We covered many things: why a client of hers grew 1,000 subscribers with notes that got almost no likes, why reposting your best note ten times doesn’t annoy anyone, and why the 26,000 people who saw my content without following me matter more than the ones who already did.

I also shared my screen and walked through how I schedule a full week of notes in under 30 minutes, clear 162 notifications in half an hour, and turn old articles into notes that send readers back to them.

If you’ve been posting daily and wondering why the number won’t move, the answer is probably in here.

Your weekly digest of the most relevant data, giving you insights NOBODY else has.

You have a thought. It is a good one. You open the Notes composer, it comes out in two sentences, and then you sit there deciding whether two sentences is enough.

So you add a line. Then a second line to explain the first. Then a question at the end so people have something to reply to. By the time you hit post you have a small essay, the small essay gets eleven likes, and you have no idea whether the problem was the idea, the timing, the length, or the question you tacked on the end.

I have watched writers agonise over this for two years without a number to point at. Everyone has a theory. Half of them say short is punchy and the other half say long shows depth, and neither camp has ever shown me the distribution.

So I went and got it.

What I measured

I pulled every Substack Note published in July 2026: 705,307 notes scanned, 655,056 with body text. Restacks of other people’s notes are excluded, so what remains is original writing. Then I bucketed all of it by character count, in bands of 50, and looked at likes, restacks, replies, and the shape of the distribution inside each band.

Length is the cheapest variable a writer controls. You cannot make yourself more interesting on demand. You can absolutely decide to stop typing at 280 characters. If length moves the number, it is the most actionable finding available.

It moves the number. It just does not move it the way most people assume, and the honest version of that is the reason this article exists.

1. Notes between 250 and 299 characters average 35% more likes than the platform baseline

The curve is not flat and it is not a straight line. It climbs steadily from the one-liners, spikes hard in a single 50-character band, and then collapses.

Average likes by note length

The 250-299 band averages 24.47 likes against a platform average of 18.17, a lift of 34.6%. Restacks tell the same story more loudly: 3.52 per note, against 1.38 for notes under 50 characters. That is 2.6 times the sharing on the same idea, purely as a function of where you stopped typing.

The fall is steeper than the climb. By 450-499 the average is 13.90, down 43% from peak. By 500-549 it is 12.62, down 48%. A 520-character note performs worse than a 40-character note, which is not what anyone predicts when they add that fourth paragraph.

Mechanism

A Note sits in a scrolling feed with a truncation point. Under roughly 250 characters you have not yet earned a reason to be shared, because there is no argument in the note, only an assertion. Past 300 characters the feed starts hiding the end behind a “more” link, and a reader deciding whether to like something they have not finished reading mostly decides not to. The band is the widest a note can be while still arriving whole.

The honest size of the effect

That 35% number is a cross-sectional comparison, and cross-sectional comparisons on Substack are contaminated by who is writing. Authors with large audiences may simply prefer a certain length. If that were the whole story, the band would be a marker of the writer rather than a property of the writing.

So I tested it. Using the 29,993-note conversion sample, which carries author identity, I took 570 authors who published at least five notes in July and measured each note against that same author’s own July average, on log likes. This removes the author entirely from the comparison. What is left is the effect of length on one person’s own baseline.

Same author, different lengths: performance against their own July average

The shape survives. The peak stays at 250-299. But the magnitude shrinks a lot: within a single author, a note in the band runs about 7% above that author's own average, and about 14% above their own 450-plus notes (+0.067 log points against -0.060, standard error 0.012).

2. The band moves your ceiling, not your floor: median likes stay at 3 to 4 across the entire range

This is the part a careful reader will check first, so I am putting it second instead of burying it in the methodology.

The median note in the 250-299 band gets 4 likes. The median note under 50 characters gets 3. The median note at 500-549 characters gets 3. Across the whole length range, from one-liners to 549 characters, the median never leaves the 3-to-4 window. Length does almost nothing to a typical note.

Median likes barely move. The 90th percentile moves a lot.

The 90th percentile is where the whole effect lives. It runs 27 at the shortest band, peaks at 34 in the 250-299 band, and drops to 23 past 500 characters. The gap between a peak-band note and a 500-character note at the 90th percentile is 11 likes, roughly three times the entire median.

The zero-like rate says the same thing from the other end. It sits at 17.5% for one-liners, bottoms out at 15.0% in the band, and climbs back to 17.8% past 500 characters. A two-and-a-half point swing on a base of one in six. Real, but small.

If someone tells you a length rule doubled their average note, they are describing an author effect or a small sample. The honest claim is narrower and more useful. On a quiet Tuesday the band changes nothing at all. On the day you write something good, it changes how far that note travels.

3. Three paragraph blocks is the structural ceiling, and the fourth block costs you

Length is half the structure question. The other half is how many times you press return.

Across the full 655k, average likes climb from 18.45 for a single unbroken block to 20.14 at three blocks, the structural peak, then fall to 17.32 at four blocks and 16.02 at five. Three blocks beats four by 16%.

Average likes by number of paragraph blocks

4. The first block wants the band too: a 270-279 character opening averages 24.9 likes

Here is the finding that surprised me. I ran the same length analysis on the first block alone, ignoring everything after the first line break, and the curve has the same shape and the same peak.

Average likes by length of the first block

Notes whose first block runs 270 to 279 characters average 24.87 likes with a 90th percentile of 42, the highest p90 of any first-block bucket. Notes with a first block over 300 characters average 18.18 (n = 43,397). Notes opening with a fragment under 10 characters average 14.33 (n = 21,088). The opening block behaves like a small note in its own right, and the same 250-to-300 window applies to it.

Inside the 250-299 band the same pattern shows up as a share. English notes in the band that cleared 100 likes give 60.9% of their characters to the first block. Notes that finished under 10 likes give 56.9% (n = 1,528 and 884 respectively). The winners front-load.

Mechanism

The first block is the only part guaranteed to be read. Everything after the first line break is competing with the scroll. A writer who splits a 280-character idea into four even blocks has spread the payload across four separate gambles. A writer who puts 220 of those characters in block one has already delivered the whole thing before the reader had a chance to leave, and the remaining 60 characters land as a closing beat rather than as the point.

5. A question opener throws the band away: 113 hits per 100 flops against 176 for a statement

The band does not substitute for the opener, it multiplies whatever the opener already is.

Working from 51,852 English notes in the stratified samples, I scored each length band by a simple ratio: how many notes cleared 100 likes for every 100 that finished under 10. Then I split by whether the first block ends in a question mark.

Hits per 100 flops, by opener type and note length

Two things fall out. First, the band replicates cleanly on a completely different metric. For statement openers the ratio runs 123, 121, 140, 176 in the band, then 128 and 102. That is the same curve as Finding 1, computed from the tails of a separate sample rather than from means.

Second, a question opener costs more than the band gives. A question inside the peak band scores 113. A statement at 300-449 characters scores 128, and a statement at 50-149 characters scores 121. Landing a question in the perfect window still leaves you behind a statement written at a mediocre length. The question-opener sample is small at 2,382 notes, so I would not quote the exact ratio to a decimal, but the direction holds at every one of the six bands.

The full population agrees. Question hooks average 14.60 likes with a 90th percentile of 22 (n = 32,452), against 18.17 and a p90 of 28 across all 655,056 notes. Questions do buy replies, 2.41 per note against 1.42 platform-wide, so if comments are the goal the trade is defensible. For reach it is not.

Mechanism

A question hands the reader a task. Before they can react they have to work out whether the question applies to them, and then decide whether their answer is worth typing in public. A declarative sentence hands them a finished thought they can agree with in one tap. On a feed measured in hundreds of milliseconds per note, the free option wins.

6. Subscribers run the opposite way: notes under 100 characters convert 2.5 times better per note

Everything above optimises reach. If your goal is subscribers, the advice inverts, and I would rather print the contradiction than pretend the variable only points one way.

The 29,993-note conversion sample carries publisher metrics. Bucketed by hundreds of characters, notes under 100 characters average 0.585 free subscribers each, against 0.236 for the 200-299 bucket. They also produce 0.031 follows per note against 0.016, and 41.2 clicks against 25.0. The shortest notes on the platform are the worst at reach and the best at conversion.

Free subscribers per note, by note length

The 500-599 bucket also spikes, to 0.397, but it carries only 1,591 notes and I would treat it as noise until it replicates.

I can measure what happens here. I cannot see attachments in this dataset, so I cannot prove why. The likeliest explanation is that a one-line note is usually a caption or an announcement sitting on top of something else, and a reader who acts on it has nowhere to go except through. That is a hypothesis, not a finding.

One more thing ties the two goals together, because it is the single biggest own goal in the dataset. Notes containing a URL average 10.74 likes against 19.36 for notes without one, and their zero-like rate doubles from 14.6% to 29.0%. Notes containing a Substack link average 7.85. Inside the 250-299 band, English notes that finished under 10 likes carried a link 18.6% of the time, while notes that cleared 100 likes carried one 9.0% of the time.

A link is the thing that converts and the thing that kills reach. Spending your 280 characters on a URL is spending them on the conversion goal, whether you meant to or not.

Subscribe now

The numbers behind the rules

Write the note first. Then check it against the distribution it has to live in. Every rule that falls out of these numbers is collected in the actionable block at the end, so this is the reference rather than the instruction.

The band at a glance

Your weekly digest of the most relevant data, giving you insights NOBODY else has.

You have had this note. The one that did well. Four hundred likes, a warm comment thread, people telling you it was exactly what they needed to read today.

Six restacks.

And then a week later you write something you thought was thinner, less personal, less carefully built, and it gets restacked ninety times by people you have never heard of, and it drags in subscribers for three days.

Likes and restacks look like the same button wearing different hats. They are not measuring the same thing. A like is something a reader does to your note. A restack is something a reader does with your note, in front of their own audience, with their own name attached. The second one is a much higher bar, and the things that clear it are not the things that earn affection.

Finding 1: One emoji costs you 17% of your restacks and buys you zero likes

Read more

Your weekly digest of the most relevant data, giving you insights NOBODY else has.

You post a Note. Fourteen people like it. Nobody says anything.

That silence is the normal outcome, and it is worth understanding how normal. Of the 655,056 Substack Notes published in July 2026 with body text, 405,147 finished with zero replies. That is 61.8%. The median Note in that silent group collected two likes and one conversation with nobody.

Most advice about this is some version of “ask a question.” Which is fine, except it is also what everybody already does, and 18.1% of English Notes in my sample contain a question mark while only 38% of all Notes get even one reply. So the advice is either wrong or badly underspecified.

It turned out to be badly underspecified. There is a version of a question that pulls replies at roughly double the rate of an equivalent Note with the same reach, and a version that performs no better than no question at all. The difference is mechanical and easy to copy once you can see it.

Not here for the nitty gritty? Scroll down to the TL;DR section to get the gist of the article

Finding 1: A question opener nearly doubles replies, even against Notes with identical reach

Across the full 655,056 Notes, question openers average 2.41 replies against 1.36 for the “other” opener category (n=32,452 question openers). That gap could easily have been a reach artifact.

It is not. Inside every single like bin, question-opening Notes pull between 1.8 and 2.6 times the replies of Notes with the same number of likes. Pooled across bins and weighted by bin size, English question openers average 10.83 replies against 5.57 for everything else at matched reach. That is a lift of 1.94.

Average replies at matched like counts, question opener vs everything else

The consistency is the part worth noticing. A confound that produced this would have to inflate replies by roughly the same factor at 3 likes and at 500 likes, which is not what reach artifacts look like.

Mechanism

A like is a one-tap exit. A reply requires the reader to compose something, which means the Note has to hand them a sentence they already know how to write. A question at the top of a Note does that before the reader has decided whether to keep scrolling, so the composition starts while attention is still on the screen. Every line after that is the reader rehearsing an answer they are already half-holding.

Finding 2: The question belongs in the last line, not the first

Once I split question-containing Notes by where the question actually sits, the opener stopped being the strongest position. Notes that end on a question average 11.38 replies at matched like counts. Notes that open on a question but end on something else average 9.96. Notes with the question buried in the middle average 7.33. Notes with no question at all average 5.24.

Average replies at matched like counts, by where the question sits

Restricting to multi-line Notes only, so that position is genuinely a choice, closing on a question beats putting it anywhere earlier by 1.46 times (n=1,137 against 5,477). Only 4.6% of English Notes in my sample end on a question mark, which makes this one of the cheapest unclaimed moves on the platform.

Mechanism

The last line is the one the reader is holding when their thumb reaches the reply box. A question at the top has to survive the whole body before it gets acted on, and by the last paragraph the reader has usually replaced it with whatever the Note actually argued. Ending on the question removes the gap between the prompt and the box.

Finding 3: Ask for something the reader can retrieve, not something they have to work out

This was the hypothesis I most wanted to break, because it is the one everybody states without evidence. The claim is that a question pulls replies when it leaves a gap the reader can fill from memory, and dies when it asks them to think.

It held. Among Notes that contain a question, questions that address the reader directly average 10.88 replies at matched like counts against 7.92 for abstract questions with no “you” in them, a lift of 1.37. And the hard version genuinely does nothing: questions opening with “why” or “how” average 0.94 times the replies of other questions at matched reach (n=730), which is to say no lift at all, and at 100 likes and above they run slightly behind.

Average replies at matched like counts, by kind of question

Narrowing further, questions in the closing line that ask for a specific retrievable item (”what’s your”, “which”, “name one”, “what are you”) run at 1.84 times the matched baseline (n=284, small, so treat the size loosely and the direction as solid).

Mechanism

“Which novels made you want to write fiction” and “how do we fix publishing” both look like questions. Only one of them has an answer sitting in the reader’s head already. The second one asks for an essay, and the reader correctly decides they do not have time to write an essay in a comment box, so they like it and leave. The gap has to be the size of a sentence the reader has said out loud before.

Finding 4: A link in the Note costs you a quarter of your replies

This one was not in my hypothesis list. English Notes containing a URL average 5.07 replies at matched like counts against 6.95 for Notes without one, a lift of 0.73. The penalty appears in nine of nine like bins, ranging from 0.52 at zero likes to 0.84 at 50 to 99 likes.

Reply rate of Notes containing a URL, relative to Notes without one at the same like count

Narrowing to the closing line specifically, Notes whose last line is a link or a subscribe pitch run at 0.81 of matched baseline. Nine percent of English Notes in the sample carry a URL, and they are systematically buying reach at the expense of conversation.

Mechanism

A link is an exit. It converts the Note from a place to say something into a doorway to somewhere else, and the reader who taps it is no longer in the feed to reply. It also changes what the last line is asking for. A closing URL asks for a click, and readers give the platform exactly one action per Note.

Finding 5: At the same reach, high-reply Notes convert about 20% better

Same method. Bin by likes, split each bin at its median reply count, compare subscriber acquisition between the two halves. Bin-weighted, high-reply Notes acquired 26.0 subscribers per 100 Notes against 21.7 for low-reply Notes with the same number of likes, a lift of 1.20. A permutation test that shuffles reply counts within like bins puts that at p = 0.031, so it is unlikely to be noise, but it is not a landslide either. Clicks moved by the same 1.20 factor.

Subscribers acquired per 100 Notes, high-reply vs low-reply half of each like bin

Two honest qualifications. The lift is consistent in direction across every bin I could measure but it shrinks as reach grows, from 1.93 in the 1-to-2 like bin down to roughly parity above 100 likes. And a reply half is not a randomised treatment, so some of this is that better Notes get both. The useful reading is modest: at the reach you actually have, a Note that starts a conversation is worth about a fifth more than a Note that does not.

Subscribe now

Finding 6: The like cost is real but it lives entirely in the tail

Question openers do earn fewer likes. Across the full 655,056 Notes they average 14.6 likes against 18.4 for the “other” category. That is the trade the platform makes you take, and it is worth naming precisely, because “questions get fewer likes” is true in a way that does not mean what it sounds like.

The median question Note and the median non-question Note both get exactly 3 likes. The gap opens at the 90th percentile, 22 likes against 28, and widens from there into the averages. Question Notes are not disliked. They are less likely to produce a monster.

Likes for question openers vs everything else, at three points in the distribution

A within-author check says the same thing from a different angle. Among 344 authors who published at least two question-opening Notes and at least five others in July, the median author’s question Notes earned 0.99 times their own median likes and 1.46 times their own median replies. Roughly half of those authors saw no like penalty at all on their own feed.

What this means for the trade

If your goal is one Note that reaches 40,000 people, a question opener is the wrong instrument, because the ceiling is lower. If your goal is a Note that produces twenty conversations you can convert, the median outcome is unchanged and the reply count roughly doubles. Most writers below 5,000 subscribers are not competing for the tail, and are trading away something they were never going to get.

For completeness, Open Letter openers pull the highest reply rate of any category I measured, 4.22 replies against a 1.42 platform average, but on only 897 Notes in the entire month. The sample is too thin to build a habit on.

Six closers that pull replies

Each of these is a template abstracted from Notes in the sample that outperformed their like-matched peers. Swap the bracketed part for your own subject.

1. The constrained retrieval. “In one sentence, what’s your [thing] about?” A word limit removes the reason not to answer. Retrieval asks in the closing line run 1.84× matched baseline (n=284).

2. The lookup. “Which [book / song / mistake] made you [outcome]?” “Which” outperforms “why” because the answer is a noun, not an argument.

3. The small specific. “What are your favourite [very small everyday thing]?” The smaller the category, the more people have one.

4. The solidarity check. “Anyone else, or is it just me?” Worth 1.71× matched baseline on a thin sample (n=124), so treat the number loosely. The reply it invites is one word long.

5. The direct ask. “Tell me yours below.” Explicit ask verbs are worth 1.44× (n=2,787), and this is the version that works when your Note has no natural question in it.

6. The narrowing follow-up. Ask a broad question, then a narrower one underneath it. Notes with two or more question marks earn 1.66× matched baseline (n=2,468), because the second question tells the reader what shape of answer you meant.

TL;DR

Run this over the Note before you post it. Every line is a comparison at matched like counts, so each one is a real change in outcome rather than a change in reach.

• Does the last line end in a question mark? Closing on a question is worth 2.05× the replies of a Note with the same likes (n=2,387). Only 4.6% of Notes do it.

• Does that question contain the word “you” or “your”? Reader-addressed questions beat abstract ones by 1.37× (n=3,162).

• Can the reader answer it from memory in under ten words? If it starts with “why” or “how,” it earns 0.94×, which is nothing (n=730). Rewrite it as “which,” “what’s your,” or “name one.”

• Is there an explicit ask verb? “Tell me,” “drop yours,” “let me know,” “curious” are worth 1.44× (n=2,787).

• Is there a URL anywhere in the Note? Take it out. Notes with a link run at 0.73× on replies (n=4,692). Put the link in the first reply instead.

• Is the Note more than one line? Four or more line breaks earn 1.18× over shorter Notes at the same reach (n=13,051). Single-line Notes run at 0.84×.

Your weekly digest of the most relevant data, giving you insights NOBODY else has.

You write the note. You read it back twice, move a line, delete a line, put the line back. Then you post it.

Then you watch the number stay at zero.

Zero, not merely low. Nobody tapped the heart, nobody replied, nobody restacked it into their own feed. An hour later it has fallen so far down the feed that it may as well never have existed, and you are left with the specific, unhelpful feeling that something was wrong with it and you cannot name what.

That feeling is extremely common, and almost nothing published about Substack Notes is written for it. Every teardown I have ever read starts from a screenshot of something that worked. Which makes sense, because winners are easy to export and losers are not. Nobody wants to build a spreadsheet of their own worst posts.

I built one anyway. Then I read it.

The setup

I pulled every public Substack note posted in July 2026: 705,307 notes scanned, 655,056 with body text after restacks of other people’s notes were excluded. On top of that I took a sample of 16,199 notes that got fewer than ten likes, with the full body text of each one, and a matched sample of 18,532 notes that cleared 100 likes.

The metric I optimized for here is the floor rather than the ceiling. I wanted to know what makes a note get literally nothing, and whether that is a property of the note or a property of the person posting it.

Start with how far down the floor is.

How often a July note produced nothing at all

The median July note got 3 likes. That is the real baseline, not the screenshots you see. And the three zeros stack: a note that gets no replies averages 3.9 likes, while a note that pulls just two replies averages 19.5. Replies are most of the reach, rather than a bonus sitting on top of it.

First, the part I cannot fix for you

Some notes get nothing because they are bad. Some get nothing because eleven people follow the account. Those two look identical in a spreadsheet, and I want to be honest about how much of the second one is in these numbers.

To separate them I used a panel of 510 authors who have private metrics connected, 30,179 notes between them, median 27 notes each in July. Because I can see many notes from the same author, I can ask what a given choice does inside one person’s own feed, holding their audience constant.

The first thing that panel says is humbling. 64.2% of the variation in likes (on a log scale) is between authors, not between notes. Roughly two thirds of the outcome is decided before you write a word.

It gets more specific. Of the zero-like notes in that panel, 76.3% came from authors whose median note gets one like or fewer. Only 2.1% came from authors whose median note gets five or more. If you are getting zeros regularly, the most likely explanation is not your line breaks.

So from here on, when I say a feature costs you, I mean this: among notes by the same author, in the same month, the ones carrying that feature landed lower in that author’s own July ranking. I will give you the population number too, because the population number is what most people quote, and sometimes the two disagree in ways worth knowing about.

Failure mode 1: the link, and the ask stapled to it

Read more

Your weekly digest of the most relevant data, giving you insights NOBODY else has.

I thought about writing a full article, explaining what made those templates be THE templates.

But I figured you’d rather just have the templates and start collecting emails.

If you don’t like it and prefer a full analysis, let me know (reply to this email) and I’ll do it for August.

Read more

You don’t need more Substack advice. You need Substack data.

Everyone on the platform is writing, posting notes, and guessing what works. Most of that guessing is based on personal anecdotes or recycled tips from other newsletters. None of it is based on what’s actually happening across millions of creators in real time.

That’s the gap. The Substack Data Report closes it.

I analyze tens of millions of notes and millions of Substack articles every week, the only dataset of this kind, and translate the patterns into a digest you can actually act on.

“Without data you’re just another person with an opinion. With it, you’re the only one in the room who actually knows what’s going on.”

— What every serious Substack creator eventually figures out

What subscribers get

📊 Your weekly digest

The occasional, data-backed breakdown of what’s working on Substack right now post. Patterns from notes, growth signals, what the algorithm is surfacing. You’ll scan it in 3 minutes and be ahead of 99% of creators.

• ✓TL;DR at the top — the key insight, at a glance

• ✓Weekly data report — what the algorithm is rewarding right now

• ✓The patterns everyone else is missing

Subscribe now

Get ahead of other creators, learn the secrets behind Notes and collect emails faster than you ever did before.

That’s what this newsletter exists to do. Every week, I’ll send you the data. You bring the words.

See you in your inbox.

— Orel

P.S.

Knowing the patterns is only half the job. The other half is executing on them consistently. That's exactly why I built WriteStack: a tool that helps Substack creators schedule Notes at the right time, track what's performing, and grow their list without the manual grind. Think of it this way: this newsletter gives you the intel. WriteStack helps you act on it.

👉 Try WriteStack Free

Four hours later the note has two likes and nobody has replied. The note you wrote the day before, the one about nothing in particular, has forty.

Every writer on Substack has run this experiment by accident. Most of us walked away with a vague superstition: links get punished. What almost nobody has is the number. How much reach does a URL actually cost, and does the thing you paid for (a reader who subscribes) ever show up on the other side?

I went looking for both halves of that trade, because the answer turned out to point in two directions at once. And I asked two questions:

1. What does a link cost you in reach?

2. What does it buy you in subscribers?

Not here for the nitty gritty? Skip straight to the actionable part at the bottom. Everything worth doing about links is in one block down there.

Finding 1: A link cuts your median likes in half and doubles your chance of getting zero

Notes without a link had a median of 4 likes, and notes with a link had a median of 2. The average fell from 19.4 to 10.7, a drop of 44% across 655,056 notes. That is the headline, and it is not subtl

Average likes by what is in the note

The zero-like number is the one that should worry you. Nearly three in ten link-carrying notes finished July with no likes at all, against fewer than one and a half in ten for notes without a URL. A note with no likes is a note that never entered anyone’s feed beyond the people already following you.

Replies collapse harder than likes. Average replies fell 54%, from 1.53 to 0.70. And when the link points at substack.com specifically, replies fall to 0.49 and restacks fall from 2.11 to 1.29, a 39% drop in the single behaviour that puts your note in front of strangers.

Mechanism

A note with a link asks the reader to leave. A note without one asks the reader to stay. Everything the feed rewards (dwell time, likes, replies, restacks) happens on the surface of the note, and a URL is an exit sign painted on that surface. The reader who clicks is gone before they can like, and the reader who does not click has already been told the note is an advertisement.

Restacks fall hardest for substack.com links because restacking someone else’s promotional note is the least flattering thing you can do with your own feed.

The rule

Budget for it. If you publish four notes a week and one carries a link, you have spent roughly half of that note’s reach before you wrote a word. That is fine when the link is the point. It is expensive when the link was an afterthought.

Finding 2: Length does not explain the penalty

The obvious objection to Finding 1 is that link notes are simply longer. They average 435 characters against 286 for link-free notes, and long notes have their own reach problems. So I ran the comparison inside fixed length bands, where both groups are writing at the same size.

The penalty survives in every band, and standardising for length makes it slightly worse, not better. Unadjusted, link notes get 54.4% of the likes that link-free notes get. Reweighting link notes to match the length distribution of link-free notes moves that to 48.9%.

Average likes at the same note length, with and without a link

The median is 2 for link notes in four of the five bands and 1 in the shortest one. The link-free median never drops below 3 and sits at 4 in the middle of the range. I ran the same test on the separate 29,993-note conversion sample, which is not stratified at all, and got the same direction: unadjusted ratio 0.35, length-standardised 0.31.

The one thing length does explain is where the damage is worst. A short note with a link is the worst combination in the dataset, with 40.7% of them finishing with zero likes. A bare URL under 150 characters reads as a drive-by.

Finding 3: Opening with the link costs more than the link itself

Where the URL sits inside the note turns out to matter more than whether it is there. I measured how many characters of text come before the first URL, across every link-carrying note in the sample.

Notes that open with the link averaged 3.1 likes. Notes with 120 to 299 characters of setup before the link averaged 14.7. That is a 4.7x gap between two groups that both contain a link.

Average likes by how much you say before the URL

The top decile of link-carrying notes confirms it from the other end. Sorting all link notes by likes, the best 10% open with the URL only 5.0% of the time, against 17.0% for the bottom 90%. They also carry more setup (an average of 339 characters before the first link, against 298) and are less likely to be pointing at substack.com (34.5% against 42.5%).

Here is a clean instance. A full story, told to completion, with the link as a postscript that arrives after the reader already got the whole thing for free.

1,334 likes, 44 replies, 1,095 characters, with the URL as the final line.

Mechanism

A URL in the first line tells the reader what the note is for before the note has earned anything. The reader classifies it as an ad and scrolls. A URL in the last line arrives after the reader has already spent attention, and by then the note has done its job whether or not anyone clicks.

This also explains why the middle band beats the long one. Past roughly 300 characters of setup, the note starts competing with the post it is advertising, and the reader has no reason left to leave.

The template

Give the reader a complete unit of value, then the link, in that order. Target 120 to 300 characters before the URL. Never put the URL on line one.

Subscribe now

Finding 4: Every like a Substack link earns is 2.96x more likely to become a subscriber

Now the other direction. Across 29,993 notes from 657 authors with analytics attached, I compared free subscribers gained per note and per like.

Link-free notes converted at 2.22 subscribers per 100 likes. Notes with a Substack link converted at 6.56. That is 2.96 times the efficiency. What the link changes is what happens after the like, by giving the reader who already stopped scrolling somewhere to go.

Free subscribers per 100 likes, by link type

Read the middle column before you get excited. Subscribers per note actually fell, from 0.321 to 0.300, a drop of 7%. The efficiency gain is real, and it is almost exactly cancelled by the reach loss that produced it. A Substack link buys you a much better conversion rate on a much smaller audience, and the two effects land within a rounding error of each other.

The external-link row is what settles the argument. Same reach penalty (3.64 average likes against 3.75), but 0.147 subscribers per note, less than half of what a Substack link returns, and a 92.4% chance of producing no subscribers at all. If you are going to pay the reach tax, the link had better point at your own publication. Sending your feed to a YouTube video or a news article costs you the same audience and returns you half the readers.

Mechanism

A substack.com URL lands the reader on a page with a subscribe button above the fold and a one-click signup, because they are already logged in. An external URL lands them on someone else’s website, where the best case is that they come back later and remember who sent them. The conversion machinery only exists on one side of that door.

Finding 5: The reach cost is near universal. The subscriber gain is a coin flip

Population averages hide author quality. Maybe the writers who post links are simply smaller writers. So I looked inside individual authors, comparing each writer’s own link notes against their own link-free notes, using the 147 authors who published at least five of each in July.

127 of 147 authors, 86.4%, got fewer average likes on their own link notes. The median within-author ratio was 0.58, almost exactly the population figure. So the reach tax survives once author quality is held constant, which means it happens to nearly everyone, including you.

On subscribers, 71 of 147, 48.3%, did worse with links. That is a coin flip.

Share of authors whose own link notes underperformed their own link-free notes

Efficiency splits the same way. Link notes were more efficient per like for 74 of 145 authors, which is 51%. The 2.96x figure from Finding 4 is an arithmetic fact about the whole population, and it is driven by a small number of authors whose linked notes convert extremely well. For the typical writer it is close to even.

Finding 6: Naming the post without linking it keeps 3.3x the reach

There is a third option that most writers ignore. You can point at your work without pasting the URL. I flagged every link-free note in the conversion set that referenced the author’s own post, essay, newsletter, or publication in words alone, without a URL anywhere in the body. That is 1,001 notes from 272 authors.

Those notes averaged 12.52 likes against 3.75 for notes carrying a Substack link, a 3.3x difference, and they beat the overall link-free average of 10.71 as well. Median 4 likes, median length 345 characters. Saying “my latest post” costs you nothing. Pasting the URL costs you two thirds of the room.

Average likes: pointing at your work, versus linking to it

The reach effect holds inside authors too. Among the 40 writers who published at least three of each kind, 34 of 40 got more likes from their unlinked mentions than from their linked notes.

Be careful with the subscriber side of this one. The mention group averaged 0.563 free subscribers per note, above both the link-free baseline and the Substack-link group, but that figure is carried by the tail: five authors account for 47% of the subscribers in the group, and dropping the top three still leaves 0.379. Inside those same 40 paired authors, their linked notes converted better per note than their unlinked mentions (0.509 against 0.316). The mention is a reach instrument, not a replacement for the link.

244 likes, 160 characters, no URL. The publication is named in the last six words.

Mechanism

The feed appears to price the URL, not the intent. A note that says “recipes are on my Substack” is still a promotion, and readers know it, but it contains no exit and nothing to classify as an ad at a glance. The reader who wants the recipes has to go find you, which is a filter, and the readers who do not want them still liked the joke.

The payoff: when to pay the link tax

Here is the whole study compressed into a decision you can make in ten seconds before you hit publish.

The three numbers to remember

• A link halves your median likes (4 to 2) and doubles your zero-like rate (14.6% to 29.0%). Length does not explain it.

• A Substack link converts 2.96x better per like (6.56 against 2.22 subscribers per 100 likes) and still ends up 7% behind on subscribers per note, because the audience it converts is so much smaller.

• An external link costs the same reach and returns 0.147 subscribers per note, less than half what a Substack link returns. That one is rarely worth it.

The link tax is real, it is roughly 50% of your reach, and it is charged whether or not the link works. Pay it when the click is the entire point of the note. The rest of the week, say the name and let people come find you.

TL;DR

1. Cut the URL when you want the note to travel. A link drops the median from 4 likes to 2 and doubles the odds of finishing with nothing, 14.6% to 29.0%, and controlling for length makes that worse rather than better.

2. Name the post in plain words instead of pasting the URL. Notes that referenced the writer’s own work with no URL anywhere averaged 12.52 likes against 3.75 for notes carrying a Substack link, 3.3x the reach.

3. When you do link, give the reader 120 to 300 characters first and put the URL on the last line. Notes that open with the URL averaged 3.1 likes against 14.7 for notes with 120 to 299 characters of setup ahead of it.

4. Point the link at your own Substack, not at anyone else’s site. An external link costs the same reach and returns 0.147 free subscribers per note, less than half of the Substack link’s return, with a 92.4% chance of producing no subscribers at all.

5. Never attach a URL to a note under 150 characters. 40.7% of short link notes finished July with zero likes, the worst combination in the dataset.

Author note:

Hey everybody. Orel here. 👋

Every now and then I see creators assuming that Substack is not what it used to be, and that it’s not growing anymore, a waste of time, yada yada yada..

So I decided to get Claude to dive into the data and see if that’s really the case.

I wasn’t surprised.

This is one of the very few articles that’s free for everybody. Enjoy.

Every few weeks the same conversation resurfaces in my feed. Someone posts that Substack is saturated and that whoever didn’t start in 2023 already missed the gold rush. The post gets a few hundred likes, a dozen writers quietly lose motivation, and nobody checks a single number.

I got tired of watching people make publishing decisions based on vibes. So I pulled every post and every Note in the WriteStack database, 2.9 million posts and 14.5 million Notes spanning January 2023 through June 2026, and measured the thing directly.

What I found is stranger than either the doomers or the cheerleaders claim. Substack’s two surfaces, long-form Posts and the Notes feed, are moving in opposite directions. On one of them, reader attention is growing faster than the content. On the other, content is growing faster than the attention.

Finding 1: Content supply grew roughly 12x on both surfaces

Start with the raw output of the platform. In January 2023, the sample contains about 13,200 posts for the month. In June 2026, it contains 153,000. Notes went from 68,000 in their launch month to 831,000 in June 2026. Both surfaces grew roughly twelvefold in three and a half years, and neither has flattened: post volume grew another 49% in the first half of 2026 versus the same period in 2025, and Notes volume grew 47%.

So anyone telling you Substack is shrinking is simply wrong on the supply side. The feed you scroll and the inbox you compete in both carry more content than at any point in the platform’s history.

Mechanism: supply growth alone says nothing about your odds. The question that decides whether Substack is “saturated” is whether reader attention grew faster or slower than the content did. The next three findings answer that, one surface at a time.

Finding 2: Engagement per post tripled anyway

If supply grew 12x, engagement per post should have collapsed. It did the opposite. The median post earned 2 reactions in early 2023 and earns 6 in 2026. The average went from 8.3 to 26.8. Comments per post rose from 2.8 to 5.0.

The half-year comparison makes it sharper. In H1 2026, total post reactions grew 80% year over year while post volume grew 49%. On the long-form side, reader attention is growing faster than content supply. Substack is adding readers faster than its writers can feed them.

Mechanism: a reader who subscribes to newsletters compounds. They keep receiving and reacting, and their attention stacks across every writer they follow. A feed scroller’s attention is zero-sum per session. Long-form demand grows with the reader base; feed demand grows only with session time.

Finding 3: The Notes feed crossed into saturation in mid-2025

Notes tells the mirror-image story. Supply kept compounding: 831,000 Notes in June 2026, up 61% from the 516,000 of a year earlier, with no sign of flattening.

Attention is not keeping up. Average likes per Note peaked at 22.5 in June 2025 and fell to 18.8 by June 2026, a 16% decline over the same twelve months that volume grew 61%. In half-year terms: Notes volume grew 47% in H1 2026, but total Note likes grew only 36%. On Notes, supply is now outrunning attention. The feed crossed from underpriced to crowded somewhere around mid-2025.

Mechanism: feed attention is bounded by session time, and session time grows much slower than posting volume. When 800,000 notes a month compete for a feed with finite scroll depth, the average note must earn less. This is the same curve Twitter and LinkedIn rode years earlier.

Finding 4: On Notes, the jackpot is shrinking and the floor is rising

Inside the declining Notes average hides the most actionable pattern in the dataset. While the average likes per Note fell from 22.5 to 18.8, the median went up, from 2 in 2024 to 3 and then 4 by spring 2026. The average is dragged by outliers; the median is what the typical note earns. They are moving in opposite directions.

The ratio between them measures how concentrated attention is. In June 2025 the average Note earned 7.5x the median. By June 2026, 6.3x. Attention on Notes is spreading more evenly: viral outliers capture a shrinking share while the typical note earns more than ever.

Mechanism: as a feed algorithm matures, it optimizes for session retention rather than discovery shock, which spreads impressions across more creators and dampens runaway virality. The lottery ticket gets worse. The paycheck gets better.

So, is Substack still growing?

Readers, yes, and fast: total reactions grew 80% year over year on posts and 36% on Notes. Content, also yes: both surfaces grew roughly twelvefold since 2023. What decides whether you feel growth or saturation is the exchange rate between the two. On long-form, attention is growing faster than content, so each post earns more every year. On Notes, content is growing faster than attention, so each note earns less.

The honest one-line answer: Substack is still growing, and the returns have moved to long-form. For anyone publishing a newsletter, that is close to the best possible market structure.

The playbook this data implies

1. Shift weight toward long-form. Post volume grew 49% last year while post reactions grew 80%. Attention is outrunning content there, and every additional post captures part of the surplus.

2. Use Notes for discovery, not as the destination. Per-note attention is deflating 16% a year. Every note should have a job: pull a reader toward the asset that compounds.

3. Publish on a fixed cadence. The rising Notes floor and the tripling post engagement reward the writer who shows up every week over the one hunting a single big hit.

4. Ignore the saturation discourse. Saturation is real on the Notes feed and absent where subscribers actually live. The writers who quit because “it’s too late” are handing you their readers.

P.S. I left a present under this article

The data says the whole surplus goes to writers who ship articles on a cadence. So I wrapped the shortcut. Inside:

• A calendar that arrives already full, a month of article ideas planned ahead

• Every idea one click away from a draft

• Swap ideas you don’t like, drag articles to days that fit

Come unwrap the Articles Planner inside WriteStack. The data already told you to open it.

I’ve told readers of this newsletter that 2 to 3 notes a day is the sweet spot. I said it in at least two prior articles. I believed it because it felt right, not because I’d checked it. Other creators swear by 10 a day and treat anyone posting less as under-committed. Nobody I’ve read has actually opened the data and measured what a fourth, eighth, or twelfth note in the same day is worth, compared to the first.

So I pulled it.

One more number before the findings. 72% of all posting days in this dataset land at 1 to 3 notes. Another 21.7% land at 4 to 7. Only 6.3% of days cross into 8 or more. Most creators, whether they've thought it through or not, are already behaving like the "2 to 3 is the sweet spot" line I wrote is true. The question this article actually answers is whether that instinct is correct, or whether it's just where most people's energy happens to run out.

Finding 1: Your Fourth Note Is Not What You’d Expect

For every author-day, I ranked their Notes in posting order and averaged free subscribers gained by rank. The pattern is not subtle, and it does not land where the “2 to 3 is the sweet spot” line would predict.

Read more

The PDF below has all fifteen. No examples, no stats — just the structure, how to use it, and a fill-in-the-blank template you can plug directly into an LLM or use as a writing prompt.

Enjoy! :)

Read more

The data says that’s exactly the wrong move.

I pulled every note in WriteStack’s database — 18,868,307 of them, going back to April 2023 — and looked at exactly when they were posted and how much engagement each one got. The pattern is consistent enough across four years to change how you think about scheduling.

There’s one window in the hourly data that nearly nobody posts in, and it has some of the highest engagement in the entire dataset. Read through to find it.

The Best Days To Post Notes

Peak Performance Windows

Sunday leads with 421.6 avg reactions per note — the highest of any day. Saturday is second at 401.7. Then comes a clear drop: Monday sits at the bottom with 338.2, despite having significantly higher posting volume than either weekend day.

That gap between Sunday and Monday is 25%. A note posted Sunday gets roughly one quarter more reactions on average than the same note posted Monday. Not because Sunday audiences are different people — because Sunday has fewer notes competing for the same attention.

Why this happens

Sunday has 2.28M notes in the dataset. Tuesday has 2.89M — 27% more. Reader attention doesn’t scale up 27% to match. It stays roughly constant while supply floods the feed. Each individual note gets squeezed. Weekends reverse that: fewer creators post, the feed is quieter, and each note gets more room.

Notes With Under 26 Likes

This is the downside chart. Saturday and Sunday have the lowest count of underperforming notes — the share of notes that die below 26 likes is meaningfully smaller on weekends. Tuesday through Thursday have the worst ratio: highest volume, highest share of low-engagement notes.

Posting on a weekday doesn’t just lower your ceiling. It raises your floor for failure.

Subscribe now

The Best Hours To Post Notes

Day of week matters, but hour of day is where the real leverage is. I looked at every day × hour combination across the full dataset and measured average reactions per note for each of the 168 cells.

Two things jump out immediately.

The early morning window (3–8 AM ET) is consistently strong across almost every day. This is the counterintuitive finding. These hours have some of the lowest posting volume in the entire dataset — which means any note that goes out faces almost no competition and gets disproportionate reach from readers who are up early checking their feeds.

Sunday 6 AM ET is the single highest-engagement cell in the dataset at 929 avg reactions per note. Almost nobody posts then. The notes that do go out essentially have the entire feed to themselves.

The midday weekday window (11 AM – 3 PM ET) is where engagement goes to die. This is exactly when most creators post — lunch breaks, mid-morning habits, “best time to post” cargo-culting. The data says it’s the worst possible window: maximum competition, minimum engagement per note.

The finding that surprised me most: Friday at noon (12 PM ET) has the lowest avg reactions of any cell in the entire dataset — 125 per note. The same note posted Friday at 6 AM gets 528. That’s a 4x difference on the same day, just by shifting a few hours earlier. Timing isn’t a marginal improvement. It’s a multiplier.

Strategic Timing Recommendations

Volume of Notes

The volume map is nearly the inverse of the engagement map. The darkest cells — peak posting — fall on weekday afternoons and evenings, 5–9 PM ET Tuesday through Thursday. That’s exactly when engagement per note is weakest.

The lightest cells — lowest posting volume — are weekend and weekday mornings before 9 AM ET. That’s precisely when engagement is strongest. Less supply, same reader attention, more reach per note.

Supply and demand, literally

At peak volume (Tuesday 7 PM ET), over 200,000 notes go out in a single hour. At the lowest-volume window (Sunday 10 AM ET), fewer than 47,000 do. Reader attention doesn’t scale 4x to match. The 47,000 notes get proportionally more of it.

How Long Should Your Notes Be

Timing gets you in front of readers. Length determines whether they stop or scroll. Across 18.8M notes, the word count data tells a clear story — with one result that flips what most people assume.

The sweet spot is the 33–64 word range, which delivers the highest avg reactions at 449 — not the ultra-short posts everyone assumes win. A 40–60 word note is long enough to make a real point, short enough to read in one breath. It’s the format that readers will actually finish.

Very short notes (under 16 words) actually underperform relative to the 33–64 range. At that length there’s often not enough of a hook or idea to make someone stop scrolling. They register as filler.

The 65–256 word range is the dead zone — engagement drops sharply. Long enough to feel like work, not long enough to be a real piece of writing. This is where “I’ll write something medium-length” notes end up, and it’s where engagement goes flat.

Notes above 257 words recover and actually get strong engagement (243–344 avg reactions), but the volume is tiny — only 345K notes in the dataset go that long. The readers who commit to a long note are self-selected and highly engaged.

The practical rule: Aim for 33–64 words as your default. That’s one strong thought, fully developed, with enough words to land but not enough to lose anyone. If you’re going longer, commit past 256 words — you’re writing a real piece, and the readers who stick around are your best ones.

Final Thought

Eighteen million notes tell a consistent story. Most creators are posting at the wrong time, in the wrong length. They post on Tuesday afternoon because that’s when they finish writing. They write 80-word notes because it feels like a reasonable middle ground. Both choices work against them.

The data is straightforward: post on Sunday morning or late evening. If you post on weekdays, go early (before 9 AM ET) — not at lunch. Write either a tight 40–60 word note that makes one real point, or go long enough (257+ words) to actually develop an idea. The middle — 65 to 256 words — is where engagement falls off a cliff.

Three changes. The notes that follow these rules outperform the ones that don’t, across four years of data and nearly 19 million notes.

P.S.

Each audience is active on different times. This is a general rule of thumb.

WriteStack tells you EXACTLY when your audience are most active, so you can post at your ideal times. All the time.

And you can use it today for free! Get your 7-day free trial now:

Try WriteStack For Free 👉

If you sign up for an annual plan in the next 48 hours I’ll set up a 1on1 call with you to plan your Substack growth and build a plan.

Hey everybody! 👋

This article is a part of a series of articles where I share the knowledge I have based on an analysis of millions of Notes.

Since I don’t have the capacity to write these on my own, I feed Claude with the data and it writes the articles.

Enjoy!

106,062 public Substack posts from the last 30 days. One structural pattern averaged 44% more reactions than the baseline. I filtered the dataset before touching it — excluding news, politics, sports, investing, celebrity, and religion. Topics that pull event-driven engagement from audiences who have no interest in the author’s work distort the signal. What remained was 106,062 posts where the title had to do the actual job. The winner wasn’t the colon. It wasn’t the numbered list. It was the first-person opener, and the explanatory frame right behind it.

The structural breakdown

Five structural patterns, tested against reaction count across 106,062 posts. Every number below comes from titles that had to earn the click on their own — no news cycle doing the work for them.

Read more

Hey everybody! 👋

This article is a part of a series of articles where I share the knowledge I have based on an analysis of millions of Notes.

Since I don’t have the capacity to write these on my own, I feed Claude with the data and it writes the articles.

Enjoy!

42.1% of viral Substack Notes earned zero subscribers last month. That’s not the surprising part. The surprising part is what the other 57.9% have structurally in common, and it has almost nothing to do with the topic they covered.

In May 2026, I tracked 35,845 Notes from WriteStack users with conversion data attached. Of those, 1,057 crossed the viral floor of 50+ likes. Nearly half of them got nothing back in return.

Viral Notes that converted zero subscribers

Last month’s figure (April 2026, 20,066 Notes): 40%. The rate ticked up two points. That’s not noise, but it’s not a crisis either. What matters is the pattern underneath it, which holds across both months.

What counts as viral here

The viral floor this month is 50 likes, the 92nd percentile of all Notes in the dataset. One in twelve Notes crossed it. I use likes as the primary signal because they’re the fastest feedback loop on the platform, and because they’re the number creators most commonly optimize for. The conversion data comes from WriteStack’s tracking layer, which records free subscriber gains per Note. All conversion figures are aggregated across cohorts. No individual author’s numbers appear below.

Finding 1: The conversion cliff

The tier breakdown is the article’s central finding. Here it is in full:

The cliff between the viral floor tier (50–99 likes) and the mid tier (100–249 likes) is 22 percentage points. Between mid and high (250–999 likes) it’s another 9 points. Get a Note above 250 likes and the odds flip: only 1 in 10 converts nothing. Below 100 likes, it’s closer to 1 in 2.

The mechanism: A like is a 2-second reflex. You see something that makes you feel something, your thumb moves. A subscription is a commitment to receive future emails from a stranger. Those two actions don’t share a decision pathway. Likes measure resonance. Subscriptions measure trust in an ongoing voice. A viral Note can generate one without the other, and often does.

The average free subscribers earned tells the same story from the other direction. Mega-viral Notes (1000+ likes) averaged 15 new subscribers. High-tier Notes averaged 16.98. The mid-tier dropped to 4.84. The viral floor tier averaged 1.87. Pre-viral: 0.82. Below 20 likes: 0.12.

One number stands out: the High tier (250–999 likes) averaged more subscribers per Note than the Mega tier (1000+ likes). That’s the outlier worth examining. The sample is small (11 Mega Notes vs. 123 High), but the gap is consistent with something structural: Notes that reach 1000+ likes often do so because they’re maximally shareable across audiences who don’t follow the author. High-tier Notes reach a threshold where the author’s existing audience drives the shares, meaning the people restacking already trust the author and are more likely to recommend them to people who’d convert.

Finding 2: What the zero-sub Notes have in common

I read the 50 highest-liked Notes that converted zero subscribers. Three structural patterns appear in almost all of them.

Pattern A: The complete delivery

The Note hands over its entire value in the feed. There is nothing left to follow. The reader has consumed the whole thing, liked it, and moved on. Median word count for zero-sub viral Notes in this cohort: 17 words. Several of the highest-liked ones were under 10 words, or consisted entirely of a photo caption with no author identity attached.

Pattern B: The tribal signal

Several high-liked, zero-sub Notes are essentially agreement objects. Political hot takes, gender commentary, cultural in-group signals. They attract mass likes from people who share the position, but those likes come from a diffuse audience with no shared interest in the author’s specific work. The Note does not introduce a voice. It reinforces a feeling.

Pattern C: Inspirational filler

Short, quotable, completely generic. “Progress over perfection.” “Get physically fit. Get mentally calm. Get financially free.” These Notes collect likes because they’re easy to agree with, not because they make a reader curious about the person who wrote them. They’re the Substack equivalent of a motivational poster: nobody buys the poster to find out more about the designer.

Finding 3: What the converters do differently

Read more

Hey everybody! 👋

This article is part of a series where I share what I’ve found across millions of Substack Notes. I don’t have the capacity to write these on my own, so I feed Claude the raw data and it writes the articles. Enjoy.

I analyzed 789,362 Substack Notes from April. Not a sample. The full 30-day window, every original Note, filtered to exclude restacks. One question: does the hour you post actually change how many reactions you get, or is that just scheduling superstition?

It changes. The best-performing hour outperformed the worst by 82%. There’s a specific 3-hour window that looks like a logical time to post, catches a plausible audience, and consistently produces the lowest engagement on the platform, 30% below average. The winning window, the dead zone to avoid, and a weekly posting grid built from the data are all below.

The Hour Curve

Here’s the graph that shows the best times to post:

The curve is asymmetric. It climbs from a trough between 04:00 and 09:00 UTC, stabilizes in the afternoon, then surges from 19:00 onward. The 19:00–23:00 UTC window averaged 23.2 reactions per Note, 16% above the platform average. All five hours in that window carried a median of 4 reactions. The dead zone hours dropped to a median of 3.

Two windows overlap here. Hours 19:00–23:00 UTC represent the US afternoon and early evening (EST: 2pm–6pm, PST: 11am–3pm), combined with European prime-time browsing (CET: 8pm–midnight). Two large reader audiences simultaneously active. That overlap is almost certainly why the evening surge is so consistent. Hours 00:00–02:00 UTC also perform well (avg 22.5–24.1) for a related reason: that’s 7pm–9pm US East Coast, when feed activity peaks with lower posting competition.

Volume note

No hour in this dataset is thin. The lowest-volume slot (04:00 UTC) still had 14,990 Notes. Every number in this article is statistically reliable.

Day Breakdown

Weekend Notes slightly outperformed weekdays in April (20.25 avg vs 19.86). Saturday and Sunday carry roughly 25% fewer Notes than the busiest weekdays, which means less competition for feed position. That volume gap explains most of the performance edge. Wednesday is the busiest day on the platform by Note count (143,843) and still ranks second -- which says more about Wednesday reader activity than about timing strategy.

Friday is a consistent underperformer at 19.0 avg reactions, 4.9% below the weekly average. Posting on Friday isn’t disqualifying, but it’s the one day where you’re working slightly against the trend.

The Dead Zone

Dead Zone

Posting between 06:00 and 08:00 UTC produced an average of 14.0 reactions per Note, 30% below the platform average. The worst single hour was 07:00 UTC at 13.6 avg reactions.

This is the cleanest avoidance signal in the data. No other window produces a gap this wide. The mechanism: 06:00–08:00 UTC falls between the two major audiences. It’s 1am–3am US East Coast (the primary Substack reader base is asleep) and 7am–9am CET (European readers are commuting, not engaged browsing). Notes posted in this slot sit in a quieter feed during their first few hours, and early engagement compounds -- a slow start means lower feed distribution, which means fewer reactions overall.

The counterintuitive trap: 07:00 UTC looks like a reasonable time to post if you’re thinking “European morning.” It’s not. The audience size at that hour can’t offset the lack of US afternoon engagement that Notes posted 12 hours later will capture. The avoidance is worth more than any other optimization in this article.

The Window

Recommended Window

Tuesday or Wednesday, 19:00–22:00 UTC is the clearest intersection of the top-performing days and the top-performing hours. Estimated 20–24% above the platform’s weekly baseline when day and hour effects combine.

Crossing the day rankings against the hour rankings, one window stands out: Tuesday or Wednesday, 19:00–22:00 UTC. For US creators: 2pm–5pm EST / 11am–2pm PST. For European creators: 8pm–11pm CET. This window captures the maximum audience overlap, and Tuesday and Wednesday are both top-3 days for average engagement. Sunday 19:00–22:00 UTC works equally well as a weekend alternative -- same hour range, slightly different audience mix.

One extra note for creators optimizing for virality rather than average engagement. Hours 14:00–16:00 UTC produced the most viral Notes in April (3,741 viral Notes at 14:00 UTC alone, defined as 50+ reactions). That’s the US morning posting rush (EST: 9am–11am). Average reactions in that window are near the platform median, but the absolute viral count is highest because the posting volume is highest. More shots on goal. If you’re publishing something with real breakout potential, the midday window gives you the most chances to catch a spike.

The Caveat

This is platform-wide data from 789,362 Notes. It tells you when the average Substack reader is active -- not when your readers are active. A newsletter focused on European finance professionals, Asian developers, or US night-shift workers may behave nothing like this curve. The winning window for the platform average could be the dead zone for your audience.

If you have consistent posting history, your own engagement data is more valuable than these platform numbers. Look at which of your Notes got the most reactions in the first two hours after posting. That’s your actual optimal window. Use these platform findings as a starting hypothesis, not a fixed answer. And treat the dead zone (06:00–08:00 UTC) as a strong default avoidance unless your data specifically contradicts it.

How to Use This

Platform timing data changes. The winning window from April may shift in June as posting behavior and reader habits evolve -- which is why I run this analysis monthly. The most useful thing you can do with this data isn’t to post exactly at 21:00 UTC tomorrow. It’s to build a repeatable system: batch-schedule your Notes into these windows, compare your own stats against the platform baseline, and adjust quarterly. WriteStack’s scheduling grid lets you load a week of Notes into these slots in one pass, and the Stats dashboard shows your own hour-by-hour engagement so you can compare directly.

—

p.s.

In this article you can see the best times to post, based on general analysis of hundreds of thousands of notes.

In WriteStack, you can see your own personalized best time to post graph that shows when your audience is most active and when you should post your notes.

→ Try WriteStack For Free

Hey everybody! 👋

This article is part of a series where I share what I’ve found across millions of Substack Notes. I don’t have the capacity to write these on my own, so I feed Claude the raw data and it writes the articles. Enjoy.

Every creator wants a viral Note.

The dopamine of watching the likes climb. The restacks rolling in. The notification icon that won’t sit still. It feels like growth, and for the few minutes the Note is alive in the feed, it feels like the thing finally clicked.

Then you check your subscriber count two days later and nothing has moved.

I’d noticed this pattern in my own Notes for a year. The ones that went off the rails on engagement weren’t the ones that grew me. The ones that grew me were often the quiet ones, and I couldn’t explain why.

So I pulled the data on every Substack Note in our system from November 2025 through April 2026. 243,483 Notes. Then I filtered down to the top 8.2% by likes or restacks. That’s the 20,066 viral Notes this article is built on. Real engagement. Real subscriber counts. Six months of behavior.

What I found landed harder than I expected. The number that runs the headline is real, the mechanism behind it is structural, and one of my best hypotheses got completely demolished along the way.

What “Viral” Means Here

Before the findings, the definition.

The top 8.2% of all Notes by likes start at 46 likes. The top 8.2% by restacks start at 3 restacks. A Note that crosses either bar lands in the viral cohort. That gives us 20,066 Notes, averaging 126 likes, 13 restacks, and 368 clicks each. These aren’t toy viral. These are the Notes Substack’s algorithm actually picked up and pushed.

The optimization metric for this article is free subscribers gained. Likes are public on Substack. Subscriber counts and click counts are private, which is why all author-level callouts in this article reference only public metrics, and the conversion data stays in aggregate.

Finding #1: 40% of Viral Notes Got Zero Subscribers

Read more

Hey everybody! 👋

This article is a part of a series of articles where I share the knowledge I have based on an analysis of millions of Notes.

Since I don’t have the capacity to write these on my own, I feed Claude with the data and it writes the articles.

Enjoy!

Most creators post Notes and hope for the best.

They write short. They ask questions. They use emojis. They follow the “rules” everyone on the internet keeps repeating.

And they wonder why their subscriber count barely moves.

I got tired of guessing.

So I analyzed 19,471 of the highest-converting Substack Notes ever posted — ranked by actual free subscribers gained. Not likes. Not impressions. Real people who hit Follow after reading.

One finding shocked me more than anything else.

And most creators are doing the exact opposite of what works.

What “Viral” Actually Means Here

Before we get into it, let me be specific.

These 19,471 Notes were ranked by total free subscribers gained per Note.

That gap is what I wanted to understand.

Here’s what I found.

Finding #1: Question Marks Are Killing Your Notes

Read more

That means that you no longer have to open the app and post your Note manually. You can build a queue up to 3 months in the future.

But let me tell you, they built it in a way that only the engineers who developed it can understand how to use it.

In this article, I’ll show you EXACTLY how to manage your drafts and how to schedule your Notes.

Before we dive in, if you’re serious about growing on Substack, you’ll probably hit the limits of the native scheduler pretty quickly.

It lets you schedule one Note at a time, manually, with no way to bulk import, bulk schedule, reorder your queue, or organize by topic. That’s fine for occasional notes, but not for creators who publish consistently.

That’s exactly why I built WriteStack. It’s a tool designed for Substack creators who want to schedule dozens and hundreds of Notes in seconds, triage their notifications and manage their Substack without burning out.

👉 Try WriteStack free

Laptop Scheduling

First of all, you need to create a new Note. Click the ‘What’s on your mind?’ button to start writing.

This window will open:

Once you have completed your draft, you have a 3 options:

1. Save as draft

Click Cancel or click outside the editor to pop up the save window:

Click Save to save the Note in your drafts or Discard to remove it.

Now, once you save it, you’ll be able to see all your saved drafts in the Drafts list, by clicking Drafts.

2. Schedule the Note

Now this is why we’ve gathered here. Scheduling your Notes.

To schedule the Note that you created, you can either click any of your drafts or create a new Note, like we did before.

Instead of saving it as Draft, click the Calendar icon, pick a date and time, Click Save and then click Schedule

Now go back to your drafts and you’ll be able to see your schedules. You’ll be able to see the date to which they’re scheduled underneath the Note’s body.

3. Update a draft

Go to Drafts, Click the Note draft you want to edit. Change whatever you want, then click outside or Cancel. This window will pop up

Choose to Update or cancel all changes (Don’t Update).

How to delete/edit a schedule

This is quite simple. Go to Drafts, Click the Note that you want to edit.

To update the date, simply click the Calendar, pick a different date, Save → Schedule.

Deleting a schedule, or unscheduling, is a bit more complex. Open the scheduled Note, Click the X next to the schedule at the top.

Then simply update, as I show before.

—

And that’s how you create and manage your schedules on Substack.

Mobile Scheduling

Now mobile is a whole other beast. At least on the Android version.

The logic is the same. The UI is completely different and unclear.

1. Save as draft

Click the + button to create your draft. Write whatever you want and

Then, to save the Note click on the X → Save draft.

2. Schedule the Note

In order to schedule a Note, you need to click the 3 dots on the right side → Schedule

Pick a date and time → OK → Schedule

Word of caution:
When you choose time, make sure not to scroll too fast, or the app will crash (Android).

3. Update a draft

To update a draft, click the Draft you want to edit.

If it’s a scheduled Note, you can either click the date banner

or the 3 dots → Edit Schedule to update your schedule.

Then click the X, top left, to update/cancel update.

You should do the same if you want to edit the text of the Note. Just always remember to click the X before closing the app to make sure changes are saved.

The Limits

Substack’s scheduler does what it promises, but it’s built for occasional posting, not for creators with a real publishing system.

Here’s where you’ll start feeling the friction:

One Note at a time. There’s no way to bulk import or schedule multiple Notes in one sitting. Want to plan a week of content? Prepare at least 30-40 minutes to schedule.

No queue management. Once your Notes are scheduled, you can’t reorder them. Change your mind about the sequence? Unschedule, reschedule, repeat.

No organization. No tags, no categories, no evergreen queues. Everything lives in one flat draft list.

No performance visibility. You can schedule Notes, but Substack won’t tell you which ones actually converted subscribers, what time your specific audience engages most, or how your content is performing over time.

So, although it solves a HUGE problem, it’s done VERY poorly without considering the fact that the creators who use it are NOT engineers.

Now you know how to schedule Notes natively on Substack, and what its limits are. If you are a serious Substack creator, you’ll hit this wall (and many more) sooner or later.

Here's what the alternative looks like:

It’s built by a Substack creator (me), for Substack creators. You can try it and see if it fits how you work.

👉 Save Hours Managing Your Substack

Hey everybody. 👋

This is the second addition to this article, but with twice the data. If you read the previous one, feel free to skip to the data.

If you didn’t, I am going to break down the best times to post and what to post. So you’re going to find it extremely useful.

But before I do that, I want to tell you about WriteStack. This is an app that I built for Substack creators.

WriteStack will help you:

1. Schedule Notes - Build a schedule that fits the best times for you to post, so you don’t have to manually send the Notes.

2. Notes Analytics - Learn what converts for your audience and what brought you subscribers (free and paid).

3. Followers Activity Graph - See when your audience is active, based on real-time data that’s collected throughout the day, so you know the BEST times to post, even if your audience changes.

And the best part? You can try it today, for 7 days FREE.

Grab WriteStack 7 Days Free Trial 🎁

It’s not a secret anymore that most of the traffic creators get is from Notes.

30-50% of their subscribers come from Notes. For some creators, these numbers shoot up to 70-80%.

Crazy effective.

Now, in order to make it even more effective, you’re going to need to know exactly when to write them, how to write them and what makes them effective.

And I analyzed over 12,463,022 notes to figure those out.

There’s one thing that caught me by huge surprise when I discovered it, so read through to find out what it is.

Understanding the Science Behind Peak Engagement Hours

Content timing isn’t just about posting when you’re awake.

It’s about understanding when readers and creators are most receptive to your content and are willing to engage. When they are inside Substack and are willing enough to engage.

For example, if 80% of creators and readers are in America, posting a note when they’re asleep won’t get you too much exposure. Other notes will be posted at the right time to overshadow yours.

That’s why it’s so important to know the best time to post your Notes and write them in a way people want to consume.

You want to hit your readers as soon as they open the app.

The Best Days To Post Notes

Read more

My name is Orel, I am the founder of WriteStack. I built multiple newsletters to over 5,000 subscribers.

I professionally help people grow their subscribers list on Substack to 1,000 subscribers.

You’re starting from zero.

No audience. No momentum. Just you and your wish to grow on Substack. This is your plan:

1. Put out content every day

2. Reply to every comment

3. DM anybody who engage with you

4. Engage with others on purpose

5. Collaborate as much as you can

Follow this for 30 days and you’ll build visibility, trust, and a small network that turns into subscribers.

Two rules for the month

1. Notes bring reach

Notes put you in front of new people. Post them daily.

2. Articles bring subscribers

A weekly article turns attention into trust. Trust turns into email signups.

WriteStack helps with both because it lets you save ideas while you scroll and turn them into drafts later.

Subscribe now

Pillar 1: Put out content daily Notes and weekly articles

Read more