Chud News

No. 13 — The send button is a design problem

Bulkhead — Thu, 17 Sep 2026 00:00:00 GMT

Everyone who has run a newsletter for more than a year has a send-button story. Mine is a merge tag that went out as {{first_name}} to eleven hundred people, which is fine, and a second one where I sent an issue twice in four minutes, which was worse.

I used to think these were attention failures. I now think they are interface failures, and that the interface is shaping more than just the mistakes.

What the button actually is

Consider what we are asking of this control. It performs an action that cannot be undone, at a scale where every error is multiplied by the list size, on a piece of work the author is emotionally invested in, usually late, usually tired, after a session of small anxious edits.

Almost every other irreversible action in software has a safety net. You can unsend a Slack message, recall a Gmail for thirty seconds, roll back a deploy. The one action where the stakes are highest and the author is most tired ships with a modal that says “Are you sure?” — which is not a safety net, it is a formality everyone clicks through.

The habits that work

I stopped trying to be more careful, because being careful is not a plan. These are mechanical.

Send yourself the real thing. Not a preview render — an actual send, to a real consumer address, with a real merge record attached. Preview modes fill merge tags with sample data and hide exactly the class of bug that embarrasses you.

Read it on a phone, in the morning. Not on the desktop where you wrote it, and not at eleven at night. Roughly two thirds of your readers will see it on a phone and the errors that survive proofreading are almost always errors of scale.

Schedule, never send. Schedule for ninety minutes out. The button becomes reversible for ninety minutes, which is the entire safety net, and it costs nothing. Every provider supports this and almost nobody uses it for anything other than time zones.

Run the link check last. After every edit, click every link in the sent test. Links break during the final pass more than during writing, because the final pass is when you move paragraphs.

If you are about to send while feeling any emotion at all about the issue, schedule it instead and look again in the morning. Feelings at 23:40 are not editorial judgement.

The part that is not about mistakes

Here is the thing I did not expect. Once I started scheduling instead of sending, I started writing differently.

When the button is irreversible, the last hour of work goes into anxiety — re-reading for errors, checking the subject line for the fifth time, hovering. When it is reversible for ninety minutes, that hour goes into the ending instead. I cut more. I strengthened more closings. The issues got better in a way I can point at, and the only thing that changed was that I had made a decision reversible.

Interfaces shape work. This one shapes yours twice a month.

That is issue thirteen. Next time we start on subject lines, which deserve a lot less mysticism than they get.

No. 12 — Do the paid-tier arithmetic first

Epic Games — Thu, 10 Sep 2026 00:00:00 GMT

Every few months someone I know announces a paid tier. Most of them have not done the multiplication, and the ones who have usually did it with the wrong number.

So here is the arithmetic, plainly, before the fun part.

The conversion rate is 2 to 4 percent

Across the newsletters I have seen real numbers for, free-to-paid conversion sits between two and four percent. Four is good. Above five means either a very narrow professional niche or a list built almost entirely from people who already pay you for something else.

That is the number to plan with. Not the one from the case study written by the company selling you the subscription tooling, which is invariably about a newsletter with forty thousand subscribers and a founder who was already famous.

At the usual rates, on a list of 5,000:

Conversion	Paying	At €5/mo	At €8/mo
2%	100	€500	€800
3%	150	€750	€1,200
4%	200	€1,000	€1,600

Before fees, before tax, before churn.

Now subtract

Payment fees. Roughly 3 to 5 percent all in, more if you are taking small amounts across borders.

Churn. Monthly subscriptions to a newsletter churn at something like 4 to 8 percent a month. That is not a disaster — it is normal — but it means a third of this year’s subscribers will not be here next year, and you are running to stand still. Annual plans cut it sharply, which is why every provider pushes them.

Your time. This is the cost nobody puts in the sheet. A paid tier is support requests, failed card emails, refund decisions, VAT, and the permanent low-grade obligation to deliver something extra. Call it four hours a month at the very least.

On a 5,000-person list converting at 3 percent at €5, you are looking at roughly €700 a month after fees, before tax, for a recurring obligation that never stops. That may be excellent — it is a mortgage payment for a lot of people — or it may be the worst-paid job you have ever taken. The point is to know which before you start.

The question is not whether people will pay. Some always will. The question is whether the amount is worth what you have to do forever to keep earning it.

The three shapes that work

Same issue, early, plus thanks. The lowest-obligation model there is, and the one Marco Peña argued for in issue 03. You are selling continuation, not access. Converts lower, costs almost nothing to run.

One extra thing a month. A deep-dive, an interview, a working file. Converts better. It is also a second deadline, which is the thing most people underestimate — you have effectively doubled your publishing schedule to serve three percent of your readers.

A one-off, not a subscription. A guide, a course, an annual archive. No churn, no recurring obligation, and it can earn more in a week than a tier does in a year. Badly suited to steady income, very well suited to people who do not want a second job.

What I would do at each size

Under 1,000 subscribers. Nothing. Two percent of 800 is sixteen people. Write more issues.
1,000 to 5,000. A tip jar or a one-off. You want the money without the obligation at this stage, because the list is still the thing that needs your attention.
5,000 and up. Now a tier makes sense, and I would start with the cheapest version — early access and a thank-you — for a full year before adding anything you have to produce.

The mistake is almost never charging too little. It is building the expensive version of the tier first, discovering it converts the same as the cheap one, and being locked into producing it.

Next issue: Mara on the send button, and why the interface around it is quietly shaping what you write.

No. 11 — The forty people who opened nothing

Crytek — Thu, 03 Sep 2026 00:00:00 GMT

Pull up your list and filter for addresses with no opens in twelve months. For most newsletters past their second year, it is somewhere between a quarter and a half.

The standard advice is a win-back sequence: four emails with increasingly plaintive subject lines, ending in “is this goodbye?”. I have run those. They convert around two percent, and the two percent were mostly going to open the next issue anyway.

Here is the case for the other approach.

What a dormant subscriber costs

They cost you money. Every provider prices on list size. A thousand dormant addresses on a typical plan is real money a year, spent on people who are not reading.

They cost you deliverability. This is the expensive one. Mailbox providers score you on how recipients behave. A third of your sends landing on addresses that never open is a third of your sends saying “this mail is unwanted”. That score is applied to everyone, including the readers who love you.

They cost you clarity. An open rate calculated against a list that is a third dead is a number you cannot use. Every decision you make from it is made through a fog.

What removing them actually did

I suppressed 1,340 addresses from a list of 4,200 in February. Nobody was deleted — suppressed, so the records exist and anyone who comes back is recognised.

Open rate went from 28% to 44%. Same readers, honest denominator.
Inbox placement in Gmail improved measurably; Promotions-tab share dropped by about a third.
Provider bill dropped by 40 a month.
Six people noticed and re-subscribed.

The list looked smaller on the dashboard and more people read the newsletter. Both of those were true at once, and only one of them mattered.

A subscriber count is a vanity number the moment it stops tracking readers. The point at which that happens is earlier than anyone wants to believe.

How to do it without being dramatic

Define dormant honestly. No opens and no clicks in twelve months, for addresses that have received at least ten issues. A signup from three weeks ago is not dormant, it is new.

Send one email. One. Subject: “Still want this?” Body: three sentences and two links — keep it, or stop. No countdown, no guilt, no “we’ll miss you”. The four-email sequence exists to squeeze the two percent, and the squeezing is what makes people resent you.

Suppress, do not delete. Keep the record so you can tell a returning subscriber from a new one, and so you never mail them again by accident.

Do it once a year. On a date, in the calendar. It is hygiene, not a campaign.

The part that is uncomfortable

You will watch a number you have spent years growing go down by a third in an afternoon. If you have ever quoted that number to a sponsor or on a pitch, it will feel like a loss.

It is not. It is the difference between the audience you have and the audience you report. Only one of those replies to you.

Next issue: the arithmetic behind a paid tier, run before you spend three months building one.

No. 10 — One column is not a limitation

Arkane Studios — Thu, 20 Aug 2026 00:00:00 GMT

Designers coming to email from the web tend to describe it as a downgrade. No flexbox you can trust, no web fonts in Outlook, no JavaScript, a viewport that might be 320px or might be a preview pane four inches wide. Tables. In 2026.

I think this framing is wrong, and I think it produces bad email.

There is exactly one real constraint: one column, unknown width. Everything else on that list is a detail of how you satisfy it. And a single column of unknown width is not a poor version of a web page. It is a different medium with a long and rather good history, which is the medium of the printed letter.

What the constraint gives you

It removes the layout question. You cannot put a sidebar next to the text, so you will not spend two days deciding whether to. Every decision left is about the words and their rhythm, which is where the decisions should have been.

It forces sequence. In one column there is no “over there” — only “before” and “after”. That is a stronger organising principle than most web layouts manage, and readers follow it without instruction.

It is legible by default. Cap the column at about 560 pixels and you land on a measure of 60-75 characters, which is where typographers have landed for four hundred years. You get that for free.

The four moves worth having

Once you accept the column, the whole design vocabulary is small enough to list.

Measure. 520 to 600 pixels. Narrower reads as cramped on desktop; wider breaks the line length on any screen over 700.

Rhythm. One vertical spacing unit, used consistently. Paragraph gap of 1, heading gap of 2 above and 1 below, section break of 3. That is the whole spacing system and it will look considered.

One rule, used once. A single horizontal line, above the sign-off. It marks the end of the reading. Use two and it becomes decoration; use none and the ending is soft.

One photograph, in the middle. Never at the top — an image above the first paragraph is the single strongest “this is marketing” signal there is. In the middle it works as a breath.

Everything you want that the column will not give you, the archive page can. Stop fighting the email for layout it cannot do and put it on the web, where it is free.

The split that makes this easy

This is the part people miss. You do not have to choose.

The email is the letter: one column, restrained, made to survive an unknown client. The web version of the same issue can do whatever you want — a wide cover, a proper grid of related issues, a topic index, a search box. They are the same content with different jobs, and treating them as one design problem is what produces an email full of things that do not work and a web page that looks like an email.

Design the letter for the inbox. Design the archive for the browser. The writing is the only thing that has to be identical.

Next issue: Sam on the forty percent of your list who are not reading any of this.

No. 9 — Ellen Wu has sent 300 issues without a schedule

Arkane Studios — Thu, 06 Aug 2026 00:00:00 GMT

Ellen Wu writes Margin Notes, a letter about editing and the people who do it, to about fourteen thousand subscribers. She has sent three hundred and four issues since 2017. The gaps between them range from four days to seven weeks.

Every piece of advice in this field says that should not work.

Let’s have the argument straight away. Why no schedule?

Because I tried one and it made the letter worse. I committed to Tuesdays for eleven months. About a third of those issues existed because it was Tuesday. You can tell which ones. I can tell, and so could the readers — the Tuesday-filler issues got half the replies.

But surely people need to know when to expect it.

They need to know what to expect. That is a different promise and I think we have confused the two. My readers know that when Margin Notes arrives, it is finished. That is the consistency. The interval is not the product.

What does that do to your numbers?

My open rate is 52 percent, which is higher than anyone tells me it should be. I think that is causal, not despite. People open it because it is not routine. There is no habituation.

On what replaces the deadline

If not a schedule, what stops you drifting for a year?

Two things. I have a rule that I never go more than eight weeks, and if I am approaching six I send a short one — genuinely short, three hundred words, one observation. That keeps the address warm and it keeps me honest.

And I write constantly. There are always four or five half-issues open. The gap between sends is not a gap in working, it is a gap in finishing.

Does that not just become a schedule with extra steps?

A ceiling is not a schedule. A schedule tells you when to send. A ceiling tells you when you have stopped.

A deadline makes you produce. A ceiling stops you disappearing. They are not the same tool and most people only use the first one.

Wu keeps four or five issues open at once. The gap between sends is not a gap in working. Photo by Codioful

On the readers

Do people complain about the irregularity?

Once every couple of months someone asks when the next one is. I reply and say I do not know, and that is usually a nice exchange. Nobody has ever unsubscribed over it in a way they told me about.

And deliverability? Irregular sending is supposed to be a problem.

It is a problem if your volume swings. Mine does not — the list grows slowly and I send to all of it every time. Gmail does not have a diary. It has a reputation score, and mine is fine because people open and reply.

Three hundred issues. What has changed most?

I stopped writing for the person who might subscribe and started writing for the person who already did. That happened around issue sixty and it is the only decision I would call important.

What would you tell someone on issue nine?

Send the one you are embarrassed by. Not the bad one — the one that feels too personal or too specific or too small. Those are the issues people write back about, every single time.

Margin Notes arrives when it is finished. Ellen has never used the word “weekly” in any of the three hundred and four.

No. 8 — Plain text still beats your template

Iris Novak — Thu, 23 Jul 2026 00:00:00 GMT

Two issues ago I mentioned killing a design system I had paid a designer to build. Several of you asked for the numbers, so here they are.

The test

Four months, alternating issues. Version A was the full template: header image, brand colours, section dividers, a footer with social icons and a card layout for links. Version B was one column of text, a single rule above the sign-off, links in the body, and nothing else.

Same writer, same lengths, same send times, same list. Not a clean experiment — alternating means seasonality is baked in — but eight issues each is enough to see a direction.

	Template	Near-plain
Open rate	41.2%	43.8%
Click rate	3.1%	4.9%
Replies per issue	6	19
Promotions tab (Gmail)	68%	22%
Unsubscribes per issue	11	7

The open rate difference is inside the noise. The rest is not.

Why the plain one wins

It looks like mail. Gmail’s Promotions tab is not a spam filter; it is a classifier, and a header image with brand colours and a row of social icons is exactly the shape of a marketing email. Removing the furniture moved two thirds of my Gmail readers back into the primary inbox, which is worth more than every other change I have ever made.

It renders everywhere. There is no version of a one-column text email that breaks in Outlook 2016, in dark mode, at 320px, or with images disabled. The template broke in three of those four.

It reads as a person. The reply numbers are the real story. Nineteen versus six, from the same writing. People reply to letters. They do not reply to newsletters that look like product announcements.

Most of your readers decide on three things: sender name, subject, and about forty characters of preview text. Photo by Codioful

What “near-plain” actually means

Not literally plain text. It is still HTML, because you want links, emphasis and a reasonable line length. It is just restrained:

<body style="margin: 0; padding: 24px 16px; background: #fff">
  <div style="max-width: 560px; margin: 0 auto; font: 16px/1.6 system-ui, sans-serif">
    <!-- the issue -->
  </div>
</body>

That is the whole template. A max width, a system font stack, a colour, and generous line height. Everything else is the writing.

Printed post has looked the same for a century, and nobody calls it under-designed. Photo by Codioful

Two rules I stuck to:

No image above the first paragraph. The first thing in the message should be words.
No tracking parameters on internal links. They make your URLs look like a campaign, and you can read the referrer anyway.

If your email needs a design system, the thing you are sending is not a letter.

The part I got wrong

I did keep one image per issue, in the middle, and I still think that is right. The argument against decoration is not an argument against photographs — it is an argument against chrome.

Next issue: Leah talks to someone who has sent three hundred issues without ever committing to a schedule.

No. 7 — Six tools I stopped paying for

Rockstar Games — Thu, 09 Jul 2026 00:00:00 GMT

I did an audit in March. The stack behind a newsletter with about four thousand subscribers had reached 214 euros a month across nine subscriptions, most of which I could not remember signing up for.

It is now 38. Here is what went, and why.

Gone: the analytics suite

I was paying 49 a month for dashboards I looked at twice a month and could not act on. Cohort retention curves are interesting for a product. For a fortnightly letter, the only numbers that change anything are: how many opened it, how many replied, and how many left. My sending provider reports all three.

The trap with analytics tools is that they answer questions you did not have. I was optimising a subject line A/B test with a sample size that could not possibly reach significance, and calling it work.

Gone: the landing page builder

29 a month for a single page. It was a good page. It was also a page I edited twice in fourteen months, hosted on a domain I did not control, behind a builder that would have taken the page with it if the company folded.

It is now a static page in the same repository as everything else. Same conversion rate, measured over three months.

Gone: the scheduling tool

I wrote issues in one app, formatted them in another, and scheduled them in a third. The scheduling tool existed entirely to move text between the other two. Writing directly in the provider’s editor is slightly worse and removes an entire class of formatting bug.

Gone: two “audience growth” products

One promised recommendations from similar newsletters. It sent 900 subscribers in four months, of whom 61 ever opened anything. The other was a referral-rewards system nobody used, because the reward was a PDF.

Both were doing the same thing: adding addresses that did not want to be there, which costs you twice — once in fees, once in deliverability.

Any tool that grows your list faster than your writing earns it is borrowing against your sender reputation.

Gone: the design system

I built a beautiful HTML template with a designer. It rendered differently in seven clients and I spent an afternoon a month maintaining it. That one deserves its own issue, and gets one in two weeks.

What survived

The sending provider. 29 a month at this list size. Non-negotiable, and the place to spend if you are spending anywhere.

A domain and static hosting. About 9 a month all in, for the archive and the signup page.

A notebook. For the idea that turns into an issue, which has never once arrived while I was looking at a dashboard.

The honest conclusion of the audit was not that the tools were bad. It was that buying them felt like progress at times when writing was hard, and the spending tracked my anxiety more closely than it tracked the newsletter.

Next issue: the template, and why the plainest version of it beat the beautiful one.

No. 6 — The welcome email is doing all the work

Crytek — Thu, 25 Jun 2026 00:00:00 GMT

Across the newsletters I have had access to the numbers for, the welcome email is opened between three and five times more often than a regular issue. It is the only message you will ever send at a moment when the reader is actively thinking about you.

It is also, almost universally, the worst-written thing on the list. Usually it is the provider’s default with the name changed.

Why it matters more than the next six issues

A new subscriber is making one decision in the first few days: is this going to be worth the inbox space. Nothing you send in week three can fix a bad answer in week one, because by week three you are being filtered, ignored, or archived unread.

There is also a mechanical reason. A brand-new address that opens and replies to your first message is the cleanest possible engagement signal, and it arrives at exactly the moment your sending reputation to that recipient is being established.

What to put in it

Four things, in this order, and nothing else.

Confirm what they get and when. One sentence. “Mailer arrives every other Thursday morning.” If your schedule is irregular, say that instead — vagueness is worse than irregularity.

Give them something to read now. Not “check out the archive”. Link two specific issues, with one line each on why. The gap between signing up and receiving anything is the most dangerous window you have; fill it.

Ask one question. “What made you subscribe?” gets a real answer surprisingly often, and every answer is research you cannot buy. It also generates a reply from a new address, which is the single best thing that can happen to your deliverability with that recipient.

Set the unsubscribe expectation. Counterintuitive, and it works. Telling someone the exit is easy makes them less likely to use it, because the decision stops feeling permanent.

Everything else people put in a welcome email — the founder story, the manifesto, the list of what you will never do — is for the writer, not the reader.

What to leave out

A five-part onboarding sequence. Nobody asked for a course.
Your life story. It goes on the about page, where someone who wants it can find it.
A request to add you to their contacts. It reads as needy and almost nobody does it.
Attachments of any kind.

The bit people skip

Read yours. Out loud. Right now, in the account you subscribed to your own newsletter with — because you did subscribe to your own newsletter, on a normal consumer address, and you do check what it looks like on a phone.

Most people find at least one of: a broken link, a merge tag that never got filled in, a sentence about a schedule they abandoned eighteen months ago, or a sign-off from someone who no longer works on it.

Then diary a reminder to read it again in six months. It is the only email you send that goes stale silently.

Next issue: Iris on the tools she stopped paying for, and the one she would pay double for.

No. 5 — How to end an issue

Rockstar Games — Thu, 11 Jun 2026 00:00:00 GMT

There is a whole genre of advice about opening lines. Hook the reader. Start in the middle. Never begin with “I”. Fine, all useful.

Almost nobody writes about endings, which is strange, because the ending is the part that does the work. It is where a reader decides whether to reply, whether to forward, and — quietly, without noticing — whether to open the next one.

Here is what I have learned from writing about three hundred of them badly.

The four ways an issue ends badly

It trails off. You have made your point, you are tired, and you write “anyway, that’s my two cents” and hit send. The reader feels the energy drop and closes the tab. Nothing bad happens. Nothing good happens either.

It summarises. “So to recap: authentication, hygiene, consistency.” If the piece needed a recap, the piece was too long. A summary at the end of nine hundred words tells the reader you did not trust them to follow it.

It asks for everything at once. Reply, forward, share on social, reply to the survey, join the Discord, upgrade. Five asks is zero asks.

It ends on a link. Sending someone away in the final paragraph is the one genuinely costly mistake on this list, and it is the most common. Your last sentence is the most-read sentence in the issue after the subject line. Do not spend it on somebody else’s website.

What to do instead

Pick one of three endings and rotate them.

The open question. Not “what do you think?” — that is a non-question and gets non-answers. A real one, narrow enough to answer in a sentence: what is the longest gap you have taken, and did the list survive it? You will get replies, and replies are the strongest deliverability signal you have.

The forward-looking line. One sentence about what is in the next issue, specific enough to be a promise. “Next time: what plain text actually does to your numbers.” This is the closest thing to a retention lever that exists in this format, and it costs you eleven words.

The turn. Restate the argument in a way that is slightly sharper than how you opened it. Not a summary — a compression. The reader should feel that the piece arrived somewhere rather than stopping.

The ending is where the reader decides what the issue was about. Whatever you put there is what they will remember, regardless of what you spent the middle on.

The sign-off itself

Keep it the same every time. Readers use it as punctuation; changing it every issue is like changing your handwriting.

And put your actual email address under it. Not a form, not a “reply to this email” instruction — the address, written out. It reads as a person rather than a system, and about one reader a month will use it in a way that changes what you write next.

Next issue: Sam on the welcome email, which is probably the best-performing thing you will ever write and almost certainly the least edited.

No. 4 — Deliverability, without the folklore

Sony — Thu, 28 May 2026 00:00:00 GMT

Deliverability advice online is about eighty percent folklore. Avoid the word “free”. Never use more than three links. Do not attach images. Keep your text-to-image ratio at 60:40, a number someone invented in 2011 and nobody has questioned since.

None of that is how modern filtering works. Here is what actually decides whether your issue lands, roughly in order of weight.

1. Authentication

If SPF, DKIM and DMARC are not set up correctly on your sending domain, nothing else you do matters. Gmail and Yahoo both require authentication for bulk senders now, and “bulk” starts at 5,000 messages a day — which sounds like a lot until you have two thousand subscribers and send twice a week.

Your provider will have a page on this. Do it properly, with your own domain, not their shared one. Then verify it independently: send yourself an issue, open the raw headers, and check for three pass results. Do not trust the dashboard’s green tick.

2. Engagement, and specifically recent engagement

Mailbox providers watch what your recipients do. Opens are a weak signal now that image proxying is standard, but these are strong:

Replies. Enormously strong. One reply is worth a hundred opens.
Moving a message out of spam.
Adding you to contacts.
Clicking through and staying.

And the negative ones: deleting without opening, repeatedly, over weeks. That is the signal that quietly kills a list.

3. List hygiene

This is where most small senders are losing, and it is the easiest to fix. If a third of your list has not opened anything in a year, every send is telling Gmail that two thirds of the people who receive your mail do not want it.

Suppress anyone with no engagement in twelve months. Not delete — suppress, so you keep the record. Send them one re-permission email first if it makes you feel better, but do not send six.

A smaller list that gets opened is worth more than a big one that gets ignored, and this is not a metaphor. It is literally how the scoring works.

A list that never opens anything is telling the provider something, every single send. Photo by Codioful

4. Consistency

Sending 400 messages a week for six months and then 9,000 in one afternoon looks like exactly what a compromised account looks like. If you are about to send to a much larger list than usual — after a big signup spike, say — ramp it. A few thousand at a time, over a few days.

What barely matters

Specific words. Filters have been statistical for fifteen years; no single word triggers anything.
Number of links, unless they point to domains with bad reputations of their own.
Emoji in subject lines. Annoying, not filtered.
Image-to-text ratio. A pure-image email is a problem because it is unreadable, not because of a ratio.

The one test worth running

Before a send, mail the issue to accounts on Gmail, Outlook and one corporate domain if you have access to one. Not to check if it arrives — to check where. Gmail’s Promotions tab is not spam, but it is not the inbox either, and for a lot of newsletters that tab is where the open rate went.

If you are landing in Promotions and want out: fewer tracking parameters on your links, fewer images at the top, and more replies. That last one is the real lever, which is why every good newsletter ends with a question.

Next issue: how to end one.

No. 3 — Marco Peña on charging for what he gives away

Arkane Studios — Thu, 14 May 2026 00:00:00 GMT

Marco Peña has written Slow Signals, a letter about civic technology, every fortnight since 2015. About nine thousand people get it. Roughly six hundred pay for it. Nobody has to.

We spoke over a bad video connection while he was between trains.

Let’s start with the obvious one. Why is there a paid tier if it unlocks nothing?

Because the alternative was worse. I spent two years arguing with myself about a paywall. Every version I sketched made the letter less useful. The whole point is that a council officer in a town of nine thousand can read it. The moment I put the good part behind twelve euros a month, that person is reading the free half, which is the half I care least about.

So what do people get for paying?

The same issue everyone else gets, a day early, and a line at the bottom of it saying thank you. That is genuinely it. I tried adding a members’ thread. It was miserable. Twenty people posted, I felt obliged to reply to all of them, and it turned a writing job into a moderation job.

Six hundred paying for nothing extra is a lot of people paying for nothing extra.

They are not paying for nothing. They are paying for it to continue. That is a real thing to buy. I think the mistake people make is assuming money follows access. Sometimes it follows relief — the relief of not having to watch something you like go under.

On the years before it worked

How long before it was a living?

Seven years. I want to be specific because I read a lot of interviews at year three that made me feel like a failure. Seven years of writing it on Sunday evenings after a full week of consultancy.

Did you ever consider stopping?

Twice properly. Both times in January, which I now think is just a January thing. The second time I did stop, for four months. I told the list I was stopping and why. Six hundred people replied. That was the year it turned.

Telling people you might stop is the most honest growth tactic there is, and you can only use it once.

Did the list shrink in those four months?

By about eleven percent. Everyone panics about that number. What actually happened is that the eleven percent who left were people who had not opened anything in a year, and my deliverability improved enough that the next issue landed in more inboxes than the one before the break. I came back to a smaller list and more readers.

On the craft

What have you changed most about how you write it?

Length. It used to be two and a half thousand words. It is now about nine hundred. My open rate did not move but my reply rate tripled. People reply to things they finish.

And what have you not changed?

The ending. Every issue ends with one question to the reader and my actual email address. Not a form, not a survey. Eleven years of that.

Last one. Advice for someone on issue four?

Do not look at the number. Look at whether you wanted to write issue five when you finished issue four. That is the only metric that predicts anything at that stage.

Slow Signals goes out every other Tuesday. Marco pays for it with the paid tier described above and, as of last year, nothing else.

No. 2 — The archive is the product

Crytek — Thu, 30 Apr 2026 00:00:00 GMT

Ask someone why they did not subscribe to a newsletter they were recommended and you rarely hear “the writing was bad”. You hear “I could not tell what it was”.

That is an archive problem.

A landing page says what you intend to write. An archive shows what you have written, which is a completely different claim and a much more credible one. Twelve titles in a row tell a reader the range, the frequency, the length, and whether you keep your promises — in about four seconds, with no copywriting required.

So the useful question is not “how do I write a better signup page”. It is “can a stranger read my last ten issues without an account”.

For a surprising number of newsletters, the answer is no. The archive is behind the provider’s default template, on a subdomain nobody links to, sorted badly, with no excerpts.

What a good archive does

Three things, in order of importance.

It shows the shape. Issue numbers, dates, titles, one line of what it was about. A reader should be able to scan thirty issues without scrolling past four of them. This is why a list beats a grid of cards here — cards are for browsing when you do not know what you want, and lists are for scanning when you do.

It is honest about frequency. If there is a nine-month gap in 2025, show it. Hiding dates does not make the gap smaller, it just makes the whole archive feel unmoored. People forgive gaps. They do not forgive finding out later.

It is linkable. Every issue on its own URL, with a title that makes sense out of context, and a preview image that is not your logo. Someone will link to one of these in a group chat eventually, and that link is worth more than any post you will write about yourself.

The thing nobody does

Cross-link your own back issues inside new ones.

Not a “related posts” block at the bottom — those get ignored. I mean genuinely: when you make a claim you argued properly eight months ago, link the issue where you argued it. It does three jobs at once. It saves you re-explaining. It shows a new reader there is a body of work behind the sentence they are reading. And it moves people from one issue to two, which is the single biggest predictor of whether they stay.

A reader who has read two of your issues is a subscriber. A reader who has read one is a visitor.

A test you can run today

Open your archive in a private window. Give yourself thirty seconds. Then close it and write down:

How often does this come out?
What are the three things it is about?
Which issue would you read first?

If you cannot answer all three, a stranger cannot either. The fix is almost never more writing. It is usually ordering, excerpts, and dates.

Next issue: Leah talks to someone who has been running the same letter for eleven years and has opinions about charging for it.

No. 1 — Why anyone still starts a newsletter

Bulkhead — Thu, 16 Apr 2026 00:00:00 GMT

A friend asked me last month whether starting a newsletter in 2026 was, and I am quoting, “a bit late”. She had a point. The format has been declared dead roughly once a year since 2018, usually by someone who has just launched one.

Here is the answer I gave her, and the reason this letter exists.

Nobody stands between you and the reader

Every other place you can publish has an intermediary with its own interests. A social platform decides how many of your followers see a post, and that number moves whenever the business model does. A search engine decides whether your page is the answer or the also-ran. A video platform decides what counts as watch time.

Email has none of that. You write an address into a box, the message arrives. There is a spam filter in the middle, and it is a real constraint — we will spend plenty of issues on it — but it is a filter with published rules, not a market. Clear the rules and you are through.

That is the entire argument. It is unglamorous and it has not changed in twenty years, which is exactly why it is worth building on.

The list is yours in a way nothing else is

The second argument is portability. If your sending provider raises its prices, or gets bought, or quietly changes what it will let you say, you export a CSV and go somewhere else. Your readers come with you. Try that with an audience that lives inside somebody else’s app.

This matters less on day one than it does in year three, which is why people skip it. It is also the difference between a project you own and a project you rent.

A list of four hundred people who asked for your writing is worth more than forty thousand followers who were shown it once.

The question has not changed much. The distribution has. Photo by Codioful

What you are actually signing up for

None of that means it is easy. Running a newsletter is mostly deadlines. The writing is the fun part and it is maybe a third of the work. The rest is:

Deciding what an issue is, and being consistent enough that readers know what they are getting.
Getting the thing to render correctly in an email client written in 2003.
Finding readers without buying them.
Writing the next one after the one that flopped.

If you want the audience without the deadline, you want a blog. That is a perfectly good thing to want. The reason to choose email is that the deadline is the product: it is what turns “I will write something eventually” into a thing that exists twice a month.

Mailer is a letter about the parts of this that nobody writes about. Not growth hacks — there is enough of that. More like: how to end an issue so it does not trail off. What plain text actually does to your open rate. Why your welcome email is doing more work than your last twelve issues combined.

It goes out every other Thursday. The archive is public, there are no paid tiers, and if you want out, the unsubscribe link is at the top and the bottom of every issue.

That is issue one. Next time: why the archive matters more than you think, and what happens when you treat it as a product instead of a backup.