is the cheapest change on this page and one of the two or three that matter most.

Then the fingertips. Cotton stock, laid paper with its ribbed grain, sheets with visible linen inclusion: all of these register as personal. A smooth coated surface registers as advertising, because advertising is what it is usually used for. The test is blind: hand somebody two envelopes without letting them look, and they will tell you which one is important.

On the outside

The envelope carries no transparent panel. An envelope with a panel is a bill, and it goes wherever bills go, which is not where letters go. That sorting happens before anyone has decided anything, and no amount of good writing inside recovers from it.

Cream, grey-blue, deep green, and never the standard size. A square or an elongated format refuses to stack, so it sticks out of whatever it is put on top of. It also fails automated sorting, so a posted piece is handled by people the whole way. Slower, more expensive, and better. A couriered piece is handled by one person the whole way, which is the same argument taken to its end.

On anything that goes by post, a real stamp, stuck on by hand, ideally an unusual one. Metered franking says a system sent this. A stamp says somebody stood at a counter. The difference is visible before the envelope is opened and costs almost nothing to make. A couriered envelope carries no stamp at all, and no stamp is its own signal: somebody paid a person to walk this in.

On short runs the address is written by hand. This is the strongest signal available on the outside of an envelope and the one that scales worst: four hundred addresses can be written, forty thousand cannot. Where a campaign is small enough, I write them. Where it is not, I say so rather than printing a font that imitates handwriting, which fools nobody and insults the reader who notices.

The seal and the impression

Where the piece warrants it, the envelope closes with wax. A seal is the only closure that cannot be opened and shut again without saying so, which tells the recipient two things before they read a word: the letter arrived as it left me, and nobody read it first. It also has to be broken with a thumb, which is a physical act performed before any reading has happened.

The letter is printed letterpress where the run allows it. Letterpress presses each character into the paper, leaving an impression you can read with a fingertip and see against the light. Digital printing cannot reproduce it. Neither can a competitor working at volume, because the method stops making economic sense somewhere above three thousand pieces. That is the same threshold where this whole channel stops making sense.

Blind debossing goes further: a mark pressed into the sheet with no ink at all. It is read by touch before it is seen. One word, or one device, and nothing more. A page of it becomes decoration, and decoration is what I am trying to avoid.

Opening it

There is one physical action between the envelope and the letter. A sheet folded in three and tied with thread. A tissue leaf laid over the page. A folder that has to be opened rather than a page that falls out. Each movement of the hand is another second in which the reader is already involved, and involvement before reading is worth more than any opening sentence.

Smell works on memory longer than anything else in this list, and it is also the easiest thing here to get badly wrong. Perfume in a business envelope reads as manipulation and is resented. What works is the smell a material already has: cedar shavings, leather, sealing wax, ink still fresh on a letterpress sheet, cotton paper on its own. The scent should be a consequence of what the piece is made from, never an addition to it.

Something in the envelope is left unfinished. A line with a figure for the reader to fill in. A tab to be torn. A card to be sent back. An unfinished action stays in the mind longer than a completed one, and a reader who has picked up a pen is no longer reading an advertisement.

What is still on the desk on Thursday

Paper goes in the bin within the week. A key, a coin, a broken bracket does not. It moves to the edge of the desk, then to a drawer, and every glance at it is another contact I did not pay for. This is the real argument for putting an object in the envelope, and it has nothing to do with the object being clever.

What lasts longer than the object is the thing the reader did: broke a seal, untied a thread, wrote a figure, posted a card back. An action performed is remembered far better than a page read.

And what lasts longest is the story. A letter that arrives in a way somebody describes to a colleague over lunch has done something no amount of paper stock achieves on its own. That is the only part of this I cannot manufacture, and it is the part I am actually aiming at.

One desk, one working day: 117 emails and 153 Teams messages, and most of the emails are skimmed in under a minute. Microsoft measured that across its own Microsoft 365 telemetry and a survey of 31,000 workers in 31 markets, and published it in June 2025.

Post went the other way. American mail volume peaked at 213 billion pieces in 2006. In the year to September 2025 it was 109 billion, a fall of almost half, and it is still falling.

Producing a cold email became free, and landing one became a trade: warmed domains, rotated addresses, sentences written to slip past a filter rather than to convince a person. That trade is real, and everybody selling to your buyer has already bought it. It is the price of entry now, and entry is where it stops.

A physical letter cannot behave this way, and that is the whole argument. Nobody receives two hundred envelopes in a day, because nobody can afford to send them. Paper costs money. The object inside costs money. Packing, postage and a delivery record cost money. Every one of those costs is a barrier, and the barrier is what keeps the channel worth using. Email lost its barrier and lost its value at the same time.

The expense also changes what happens before the send. An email campaign has no reason to verify an address, because a wrong one costs nothing and a bounce is not an event. I check because an unverified record costs me paper, an object, postage and a delivery attempt. I would rather burn a great deal of machine time confirming a record than post one envelope to a company that dissolved last year. That is not diligence for its own sake. It is arithmetic.

I collect email addresses in the course of the work. I do not sell them and I do not hand them over, not as a bonus and not on request. What I sell is a street address that a physical letter can reach. If you want an email list there are companies that will sell you one, and the fact that they will is most of the problem.