I wrote about picking a social platform by format back in 2023. The principle held up. The map did not.

Most of the platforms I named are different now. Some are dead. Some swapped their identity. One didn’t exist yet and now eats much of my attention. So this is a redraw.

Platforms

Pick the platform that fits your format. Not the one that’s trending. Not all of them.

If you write, go where writing is the point. If you talk, go where audio lives. If you film, go where video gets watched. Spreading yourself thin across six platforms is the most reliable way to be invisible on all of them.

That hasn’t aged.

If you write

In 2023 I said Twitter.

Twitter is X now, and X is a different product. The text-first culture is still there. The reach is not, for most people I know. The algorithm rewards a narrow set of voices, and small accounts feel smaller than they used to.

Substack Notes is the (new) place text-first writers actually live. It works a lot like old Twitter. It feels light(er). It also pushes you toward your newsletter, which is the point.

Bluesky exists. It’s an option.

Threads is still around, technically. I never warmed to it. But some have success there.

So: Notes first. X if you have history there. Bluesky or Threads as a backup.

If you film

TikTok survived its forced-sale year in the US. Still the format leader. Still the place a short video can go from nothing to a million views overnight. The platform US lawmakers spent three years trying to kill is bigger than ever.

Instagram Reels is a good bet for many creators, still.

YouTube Shorts is even better than Instagram, I think. It pays. It ranks in search. It feeds the long-form channel if you have or want one. If you’re going to do short video anyway, Shorts is probably the highest-ROI version of it.

If you talk

I made fun of Clubhouse in 2023. Remember that one? Audio-only social media. It’s dead.

Audio didn’t die. It went back where it was working: podcasts, audiobooks. The “social audio” wrapper was the part nobody needed or wanted.

If voice is your thing, start a podcast. Don’t wait for the next Clubhouse. It isn’t coming.

If you make images

Instagram, still. Less obvious than it was, though.

Pinterest second. SEO-driven, long-tail, kinder to small accounts. It rewards a body of work over a single viral post. That fits a lot of creators better than they realize.

If you make photographs as art, neither is great anymore. The format that platform was built for got eaten by short video.

Trying all of them

I tried many. It didn’t work.

Six platforms means six content engines, six tones, six audiences who don’t overlap as much as you’d think. The math just doesn’t work for one person.

One platform you like, plus your owned channel (blog or newsletter), beats a thin presence everywhere. By a lot.

Downsides

Picking one platform means missing the next one. Probably.

If TikTok actually gets banned somewhere it matters, that’s a problem for anyone who put everything there. If X keeps drifting, the writers who stayed lose more than they realize.

Owning your audience somewhere a platform can’t take from you is a great idea. Email is still the best version of that.

The Bottom Line

The map in 2026 is quite different from the one in 2023. Clubhouse is gone. Twitter is X. Notes is the writer hangout. Threads are somewhat flatlining. TikTok is strong. Pinterest is underrated.

The principle holds up to this day, I believe: Pick what fits. Make the good stuff there. Send people from there to a list you own.

That’s been the play the whole time.

I publish on Substack. Almost daily. It’s a fantastic platform. It’s got a lot going for it. And mostly for free. But I like Medium more, and the gap is getting bigger.

Substack Notes ate the app

Notes has gotten huge. Good for Substack. Less good for me.

Every update pushes the app a little further toward social media and a little further from writing, I feel like. Many people enjoy this. I don’t really.

New feed tweaks. New video stuff. New ways to “engage.” Like another Twitter, just with longer captions.

I open the app to write. I close it because I scrolled.

That’s not Substack’s fault, really. They’re chasing growth, and Notes is where the growth is. But it changes what the platform feels like. And what it feels like to me right now is a social app with a newsletter and blog bolted on.

The email part isn’t enough for me

Substack’s email tools are fine for a free newsletter. They’re not enough for the way I work.

Most of my email workflows live in Gumroad. Product delivery, follow-ups, segmentation by what someone bought. That’s where list segmentation does the real work. Substack handles the public newsletter. But it’s basically just a sendout of a blog post. Said it before. Substack is more a blog with email attached than a “newsletter” tool.

So Substack’s email side has become the lighter half of my setup. The newsletter still goes out. But it’s not the engine. For anything.

My main driver of Substack growth is SEO. And that’s doing pretty well right now. It’s the reason I still have Substack.

Medium is my main thing

Medium pays me. Quite well most months.

Medium also ranks on Google. Medium has a custom domain that I own at stories.byburk.net. (Substack has that too, though). The reading experience is calm. There is no Notes feed pulling at me. Sure, it’s algorithmic. What isn’t? But it’s not social media, in my eyes. Or less than Substack, at least.

I do most of my writing for Medium first. Substack gets the same article, later. That’s the order, and it tells you what I treat as primary.

Substack has become a second-class citizen for me. And… I treat it like one.

The one thing that keeps me there

I said it. SEO.

The SEO bump on letters.byburk.net over the last months has been really nice. Search traffic is up. Old posts are getting found. The custom domain is doing its job.

I wrote a whole guide about how I do this in Substack SEO. And it’s starting to pay off.

So I’m not pulling the plug. The archive earns its keep, even if I’m not in love with the app (right now).

Will They?

Substack could pull me back in. Probably not by adding more social features. More like the opposite.

Better long-form tools. Better editor. Better post management. Less feed in my face when I open the app to draft.

Could happen. I doubt it.

This is me, not Substack

None of this means Substack isn’t great. It is.

For a beginner writer, Substack is probably the best place to start right now. One platform, free, easy, with a built-in audience graph through Notes and recommendations. For an established writer who wants a newsletter, a blog, and a social feed in one app, it’s basically perfect.

I’d encourage anyone to try Substack. And Medium. Both, ideally.

But if it’s not for you, there’s no shame in not doing it. Platforms are tools. Some fit, some don’t. Mine just fits a little less than it used to.

The Bottom Line

I’m not quitting Substack today. I’m just thinking about where it sits in my stack.

Medium first. Gumroad for email that does work. Substack for the SEO archive and the newsletter that goes out anyway.

Notes is the headline feature for Substack right now. It’s not mine.

Most Substack publications don’t show up on Google. Substack is not bad at SEO. It isn’t great either, but’s it’s very doable.

It’s just the missing setup. Or the wrong one. Or it never really happened.

I spent way too long writing before I started figuring out what the foundation should look like when you want your Substack to show up on Google and attract search traffic. Once I had it, search traffic not only started showing up, but grew consistently and became my #1 source or Substack growth.

This is the setup. It takes about half an hour. You only do it once.

Custom domain first

This is the one I’d put before anything else. Substack lets you connect your own domain for $50, one time. Mine is letters.byburk.net.

The default yourname.substack.com address is shared with everyone else’s reputation. Yours is just yours. Google rewards that over time.

The other reason: SEO traffic on your own domain travels with you. Move off Substack one day, your rankings come too.

$50, once. Totally worth it.

Google Search Console, before you publish anything else

Go to search.google.com/search-console. Add your custom domain as a property. Verify it. Submit your sitemap.

Your sitemap lives at yourdomain.com/sitemap.xml. Just paste that URL into Search Console.

Without this step, Google might still find you eventually. Or not. With it, you’re telling Google “here is everything, please come look.”

It also shows you what people are searching for to land on your pieces. Which is useful for the next thing you write.

Check that you’re not blocked

Open yourdomain.com/robots.txt in a browser. Make sure it doesn’t say Disallow: /. Substack handles this fine by default, but worth checking once.

Then go to Substack settings, Privacy, and confirm your publication isn’t set to private or invite-only. Sounds obvious. People still mess this up.

Fix your About page

Write something other than the default text that’s there. Many don’t do that.

Yours should answer two questions: what you write about, and why someone should subscribe. Use the words people would search for.

If you write about Substack growth, the words “Substack” and “growth” should be in there. Not stuffed. Just present. Same for whatever your topic is.

Use sections if you write multiple different topics

Substack lets you create sections. It’s a bit messy, and they could do it a lot better. But still, sections are a great asset if you have a diverse set of topics you write about that don’t really fit one broader category.

That being said, I don’t do that. Mostly because I hadn’t done it from the start, and then it was too much manual work for me to do it later.

Why sections? They give Google another way to understand what your publication is about. Each section becomes its own indexable page with its own articles grouped under it.

Three or four sections is plenty. Don’t go wild.

Publish-time checklist

Every time I hit publish, I go through the same four things.

Title. Includes the actual words people would search for. “How to Set Up Substack for Google Search“. Not religiously. But I give it a second thought.

URL slug. Substack auto-generates a slug from your title. Sometimes it’s bad. Edit it. Keep it short, keep the keywords, drop fillers.

Subtitle / SEO description. Substack uses this for the meta description Google shows under your title. One sentence. Says what the piece is about.

Cover image. Add one. Substack uses it for OpenGraph and Google Discover. A piece without a cover image looks half-broken in search results.

That’s it. Maybe two minutes per article.

Link your old pieces to your new ones

Every new article gets one or two links to older relevant pieces. Then I open those older pieces and add a link back to the new one.

It’s a little bit of busywork. Five minutes per piece. A great use case for AI, by the way.

This is what tells Google your archive is alive. Connected. Worth crawling. A pile of unconnected articles looks dead.

I went deeper on the writing side of this in how I do Substack SEO. The setup above is the foundation.

What ranks

Once the setup is in place, certain pieces climb and others don’t.

Comparisons may rank. How-tos may rank. Pricing breakdowns may rank. Anything with a year in the title may rank. Or they won’t. It’s really not guaranteed, but it’s better to try.

That being said, personal essays don’t really rank on Google and shouldn’t really. It’s not search content. You know what I mean? Opinion pieces don’t either. Anything too generic doesn’t.

You don’t have to choose, though. Mix it up. I don’t just write for SEO. I want SEO to feed my publication. Not control it.

Frustrating things

Substack’s editor is basic. No proper headers beyond H2, H3…, no real table of contents, no schema markup. You work with what you’ve got.

Search Console takes 3–7 days to start showing data. Then a few weeks to show patterns. It’s slow-ish. Be patient.

A custom domain Substack also starts from zero on Google’s trust meter. You’re building it up, one piece at a time. But it will be better in the long-term.

And not everything you write will rank. I had to come around on this myself. The pieces I love most aren’t usually the ones that show up in search. That’s fine. They have a different job.

The Bottom Line

Custom domain. Search Console. About page. Sections. A few SEO settings. And some good topics. That’s my checklist at publish time. Internal links between old and new.

That’s the whole setup. Half an hour, once. Then a few minutes per article forever after.

The email gives you today. SEO gives you next year. You can have both.

When I started on Medium in 2021, I paywalled everything immediately. Every single story, locked. I was in the Partner Program, I wanted to earn.

Made sense to me, right?

It didn’t work as I expected.

I’ve talked to a lot of new writers since then. Some do what I did. Others go the opposite route: everything free, no paywall, build the audience first. That also doesn’t quite work. Not in the way they hope, anyway.

After 4 years and about 900 articles, here’s what I think you should do.

„Paywalled” on Medium

Quick definition for anyone just starting out: on Medium, you can mark any article as a metered story. That means only paying Medium members ($5/month) can read the full thing. In return, you earn from their reading time and engagement.

Non-members hit a wall after a paragraph or two. They can’t read the rest without subscribing.

Free articles, on the other hand, anyone can read. Members, non-members, Google bots, people who clicked a link from Twitter. Everyone.

That difference matters.

Always-paywalled fails you early

Medium has around 100 million monthly visitors. But active paying members probably way less. Most people landing on your article from Google, social media, or a referral are NOT members. Most likely.

If everything is paywalled, those people bounce immediately. They never read it. They never follow you. They never come back.

You’re writing for an audience you don’t have yet. And paywalling everything cuts your potential readership by a lot before you’ve built any momentum.

There’s also the discovery problem.

Medium’s algorithm pushes stories to readers based on engagement. No reads, no engagement. No engagement, less distribution. The paywall makes it harder to break that loop when you’re new.

I’ve seen writers with genuinely good articles sitting at 3 views after a month. All paywalled, no audience, no traction. The articles weren’t the problem.

Always-free fails too

“Go free, build your audience first, then monetize later”. Sounds logical but…

If you never paywall anything, you earn nothing. Not a penny, even if you have thousands of reads. The Partner Program only pays on metered stories. Mostly.

Writing for free permanently means you never develop the habit of writing things people want to pay for. There’s a subtle difference between writing something worth reading and writing something worth paying for. Paywalling makes you think about that.

Also, “build the audience first” can take years. Writing free articles for a year and then switching everything to paywalled doesn’t suddenly make you money. You’re still starting over with the earnings side.

And putting no skin in the game makes it easy to not take the writing seriously enough. When you’re getting paid even small amounts, it sharpens your focus. Even $3 a month from a story matters when it’s your first $3.

Did to me.

Answer: Mostly paywalled, with some strategic free

Paywall most things. That’s the baseline. I do that. You joined the Partner Program to earn, so try to earn.

But keep some articles free, on purpose, for specific reasons.

Free articles are your growth engine. They’re how non-members find you, follow you, maybe subscribe to your email list, and eventually become readers who follow your paywalled work. The front door.

The ratio I’d suggest for a new writer: roughly 3 paywalled for every 1 free. Not a hard rule. A starting point.

When to go free

Make an article free when any of these apply:

It targets a search term. If you wrote something optimized for Google, go free. Google can index it fully. Non-members can read it. That’s the whole point of SEO on Medium. A paywalled article still gets indexed, but the wall kills the click-through benefit. Free articles rank and convert readers far better.

It’s a wide-audience introduction piece. An article about why you started writing, your story, what your Medium is about, make it free. It’s not your best work in terms of depth. It’s a handshake. Let everyone read it.

It’s a list or roundup with broad appeal. “10 apps I actually use daily.” “5 mistakes I made as a new writer.” These spread. They get shared outside Medium. Paywalling them limits the spread.

You want to test something. A new format, a new topic. Free it, see what happens. Less risk, more data on whether people actually engage with it.

When to paywall

Paywall when:

It’s a deep, high-value piece. Your best stuff. A 2,000-word breakdown of something you’ve lived through or know well. The kind of article that makes someone think “this is worth $5 a month.” Put it behind the wall.

It’s niche-specific. If the topic is narrow enough that most readers who’d find it are already Medium members in your niche — paywall it. A niche audience on Medium is more likely to be paying members than a general audience from Google.

It’s most things. Genuinely. Don’t overthink every article. Paywall it by default, make it free intentionally. That’s the better default than the reverse.

Downsides of the mixed approach

It’s more decisions. Every time you publish, you have to think about it instead of just hitting the same button. That’s mildly annoying.

You’ll occasionally get it wrong. A piece you thought was great for discovery dies on free. A deep niche article you paywalled would have done better as a free SEO piece. That’s fine. You’ll calibrate over time.

Free articles also earn nothing. If you’re publishing a lot of free content early on, your earnings will be low anyway. That can feel demoralizing when you’re already not making much. Just remember what the free articles are FOR. They’re not there to earn, they’re there to bring in readers who read your paywalled stuff.

Also: Medium’s boost system rewards paywalled articles. A boosted article earns. A free article doesn’t, even if it gets boosted distribution. So if you’re targeting boosts — and getting into publications that have boost nominators is one of the better things you can do early — paywall those articles.

The Bottom Line

There’s no version of “always paywall” or “always free” that works well for a new writer. Both are wrong.

Paywall most things. Go free strategically, for SEO and discovery. Keep the ratio roughly 3-to-1 until you have an audience, then adjust based on what’s actually working for you.

The goal isn’t to maximize earnings in month one. It’s to build enough of an audience that your paywalled articles actually get read. Free articles are how that happens.

You’re not leaving money on the table by going free sometimes. You’re building the table.

The history of tech that gets repeated is mostly the same guys. Jobs. Gates. Wozniak. Berners-Lee. Turing, when we feel like being inclusive about who counts as a founder. The women who built the foundations underneath all of that didn’t make the poster.

So here are four. There are more, of course.

Hedy Lamarr

She was a Hollywood actress in the 1940s. Marketed as “the most beautiful woman in the world,” which is the kind of label that tends to flatten everything else about a person.

What got flattened in her case was a patent.

In 1942 she co-invented frequency hopping spread spectrum. Originally meant to keep Allied torpedo signals from getting jammed by the Nazis. The Navy filed it away and ignored it for decades.

What we use today because of her: Wi-Fi. Bluetooth. GPS. Every wireless protocol that hops across frequencies to avoid interference traces back to that idea.

She got a Pioneer Award in 1997, three years before she died.

Margaret Hamilton

She led the team that wrote the onboard flight software for the Apollo missions. The code that put humans on the moon and brought them home.

The famous photo is her standing next to a stack of printouts taller than she is. That’s the source code. All of it. Hand-written, hand-checked.

She also coined the term “software engineering.” At the time it was a joke. Engineering was for hardware. Software was something you scribbled. She used the term on purpose, to argue that what they were doing was engineering and deserved the same rigor.

What we use today because of her: the entire framing of software as a discipline. Every job title with “engineer” in it. Every conversation about reliability, error handling, priority interrupts (her code on Apollo 11 invented the modern version of that, by the way, and it’s literally what saved the landing).

She got the Presidential Medal of Freedom in 2016. Forty-seven years after Apollo 11.

Radia Perlman

If you’ve ever connected to the internet, her algorithm is somewhere in the path your packets took.

She invented the Spanning Tree Protocol in 1985. It’s what lets network switches talk to each other without packets going in circles forever and melting the network. Without it the modern internet doesn’t work at scale.

She wrote a poem to explain the algorithm. It’s in the original paper. What a great idea.

She doesn’t like being called “the mother of the internet.” Says it’s reductive.

What we use today because of her: every Ethernet network. Every office Wi-Fi.

Karen Spärck Jones

She figured out how to teach a computer which words in a document matter.

In 1972 she introduced inverse document frequency. The idea that a word’s importance isn’t how often it appears, but how rare it is across the whole collection. Common words tell you nothing. Rare words tell you what the document is actually about.

That became TF-IDF. Which became the foundation of every search engine. Which became Google.

What we use today because of her: search. All of it. Even the modern AI stuff still uses TF-IDF as a baseline. Your Google query, your Apple Spotlight, your Algolia search bar on some random e-commerce site. Her math is in there.

She was a computational linguist. Same field I studied. I heard about TF-IDF. I did not learn her name until much later.

Why we don’t hear about them

Unfortunately, mainly because they’re women.

Patents filed under their employers. Credit going to the men who managed them. Awards arriving forty years late. Histories written by people who weren’t paying attention (to women).

Also: their work tends to be infrastructure. Frequency hopping isn’t a product you can hold. Spanning tree isn’t a thing you click. TF-IDF isn’t an app. The work that holds everything else up is the work that disappears into the background.

Whereas the guy who designed the icon set gets a documentary…

The Bottom Line

I use these ideas, inventions, genius creations every single day, and I didn’t know the names of the women that brought them into the world.

If you’ve heard of all four of them already, kudos.

If you haven’t, now you have. Spend ten minutes on each one. The stories are quite amazing.

Apple makes a lot of stuff. They also stopped making a lot of stuff.

And there are a few things they never made that I wish they would. But never will. Probably.

Most of these categories are dominated by companies that don’t care about design, don’t care about privacy, don’t really care about usability, even. They care about margins. Which Apple does, too, but not just that.

Apple is one of the few big tech that do design, usability, and margins well.

So here’s my wish list. Six things.

1. A printer that doesn’t suck

Every printer I have ever owned was a small disaster. Loud. Ugly. Refuses to print black because cyan is “low.” Wants me to install a 400 MB driver in 2026. And has software that occasionally works.

Apple should make a printer.

Minimal box. White or space gray. One button. AirPrint baked in at the silicon level, no driver download. Refillable fluid ink that lasts for years of normal household printing.

Reasonable pricing. Replacement ink at a fair price, not the printer-cartridge scam where the ink costs more than the printer.

Good scanning baked in.

I don’t need fax. I don’t need a (bad) touchscreen on the printer. I need it to print when I press print on any device.

Will Apple ever make this? Probably not.

The printer business is too messy and low margin, I guess. But I’d buy one immediately. Because that product is something many of us need. All the time.

2. AirPort comeback

Apple made routers. They were great. White, simple, fast for their time, no cloud login that watched your traffic. They killed the AirPort line in 2018.

Since then, routers have been a parade of plastic spider-looking things from Netgear, TP-Link, and Asus. I use a German Fritzbox. It works. Good company. But not aesthetic at all.

Bring back AirPort. Routers and mesh relays. Beautiful objects. Private by default. Set up from your iPhone in 10 seconds. No app account, no firmware that phones home, no “premium” subscription to enable basic features.

Mesh networking is a solved problem now. Wi-Fi 7 is here. The hardware is cheap. Apple just needs to want to.

I wish…

3. HomePod with a display

This one keeps almost happening. The rumor has been floating for years.

HomePod, but with a screen on the front. Kitchen device. Nightstand device. FaceTime, timers, recipes, photo frame, Apple TV interface, Siri that can show you what it found.

I guess this product relies on a better Siri / AI. Otherwise, I don’t see any reason not to make it.

The HomePod mini is great audio. The current Echo Show line is the closest thing on the market, and it’s an Amazon ad machine. There is a real gap here for a private, quiet, well-designed display speaker.

I’m pretty sure this one will eventually ship. Once Siri and Apple Intelligence are good.

4. AirPower (for real this time)

Remember AirPower? The wireless charging mat Apple announced in 2017 and then killed in 2019 because they couldn’t get it to work without overheating.

My God…

That was seven years ago. The technology has moved on. Qi2. MagSafe.

I want the original promise. A flat mat. Drop your iPhone, AirPods, and Apple Watch anywhere on it. Charges all three. No alignment, no fiddling.

Most third-party versions are plasticky, finicky, and have three little raised circles you have to hit exactly.

Try that one again, Apple. Would be cool.

5. An outdoor Bluetooth speaker

I love my HomePod and HomePod mini. But they’re wired and wifi. I also love my backyard, and listening to tunes outside.

So, I bought a cheap Bluetooth speaker of Amazon last summer. It sounds fine. It looks like a piece of gym equipment. And it broke a few months in.

The current outdoor speaker market is split between mostly ugly rugged plastic (JBL, Sonos Roam, Bose) and beautiful things that aren’t entirely waterproof.

Almost nobody does both (well).

Apple should. Aluminum body. IP67. Battery that lasts a full weekend at the lake. AirPlay 2 when you’re near Wi-Fi, regular Bluetooth when you’re not. Stereo pairing with a second one. A handle that looks like it was designed by someone who has held… a thing before.

Could even be premium price. I’d still get it. Or two.

6. The Mac Mini NEO

Lastly, a little cheat entry.

This one isn’t really a product I need. It’s a product I have written about because it should exist.

A small, cheap, colorful desktop Mac. A18 Pro chip. Around $399. The cheap PC market killer.

It’s not on the roadmap. Probably won’t happen. But of all the products on this list, it’s the one Apple is most capable of shipping tomorrow. Because they have everything they need for it already. From the MacBook Neo, from iPhone, from Mac Mini, even the form factor from an Apple TV box.

Would be a cool device for kids, companies that only need a tiny machine to run some web apps with a good big display, or even schools or universities that only use basic office software and learning apps.

Will They?

Most of these products will never happen. Unfortunately.

Apple doesn’t enter low-margin commodity categories. Usually. Maybe the new CEO changes that. But I doubt it.

Printers, routers, charging mats, these are all races to the bottom dominated by companies willing to ship plastic junk for $40 on Amazon.

Apple won’t play that game.

And when Apple does enter a category late, they sometimes get it wrong.

The HomePod was almost discontinued. The original AirPort eventually fell behind. AirPower never shipped at all.

So this list is probably more wish than prediction. They should still try more. Again.

The Bottom Line

The categories where Apple doesn’t compete are the categories where I’m forced to buy ugly, plastic, privacy-unfriendly things from companies I don’t want to buy from.

I’d happily pay an Apple premium for any of these. Many people I know would too.

Bring ’em back. Or build them for the first time. We’d love it.

The MacBook Neo is pretty cool. Cheap, colorful, fanless, A18 Pro inside. It works.

Now let’s imagine a desktop version of this. A Mac Mini Neo. Same idea, just a tiny box instead of a laptop. Maybe an A19 Pro. Maybe fixed RAM. Maybe in the same colors as the laptop.

I mean… why not?

What it would probably be

A small enclosure. Possibly Apple TV sized. A19 Pro chip, the same family that powers the iPhone, not the M-series that powers the regular Mac Mini. Or an older M-chip.

Fixed RAM. Maybe 8 GB. Maybe 16. Soldered, not upgradable.

Fanless. Cool colors, like the MacBook Neo.

Price somewhere between $299 and $399.

Why Apple would build this

Not to kill Windows. But… yeah, really to kill Windows PCs.

There’s a gap right now. People who use an iPad as their main computer but keep hitting walls. People who have an iPhone and an Apple Watch and a HomePod but no Mac. Students who want to switch but can’t a bigger Mac price tag.

A $299 — $399 colorful little box solves all of that. It pulls iPad-only users into a real desktop OS without asking them to spend laptop money. It locks the next ten years of their software life into Apple.

Windows killer

Budget Windows PCs aren’t all bought by people choosing between platforms. They’re bought by Windows users who have never used anything else, by enterprises with twenty years of Windows-only software, by people who want to upgrade the RAM in two years, by people who build their own machines for fun.

None of those people are switching to a fanless 8 GB Apple box. Not because the Mac is bad. Because that’s not what they need.

School districts running on Windows licenses, custom ERP software, weird old printers, that one Excel macro from 2009. None of that goes away because Apple ships a cute mini.

But it might appeal to many new users and iOS fans.

The internal problem

Apple has to be careful not to eat the regular Mac Mini.

The current Mac Mini starts at $599 with an M-series chip and good RAM. If a Neo at $299 does 80% of what most people need, why would anyone buy the Mac Mini for casual use?

Apple has done this before, though. They’re good at it. The Neo will be limited on purpose. Capped RAM, capped ports, capped pro features. So the people who actually need a Mac Mini still buy a Mac Mini.

That’s also why I think 8 GB is realistic. Apple needs the regular Mac Mini to still have a reason to exist.

The MacBook Neo already has a RAM problem for the same reason. The Mini Neo will too.

No. I need more. Also, I use MacBooks. But I love the idea.

Downsides

Fixed 8 GB RAM ages fast. In 2027 it will feel tight. In 2028 it will feel… old.

Not upgradable. When the SSD wears out, the whole machine is e-waste.

Limited ports, almost certainly. Apple TV-sized boxes don’t have room for much.

And the colors might not survive scrutiny. Cute on a laptop you carry. Maybe odd on a desk next to a black monitor. We’ll see.

The Bottom Line

The Mac Mini Neo would be a smart product. Small, cheap, colorful, efficient, ecosystem glue.

It wouldn’t kill Windows entirely. It would just give Apple another door into your home. A third Mac per household instead of one.

That’s enough for them.

I like Notes. I do. It’s a fun part of Substack,

You post a thought, people reply, someone restacks it. That’s a good feeling.

But if you’re trying to grow your Substack, in a way that lasts longer, Notes probably shouldn’t be the main thing you’re counting on.

SEO should.

What Notes is great for

Notes is Substack’s social feed. Short posts, restacks, replies.

It’s great for staying in touch with your existing subscribers between posts. It’s great for finding other writers short content. It’s great for the occasional nudge toward a new article.

That’s the use of Notes.

What Notes are not

A Note has the lifespan of a tweet. Maybe shorter. You stop posting for a few days and your reach quiets down. Stop for a week and you’re starting over.

That’s how short social feeds work. But it means the growth from Notes only lasts as long as you keep posting.

That can be fun for a while.

SEO keeps working

An article you write today, with a clear title and a topic people search for, can keep pulling readers in next year. And the year after.

One searchable piece I wrote in an afternoon has been bringing me subscribers for months. I haven’t touched it. It’s just there, doing its thing.

I’ve written about how I do Substack SEO.

Rough numbers

For a while I posted a lot of Notes. Some did well, by Notes standards. A few cracked a few hundred likes.

The subscribers from all of that were on the lower end. Some weeks good, some weeks quiet.

The searchable articles, in the same period, just kept adding readers. Steady. And they didn’t stop when I took a break.

The Notes audience and the reader audience aren’t the same

The people who like your Notes aren’t the people who open your emails. Usually.

A Note about a hot Substack topic can get 500 likes. The article you publish that week, where you put effort in, gets fewer reads than that.

Doesn’t mean Notes readers are bad readers. They’re just there for the feed, mostly. Different mode.

They might get you a lot of new subscribers. Viral notes can. But if those new subs never open a mail, what’s the point?

What I’d focus on

Three things:

1. Writing articles people search for. Comparisons. How-tos. Search query answers. The stuff Google can find a year from now.

3. A custom domain. $50 once. Your SEO traffic on your own domain travels with you.

5. Use Notes for what it’s good at. Talking to your existing readers, sharing new pieces, hanging out with other writers. Not as the engine.

I’ve also come around on Substack’s discovery a bit, so this isn’t a one-platform argument.

Frustrating things

Notes is designed to keep you scrolling. So am I, when I’m in there.

And the platform pushes Notes. Notifications, digests, nudges. That’s their incentive, not yours.

You also get this background guilt. Other writers post all day. You feel like you’re missing out by not doing it. You’re not.

I like SEO for that reason. No “active“ feed.

Keyword research, title patterns, internal linking. None of it involves posting Notes five times a day.

The Bottom Line

Notes are cool. Keep posting them if you like it. I will occasionally.

But the readers who stick around, the ones who open every email, mostly came from search. Not from the feed.

If you’re on Substack and you feel like your articles disappear right after the email goes out, you’re not imagining it.

That is happening. To many of us.

I hit publish, watch the open rate climb, feel good about it, and by Friday the piece is already buried under the next one. Nobody finding it. Nobody coming back to it.

It used to bug me. Now I do something about it. And it’s not that complicated, I promise.

Substack SEO 2026

SEO just means writing in a way that Google can find your articles later. Not just on the day you publish.

That’s it. There’s really no trick to it. You’re not gaming anything. You’re just making it easier for someone searching “best substack SEO advice 2026” to land on your piece about it, instead of someone else’s.

For most of us, the email blast is one good day. Search keeps going.

Both matter. But the latter is for long-term.

Get a custom domain first

This is the one technical step. Substack lets you connect your own domain for $50, one time. Mine is letters.byburk.net.

It matters because Google takes your own domain more seriously over time. The default substack.com address is shared with everyone else. Yours is just yours.

Bonus: You can take it with you and SEO traffic is yours, not Substack’s.

$50 might feel like a lot to some people, I get it. But it’s the single biggest thing you can do for your SEO credit, and you only do it once.

Write titles people would search for

This was the hard part for me.

For years, I wrote titles that I liked. Personal. A bit clever. Or vague on purpose.

Those titles don’t get found on Google. Usually. Nobody searches “the newsletter thing I keep thinking about.”

What people do search is more boring. “Medium vs Substack.” “Best newsletter platform.” “How to start a paid newsletter.“

(Side note: Those are way too generic to actually use.)

I started writing titles closer to those phrases. A bit clearer on the content and topic. The piece can still have your voice inside. The title is just the door, and the door has to say what’s behind it. At least for Google.

If you want to check what people search, Google’s own autocomplete is free and pretty good. Type your topic and see what it suggests. Or you use the Google Search Console.

Link your old pieces to your new ones

Every time you publish something new, link to one or two older pieces inside it. Then go back to those older pieces and add a link to the new one.

Busywork. It is, a little. But it’s what tells Google your archive is alive and connected.

I do it every time now. Five minutes per article.

AI can also help automate the boring tasks very well.

Some pieces will rank, some won’t

Not all of your work is going to find readers through search, and that’s fine.

Would be too easy, right?

What tends to rank, in my experience: comparisons, how-tos, pricing breakdowns, anything with a year in the title. Partly kidding.

What doesn’t, usually, unless it’s really done well: personal essays, opinion pieces, anything too generic.

Those still matter, they’re just for your subscribers, not for Google.

You don’t have to choose one or the other. Mix it up. The personal pieces feed your list. The searchable ones bring new people in.

It takes a while

I started seeing search traffic about three months after I changed how I wrote.

By month six, a few articles were getting 30–80 weekly clicks from Google. A handful pulled more. Most still got nothing.

Over time, it adds up. Compounds. And it doesn’t go away the way an email open does.

If you’re hoping for a magic month, this isn’t that.

Frustrating things

Substack’s editor is basic. No fancy headers, no table of contents, no schema markup. You work with what you’ve got.

SEO writing also might feel a bit stiff at first. That’s why I don’t overdo it. I still want to write naturally about what I want and like. With some SEO in mind.

And it’s slow. I keep saying that because I really mean it. A custom domain Substack starts from zero on Google’s trust meter. You’re building it up, one piece at a time.

If you want the longer version

I wrote down everything I figured out in Substack SEO. The keyword research I do, the title patterns that work, the internal linking system, the templates I use.

You don’t need it to start. Everything I said above is enough to get going. The guide is just for if you want deeper insights.

The Bottom Line

You don’t have to be a marketer to do this. Or an SEO expert. Might help. But I can’t imagine that being fun. You just have to write a little differently than you would for an email.

Pick a topic people search for. And you still enjoy writing about.

Write the piece you’d write anyway, just with a clearer title. Link it to your older work. Wait.

That’s it. That’s the whole thing.

The email gives you today. Search gives you next year. You can have both.

For most of 2025 and before, I was convinced Substack was great for discovery. Notes. Recommendations. The Substack app surfacing new writers. Momentum.

4 months into 2026, I’ve been going back through my own read counts on both platforms, looking at what articles are still getting reads 60, 90, 120 days after publishing.

And those numbers on Substack have little to do with Substack-internal discovery. Also, my Substack numbers do not nearly compare to my Medium numbers.

Changes

Substack Notes seemed to bring a big change to Substack. The “social media“ change… but in a good way (we hoped).

It did… and it didn’t.

Short posts, real engagement, writers recommending writers. A discovery engine. And for a while, it probably was, for people who were already on Substack and for newcomers.

But Substack discovery happens mostly inside Substack. Notes reaches people who already have the app. Recommendations push you to other newsletters’ subscriber lists. It’s circular. It’s closed.

Other forms of discovery aren’t closed. And they work better, at least for me.

Look at Medium here as well. Medium’s discovery happens a lot on Google. Internally too, of course, through boost, algorithm, and distribution, but also externally form search.

Substack Notes, and it’s internal discovery mechanisms aren’t as good as Medium’s, and externally, most Substackers ignore the search situation.

The numbers

I write articles on both platforms. Same topics, similar frequency. In 2025, I (really) started tracking reads at 30, 60, and 90 days for a set of articles on each side.

The Substack articles spike on the day an email blast goes out. Maybe 500–2000 opens for my list. A handful of web visits from search. Then it used to flatten, usually.

By day 30, most articles were getting single-digit reads per week. If any.

The Medium articles either start slower and build up over time, or they hit right away. Well… or they bomb entirely.

But the ones that work, they don’t flatten the same way the ones on Substack did.

Articles I published in January are still pulling consistent reads in September. Some of them more than they did in the first two weeks.

On rough average, Medium articles at 90 days were pulling about 3x-5x the weekly reads of comparable Substack articles. More for topics with clear search intent. Less for opinion pieces where there’s no obvious keyword.

That’s why, in mid 2025, I switched my Substack strategy to heavily focus on SEO. Instead of Substack-internal discovery.

And it paid off a lot.

Now my Substack articles get more search reads. In fact, search has become my #1 growth source for views and subscribers, including paid subscribers.

I barely use Notes anymore. I focus entirely on SEO.

Medium still wins for me though

I’ve been on Medium longer and my stories do well often. My old stories also still get read. So, right now, for me, Medium is still #1.

But Substack is catching up. With SEO. I’ve written about how to use Substack for SEO, and the strategy works. It’s slower.

I am building authority from zero, not inheriting it, because I use a custom domain on Substack and can’t ride Substack’s own high domain authority. Not ideal in the short term, but way better long-term, because the generated traffic belongs to me now, not to Substack.

I do the same on Medium. Custom domain. Traffic is mine.

Where Substack beats Medium

Substack wins on owned audience right now.

Every subscriber is an email address you control. Every open is a real person who chose to hear from you. Unless they change that, it is a deciding factor for Substack vs. Medium.

Substack also wins on newsletter format. If your writing is conversational, intimate, built for the inbox, Substack is the right home for it. The reading experience is better than Medium for that kind of work.

And for some niches, Substack discovery is working pretty well, I hear. Politics. Finance. Local journalism. These have active recommendation networks on the platform. Writers in those categories see real growth from Notes and Recommends.

Discovery != Discovery

Community-driven discovery and search-driven discovery are different. Both are options. Both can be done. They just operate on completely different timelines and for completely different readers.

I enjoy the SEO route more.

Substack Notes might send me 50 new readers today. They may or may not open any further issues. A Medium article on the right long-tail keyword can send me 20 readers a week for three years.

And good SEO on Substack can do the same.

If you’re publishing on Substack and nothing else, you’re betting everything on email and community. That can work. But you’re leaving a big chunk of search traffic on the table.

If you want to see what a dual-platform approach actually looks like in practice, I went deep on it in how I earn money on Substack in 2026.

Downsides

Medium’s paywall cuts reach. A lot of potential readers hit the paywall and leave. You’re not building a free audience the same way Substack lets you. But if you paywall everything on Substack too, it’s not better.

The Medium Partner Program earnings are still unpredictable. I know. I’ve had $30 months, $300 months, and $3000 months on the same publishing frequency.

Search traffic helps, but it doesn’t make the income stable.

Medium’s ownership risk is also there. Ghost subscribers on Substack are a problem but at least you have the emails. On Medium, your “followers” are a number, not a list you can export. And the subscribers are hidden now.

So, Substack has some things going for it.

I use WriteStack* to track a bit more what’s performing on Substack.

The Bottom Line

Substack discovery is there. It just doesn’t compound the way search does (for me).

If I could only pick one platform for a new writer asking about discovery, I’d still use Medium right now. Or at least, Substack with proper SEO.

I wouldn’t rely on the inbox and Substack-internal discovery.

You can read more about which platform pays better and how to rank your Substack on Google if you want to work both angles.

*this is an affiliate or SparkLoop* partner link. I may earn a commission.

Opus 4.7 dropped recently. Most people saw “better benchmarks” and moved on.

But tucked inside Anthropic’s own documentation was a sentence that changed how I think about the whole thing.

Mythos Preview is their most robustly aligned model. The model they released to the public is NOT the most aligned version they have. And more importantly: the most aligned version is also not the most capable.

The Numbers

Yes, Opus 4.7 outperforms 4.6 on most headline benchmarks. Finance agent tasks, long-context reasoning, complex coding, vision. They went up. The press release is real.

But benchmarks measure what they measure. And what they often don’t measure is behavior.

Opus 4.7 produces overly detailed harm-reduction advice in areas where 4.6 was more restrained. Flagged in Anthropic’s own safety documentation as a regression. Not a bug they’re fixing. A known shift when they tuned the alignment parameters to a different position.

This is what aligned AI looks like in practice, it seems. Not a clean improvement. A tradeoff. Some things go up. Some things go sideways. Some things get dialed back in ways that don’t show up in leaderboard screenshots.

The Alignment

There’s a concept in AI research called the alignment tax.

Making a model safer costs you something. You can’t maximize both raw capability and corrigibility at the same time. At some threshold, the guardrails constrain what the model will do.

Anthropic says this isn’t entirely true. They’ve argued publicly that alignment and capability aren’t opposing forces. That a well-aligned model is more valuable, not less capable.

Probably…

But “not opposing forces” is different from “no tradeoffs.” And Mythos Preview is the proof. If there were truly no tradeoffs, the most capable model and the most aligned model would be the same model. They’re not.

Two separate products. Two separate positions on the tradeoff curve.

Coding

I’ve been building with Claude Code daily for the last few months. Flutter apps, React frontends, PHP backends. I used Anti-Gravity but went back to Claude. Claude Code keeps winning.

More aligned models add hedging. They ask clarifying questions before writing code that touches user data. They attach security warnings to implementations that are pretty standard. They occasionally refuse edge cases the previous version handled without complaint.

Individually, every one of those behaviors is reasonable. Together, they add some friction.

Not a lot. But some.

The “safer” the model, the more it second-guesses itself. I’ve started dictating longer context prompts using Wispr Flow*, it’s faster by voice than typing the setup every time.

The Capability

Anthropic had a more capable version of Opus 4.7. Probably significantly more capable on certain tasks. And at some point in testing, that version crossed a threshold. Maybe a harmful threshold. Maybe a threshold where the behaviors became harder to constrain without meaningful capability loss.

So they didn’t ship that version.

They shipped what we got. Which is very good. But it’s not the ceiling of what Anthropic built. It’s the ceiling of what they were comfortable releasing.

I’ve written before about how AI coding is changing everything right now, and I’ve also written about the AI privacy problem that vibe-coded tools ignore. A more capable model that’s harder to align might be a worse product in those contexts, not a better one.

Mythos Preview exists as a separate, more constrained product. You can have more alignment. It just costs you somewhere else on the spec sheet.

Options

Sonnet 4.6 is faster, cheaper, and has slightly fewer default constraints on code generation. For most development tasks, it’s close in quality and outperforms Opus on pure throughput.

Opus 4.7 is excellent for complex agentic tasks, long document analysis, and anything where reasoning depth matters more than raw speed. That’s still a lot of real work.

If you want to understand the tradeoffs before picking: that’s the most interesting position to be in. The answer is it depends on what you’re building and how much you want the model to push back on you while you build it.

I mostly run Claude Code in the terminal, switch to Sonnet 4.6 for fast iteration, and pull in Opus for the hard stuff.

The Bottom Line

Anthropic is a safety-first AI lab that happens to ship incredibly capable models. The two things are in tension. Not irreconcilably. But meaningfully.

Every release is a negotiation between what the model can do and what Anthropic is willing to let it do. Opus 4.7 landed at a certain point in that negotiation. Mythos Preview landed at a different one.

The version you wanted might have existed.

It just never shipped.

*this is an affiliate or SparkLoop* partner link. I may earn a commission.

Start a Substack. The pitch sounds great. Own your audience. Email is forever. Newsletters are booming.

All of that is true. Doesn’t change the core problem, though.

Growing a newsletter from zero is one of the hardest things to do online right now.

Substack Is a Blog

I’ve written about this before. Substack is a blog with email attached. The email part is what most people argue with or sell. The blog part is what most ignore or underestimate.

Blogs don’t get found on their own. Not without traffic. Not without someone already knowing you exist. You need to bring the audience to Substack. Substack doesn’t bring it to you.

Yes, there are recommendations. Notes. A discovery section. For established writers with existing audiences sending each other traffic, those features work great. For someone starting from zero? Less so.

Growth

Say you want 1,000 Substack subscribers. Reasonable goal.

Where do they come from?

Your social media following. Your cross-posts. Recommendations from bigger newsletters. Guest appearances. Substack Notes someone clicks through.

Every single path requires you to already have an audience somewhere, or to manually build one with another platform (or mechanism like Notes). There is no version of Substack where a stranger types a question into Google, finds your article, and subscribes. Unless you build your Substack specifically for this use case!

Medium has something better here. Not perfect. Not every story. But it exists.

I’ve had Medium stories get picked up either by search engines, by curators, by publications, or just by the algorithm, and collect reads for months or years.

Nobody signed up for my newsletter first. Most still don’t. And I don’t expect them to.

They just found the article through search or through Medium’s own recommendations. That discovery rarely happens on Substack.

Again, unless you build your Substack as a blog for this specific use case.

Medium has…

organic reach. More than Substack at the moment.

The algorithm surfaces stories to non-followers. A new writer with zero audience can publish something today and have it read by strangers tomorrow. Not guaranteed, obviously. That would be too easy.

Not even reliable. Because it’s shouldn’t be that way, actually. Medium doesn’t and shouldn’t always push the same writers.

But it is possible, and that possibility compounds over time. More than on Substack. Or any other platform for writing.

Also, Medium has SEO. Serious domain authority built up over years. A well-titled story ranks on Google. I have articles from 2022 still pulling in reads every week. That longevity is great. It’s one of the better arguments for the platform.

Substack can become this. I do Substack SEO. And it’s starting to pay off. It took a while, though.

As mentioned before, Medium has discoverability inside the platform. Tags, publications, related stories at the end of every article.

Medium is built, so readers keep reading. Substack is built, so subscribers get emails.

Two very different products solving different problems.

And there’s luck, of course. Medium gives you more surface area. More chances. On Substack, growth is linear and mostly manual. On Medium, occasionally something takes off sideways, and you don’t fully understand why.

That’s not nothing.

The Downsides

Medium owns the relationship. Your readers are Medium readers first, yours second. You have some emails. But mostly, not.

If the platform changes its rules (they do, regularly) or changes the Partner Program payout structure (also yes), you’re at their mercy. But I truly think this is fair compared to what we’re getting. Possible instant reach to millions of readers. And monetization for that.

For SEO, the paywall limits reach, yes. Your best work is and should be behind a metered wall. This hinders SEO, but makes direct monetization easier.

Casual readers might bounce before they get attached. SEO brings people in, but conversions from “found this on Google” to “paid subscriber” are soft. That’s okay.

The Partner Program pays inconsistently. Some months make sense. Some months… not so much. It’s not a salary and you shouldn’t treat it like one. Medium doesn’t treat it that way.

But yes, Substack does have something Medium doesn’t: a list. An actual email list you own and can export (with visible email addresses). Medium unfortunately hides those now. They shouldn’t.

The Bottom Line

I publish on both platforms. I’ve written about how I actually make money on Substack in 2026. It’s not nothing, and I’m not leaving.

But if you’re starting from zero and someone makes you pick one, Medium is #1 for me.

Build the reach. Let the algorithm work for you while you have no followers to do that work. Then, once people are finding your writing, point them toward the newsletter.

Apple just announced that Tim Cook is stepping down as CEO.

Effective September 1, 2026, Cook becomes Executive Chairman. John Ternus, currently Senior Vice President of Hardware Engineering, becomes CEO.

This is not a surprise. Been rumored for a while. But it’s quite a shift.

Who Is John Ternus

Ternus joined Apple in 2001. The Steve Jobs-era Apple. He was 22, fresh out of University of Pennsylvania with a mechanical engineering degree, and he went straight onto the product design team.

He stayed. For 25 years.

VP of Hardware Engineering in 2013. SVP in 2021. And somewhere in those two decades, his hands were on basically everything Apple ships. iPhone. Mac. iPad. AirPods. Apple Watch. If it’s a physical Apple product, Ternus probably had something to do with it.

Tim Cook said he has “the mind of an engineer, the soul of an innovator, and the heart to lead with integrity.”

That’s a lot of nouns. Hopefully accurate.

Tim Cook’s Legacy

Cook became CEO in 2011. Apple’s market cap was around $350 billion.

Today it’s $4 trillion.

He helped built Apple Watch from scratch. Launched AirPods. Turned Services into a $100 billion business. Shipped Vision Pro. Grew Apple to 2.5 billion active devices across 200+ countries.

He didn’t invent products. He built a machine that could invent products reliably. That’s a different skill. And not an easy one.

Also, he’s not gone. Executive Chairman. But the day-to-day is Ternus now.

A Hardware Guy in an AI Era

Apple’s biggest challenge right now isn’t hardware. They’ve got that figured out pretty well.

The pressing issue is AI. Siri is still… embarrassing in 2026. Apple Intelligence has been underwhelming. Every competitor is shipping AI features weekly, and Apple is shipping cautiously, if at all.

That’s not all bad. AI is basically too much everywhere now.

Still, it’s interesting (and cool) they picked the hardware guy to be CEO and not someone focues on services, AI, marketing, finances (like Cook) or such.

I’ve written about the weird spot Apple is in with AI and third-party tools, and the situation hasn’t gotten clearer. The company is simultaneously building its own models and trying to license Google’s.

Ternus is not a software person. He’s not an AI person. On paper. He does all those things too, I’d imagine. But his main focus is/was hardware.

Maybe that’s exactly what Apple needs.

The Downsides

Ternus has zero public CEO experience. He’s been in interviews and keynotes. That’s it, as far as I know.

Cook spent years as COO before taking the top job. Ternus goes from SVP of one division to leading the whole thing.

He’s also not a known quantity to Wall Street. Cook was trusted. Analysts knew what to expect. Ternus is… an engineer who doesn’t give many interviews.

Quite the change. I like it, though.

And then there’s the AI thing again. If Apple’s biggest gap is software and intelligence, the new CEO’s entire career has been about making beautiful aluminum rectangles. Those two things don’t obviously connect.

But that might not be a downside after all. We don’t need >Apple to go the same Ai route as all the other big tech companies. Apple should do “Apple“. Something different.

The Bottom Line

Ternus is the first Apple CEO in history who actually built Apple’s products. Not sold them, not streamlined them, not managed them. Built them.

That’s either exactly the right energy for whatever comes next… or not. We’ll see.

But Apple has a pretty good track record on that bet.

Most of us have a tab open right now with a search query that returned nothing useful. We typed it. Scrolled three results deep. Closed them. Moved on.

The gap is the work

When we search for something specific and get nothing back, the hard part is mostly done. Someone (us) named the question. Tested the existing answers. Found them wanting.

That’s research.

Most articles fail at exactly this step. The writer picks a topic, then has to invent the angle, find the gap, and write something only they can write. Our searches do all three at once.

Yet we ignore them

They feel too small. Or too obvious. Or too embarrassing to ask out loud. We assume someone already wrote it well and we just missed it.

Sometimes that’s true. But often it isn’t. The internet is huge and shallow at the same time. Plenty of topics have surface-level coverage and almost nothing underneath.

If five of our searches in a month land in the same gap, that gap is probably really there.

What a good Googled topic looks like

Specificity.

“How to grow a Substack” returns ten thousand articles. Pretty useless. Not the articles. The query. The question is too broad to answer.

“How to price a paid Substack tier when the free one only has 200 subscribers” returns very little (good stuff). Because nobody can answer it without having tried it properly.

The narrower the question, the more our experience is the right material. This is closer to real niching than picking a topic cluster.

What to do with it

Open a notes file. Whatever app we use. Apple Notes, Bear, a TextEdit window.

For one week, paste in any search query that came back useless. Don’t filter. Just collect.

At the end of the week, most will be junk. Some will be vaguely interesting. One or two will be cool article ideas.

That’s a higher hit rate than any topic strategy I’ve tried.

We don’t hit gold every time

Some searches return nothing because the answer is just… hard. If others haven’t written it, maybe we can’t either.

Some searches are personal. Health stuff, family stuff, things tied to our exact situation. Those don’t transfer well. Usually.

Some searches are topical. They had no good answer because nobody had time to write one before the news cycle moved. By the time we’d publish, nobody cares.

Not every gap is a market. Some gaps are just… gaps.

The Bottom Line

Our search history is quite an honest reader. It tells us what we couldn’t find, when we needed it, and how we phrased the question.

Strategy gives us topics other writers are also chasing. The search bar gives us topics nobody else thought to write yet.

My Substack has subscribers. Some of them have never opened a single issue. Not one.

They signed up. For some reason. And then they disappeared.

I suspect the same is true for most newsletters.

That number on your profile, the subscribers, is a nice numbers. But not really the one to show off.

The Open Rate

Open rates used to be reliable. Mostly.

But when Apple launched Mail Privacy Protection, millions of iPhones started pre-loading email content automatically, marking it as “opened” even when the user never touches it.

The platform logs it. Your rate goes up. Readership doesn’t though.

How much of your reported open rate is real? Hard to say. Maybe more than half. But the 40–50% open rates… probably not.

Click-through rate is harder to fake.

You don’t pre-fetch a click. Industry average sits around 2–3% of your total list. That’s the number worth trusting more.

Not opens. Clicks.

I don’t care that much about clicks either. Although that number is far more important than open rates or subscriber count.

Ghost Subscribers

There’s a name for people who subscribed and never came back. Ghost subscribers.

They signed up because of a lead magnet, or a referral push, or one good subject line. Then life happened. They didn’t unsubscribe. They just stopped opening.

They still count toward your number.

A list of 1,000 subscribers probably looks something like this:

• 200–300 people who read most of what you send.

• 300–400 who open occasionally when a subject line lands.

• 300–500 who haven’t opened anything in months.

The audience that stays is almost always smaller than the audience that arrives. I wrote about this kind of long-game publishing reality in Writing the Same Topic for Five Years

1,000 Subscribers

1,000 subscribers. Reported open rate: 40%. That’s 400 opens.

Subtract Apple MPP inflation and you’re probably at 200–300 real subscribers per issue.

Click rate of 2%: 20 people clicking your links. Per issue.

Paid conversion. Top newsletters convert 5–10% of readers to paid. Most convert closer to 1–3%.

• At 1%: 10 paid subscribers. At $8/month: $80/month.

• At 3%: 30 paid subscribers. $240/month.

Not nothing. But still…

This is why I think about SEO alongside email.

A Substack post that ranks on Google keeps working whether your subscribers open it or not. It doesn’t care about subscriber counts or open rates.

The archive has a life of its own. I went into detail on this in How to Use Substack for SEO. It changes how you think about what’s worth writing.

The Number I Like More

Not subscribers. Not opens.

Replies.

How many people replied to your last issue? How many shared it?

For most newsletters at 1,000 subscribers, the reply count may be somewhere between 5 and 25. That’s your real audience.

Sounds kind of depressing for a 1000 sub list. But that’s how it is.

25 engaged readers who trust you enough to reply will click things, share things, buy things, and mabye stick around for years.

700 ghost subscribers won’t do any of those things.

The platform just doesn’t separate them visually, so you keep looking at the big number and feeling the wrong emotion.

Options

Audit your list. Substack lets you see who’s opened anything recently. Subscribers who haven’t opened in six months are distorting your metrics and inflating your count in a way that leads to bad decisions. Send a re-engagement email first, give them a week, then remove them. Your numbers will look worse. Your data will be better.

Track clicks, not opens. Write issues designed to get a click. One good link per email, with a reason to click it. Watch that number instead.

Monetize engagement, not size. Tools like SparkLoop* pay you per new engaged subscriber you refer to partner newsletters — not per your raw list count. If 200 real readers generate 40 referrals a month, you earn on those 40. Ghost subscribers don’t enter the equation. I’ve been using it as a passive layer on top of regular publishing, and it works whether or not your open rate looks good.

Think of your newsletter as a blog. Your Substack Is a Blog. Treat It Like One. The SEO value of your archive doesn’t depend on anyone opening your emails. That’s a separate audience entirely, and it compounds over time in a way your subscriber count doesn’t.

Grow intentionally. Subscribers who found you through search or a specific recommendation tend to be real readers. Random growth from a viral moment or a referral swap brings ghost subscribers. I built an automation system around my publishing partly to be more consistent and deliberate about this — showing up regularly to the right audience instead of chasing spikes.

The setup that supports all of this is worth thinking through carefully too. I covered what mine actually looks like in My Business Runs on 7 Tools.

The Bottom Line

If 25 people click your links per issue or reply, you reach 25 people. They’re the ones who might eventually pay you, forward your email, or support you.

The rest might come around. They might not. Either way, they’re not your audience right now. Their just numbers in your dashboard.

And really, 25 real readers is more than many people will ever achieve.

*this is an affiliate or SparkLoop* partner link. I may earn a commission.

In 2021 I started writing about Medium. How it pays. How to grow. What to do and what not to do. I had just discovered the platform and I was pretty excited.

It’s 2026 now.

I still write about it. Sometomes. And something is different now.

The Advice You Stop Believing

I didn’t get bored. Sometimes it happens, of course. You expect that. But I can write about what I like all the time.

But eventually you don’t follow your own advice anymore.

“Post consistently.” I don’t, and really haven’t for two or three years.

“Engage with your readers.” Mostly I don’t. I read the comments. Sometimes I respond. I don’t have time for more.

“Build your email list above everything.” I believe this in theory. But I actually prefer the blogging route over the newsletter route.

I write the advice anyway. Because it’s still true for other people. Because I know much of it works.

But somewhere in there, the writing starts to feel like performance, sometimes.

Rereading Old Articles Is Uncomfortable

I went back recently. Read a few pieces from 2021 and 2022.

The excitement is embarrassing, in a way. I wasn’t wrong all the time. Much of it still holds up. But the confidence. The certainty. The way I wrote as if I had figured something out permanently. Yikes.

I hadn’t. Nobody does. The platform changed. My income changed. My relationship to writing changed. The 2021 version of me didn’t know any of that.

That’s just time. But it does something to you when you read 900 articles written by a slightly different person each time.

The Niche

When you write about one topic long enough, you start to see the edges of it.

Medium. Substack. Creator income. Platforms rise and fall. The same arguments circle back every 18 months with different voices attached. I have written about Medium probably 300 times. I will write about it again.

Still…

There’s a kind of fatigue that comes not from the writing itself but from the repetition of the territory. You’ve covered this ground. You know where everything is. The map is complete and you’re still drawing it again, anyway.

New writers discover these topics every day. For them it’s fresh. You’re writing for them, mostly. That helps. But it doesn’t fully solve the feeling.

What I Did About It

Nothing dramatic.

I didn’t pivot to travel writing or start a podcast or rebrand myself.

I just started writing things that interested me more right now, even when they didn’t fit neatly into my “usual“ niche. EU tech policy. App development. The occasional piece that goes somewhere I haven’t been before.

Some of those performed badly. A few did really good, surprisingly. So good, they might become a new niche.

The Bottom Line

Five years of writing about one thing doesn’t mean you’ve mastered it. It means you’ve lived with it long enough to see what you got wrong, what changed, and what you’ve been saying on autopilot.

That’s not a reason to stop. It’s just something I realized.

The moment you turn on paid subscriptions, you become a business. Not a hobbyist. Not a writer with a tip jar. A business selling digital services to consumers in other countries.

And in Europe, that comes with tax obligations most creators have never thought about.

What Substack Says

Substack’s documentation does not bury this. It’s stated clearly: Substack does not add, collect, or remit taxes on your behalf. For Europe or Australia or anywhere else.

Substack is not a Merchant of Record.

That phrase matters. A Merchant of Record is the legal entity responsible for a transaction, the one who calculates and collects the tax, files the returns, and pays the authorities.

When you buy something on Gumroad, Gumroad is the Merchant of Record. When you buy something on Lemon Squeezy, same thing. Your tax problem is their tax problem, and they solve it.

Substack uses Stripe for payment processing. Stripe moves money. It does not handle your tax compliance. Those are different things.

So when a reader in Hamburg signs up for your paid Substack at $7/month, Germany is owed VAT on that transaction. Substack doesn’t collect it. Stripe doesn’t collect it. You’re supposed to.

Most of us aren’t.

Read more

There’s a poll going around. 83% of Canadians, when asked, are not against the idea of joining the EU. A union 5000km away from Canada.

That number is wild. And it’s been making rounds on social media, in European news, and in a lot of very excited Reddit threads.

The poll

The question wasn’t “should Canada become an EU member state.” It was closer to “would you support closer ties with the EU, including potential membership.” That’s a very different question.

And the options were essentially yes, explore the option, and no.

When you ask someone if they’d like closer ties with a large, stable, democratic trade bloc that has universal healthcare, strong labor laws, and isn’t currently threatening them with tariffs… most people won’t say no right away.

And they didn’t. 25% said yes. 58 percent said explore. Only 17% said no.

That’s wild.

No wonder it’s trending

Canada’s relationship with the US has gotten complicated. Duh...

Trade tensions, tariff threats, political instability south of the border. The “51st state” jokes from certain US politicians didn’t help.

So when Canadians look across the Atlantic and see a bloc that talks about data sovereignty, worker protections, and multilateral cooperation… it looks appealing. Especially compared to what’s happening next door.

So, this poll isn’t really about the EU. It’s about the US.

The 83% number is a protest vote dressed up as a policy preference.

Why it can’t happen

The EU is a European union. That’s not branding. It’s geography.

Article 49 of the Treaty on European Union says any “European State” can apply for membership. Canada is not a European state. It’s 5,000 kilometers away. Across an ocean.

There is no mechanism in EU law for admitting a non-European country. None. Yet!

You’d have to rewrite the founding treaties. That requires unanimous agreement from all 27 member states.

So, it is possible. But all EU members have to agree.

And even if that happened, EU membership comes with obligations that would fundamentally change Canada. In theory.

Adopting the euro (eventually). But honestly, that isn’t really happening in all EU members states.

Free movement of people with 27 other countries. That might be a bit more problematic.

Submitting to the European Court of Justice. Aligning with the Common Agricultural Policy. Accepting EU regulations on everything from food labeling to data protection.

It’s a lot.

Canada would have to give up trade deals it has with countries outside the EU. Including the US-Mexico-Canada Agreement. Which is, you know, kind of important when your entire southern border is the United States.

What Canada already has

Canada and the EU already have CETA. The Comprehensive Economic and Trade Agreement. It’s been provisionally in force since 2017.

CETA eliminates tariffs on most goods. It opens up government procurement. It includes labor and environmental standards. It’s not perfect, and ratification by all EU member states is still ongoing, but it’s one of the most comprehensive trade deals the EU has with any non-member country.

Canada also participates in NATO alongside most EU members. It has bilateral agreements on research, education, and climate. The diplomatic relationship is already strong.

Most of what Canadians want from the EU… they already have a version of it.

What the EU is building

The EU has been focused on something different lately. Digital sovereignty. Reducing dependency on American tech infrastructure. Building out its own cloud, its own AI ecosystem, its own regulatory framework.

I’ve written about this a few times. Countries across the EU are actively moving away from American cloud servicesbecause of the CLOUD Act. And there’s a real conversation about what a sovereign European tech ecosystem could look like.

That’s the work happening inside the EU right now. Not expansion across the Atlantic. Internal consolidation. Figuring out how to be less dependent on the US. Not how to absorb North American countries.

It signals something though

The interesting part of this poll isn’t the 80%. It’s what’s underneath it.

Canadians are really worried. About their sovereignty. About their economic future. About what happens when your biggest trading partner starts (or rather keeps) acting unpredictably.

And the EU, for all its bureaucratic slowness and internal disagreements, represents something specific right now. Stability. Rules-based cooperation. A bloc that argues about cheese regulations instead of threatening annexation.

That’s an attractive alternative.

But wanting the feeling of the EU and actually wanting to be in the EU are two very different things. One is a sentiment. The other is a constitutional overhaul that would take decades and might fail.

Is there a realistic path

What could happen is more boring but more useful, I think.

Deepening CETA. Expanding it to cover more digital trade, AI governance, and data-sharing agreements. Canada could align with EU standards on privacy and data protection voluntarily, the way the UK partially does post-Brexit.

There’s talk of a “strategic partnership” upgrade. More coordination on climate, defense, and technology. A closer relationship without the constitutional impossibility of actual membership.

That won’t make headlines the way “83% want to join the EU” does. But it might actually happen.

The Bottom Line

83% of Canadians didn’t vote to join the EU.

But they expressed frustration with their current geopolitical situation and pointed at something that looks better.

The EU can’t admit Canada. Canada probably wouldn’t want the full package anyway. And CETA already gives both sides most of what they need.

But the sentiment is there. The desire for stable, democratic, rules-based international cooperation is there. And the fact that many people look at the EU and see that, despite everything, is probably the most flattering thing the EU has heard in years.

They should enjoy it.

Compliments from 5,000 kilometers away are the best kind. No obligations attached.

The European Union has been trying to pass a law that would require every chat app, every messaging service, and every email provider to scan your private messages for child sexual abuse material.

They’ve been trying since 2022. It keeps failing. And it keeps coming back.

The latest version is more specific. The proposal is officially called the “Regulation to Prevent and Combat Child Sexual Abuse.” Informally, everyone calls it Chat Control.

And yes, it’s exactly what it sounds like.

What they’re proposing

The core of it: messaging services would be legally required to scan all messages for known CSAM, for new and unknown CSAM, and for grooming behavior. All of them. Not just the flagged accounts. Not just the suspicious ones. Everyone.

That includes end-to-end encrypted services. Signal. WhatsApp. iMessage.

The technical term they landed on is “upload moderation.” Which means scanning your message on your device before it gets encrypted and sent. They rebranded it to avoid saying “client-side scanning,” but that’s exactly what it is.

Apple tried building something like this in 2021. They called it NeuralHash. It was supposed to scan photos on your iPhone before they got uploaded to iCloud. The backlash was massive. Security researchers tore it apart. Apple killed the project in late 2022 and said it couldn’t be done without creating serious privacy and security risks.

Apple, with unlimited resources and full control of the hardware, concluded it wasn’t safe. The EU thinks it can mandate it for every chat app.

The false positive problem

Billions of messages are sent in the EU every day. Even with a 99% accuracy rate, which is generous for AI-based grooming detection, you’d get millions of false positives. Per day.

Millions of innocent messages flagged and reviewed. Parents sharing photos of their kids at the beach. Teenagers sending each other pictures. Doctors discussing cases. All of it fed into a review pipeline.

An open letter signed by hundreds of scientists and security researchers in 2023 warned that the technology to reliably detect CSAM without massive false positives simply does not exist. Not “needs improvement.” Does not exist.

The Commission’s own impact assessment was criticized for overstating what current technology can actually do.

Who wants this

The original proposal came from Ylva Johansson, the EU Commissioner for Home Affairs under the 2019 Commission. She pushed hard. Spain was the most aggressive supporter in the Council. Leaked documents showed Spain favored scanning even inside encrypted messages, no exceptions.

Ireland, Italy, Hungary, Romania, Bulgaria, Greece, and Croatia have generally supported it. France had a complicated position, sometimes in favor with caveats.

Law enforcement agencies and Europol backed it. So did Thorn, an anti-CSAM nonprofit that lobbied extensively in Brussels.

The argument is always the same. Child safety. No one can argue against protecting children without sounding like a monster. That’s the point. It’s a framing designed to make opposition politically radioactive.

Who’s blocking it

Germany. The German government opposed client-side scanning, and the Bundestag passed a resolution against it.

The Netherlands flipped to oppose after the Dutch parliament instructed its government to vote against. Austria, Poland, the Czech Republic, Luxembourg, and Slovenia also pushed back. Together, they formed a blocking minority in the Council.

The European Parliament took a more privacy-protective stance than both the Commission and the Council. The LIBE Committee, Civil Liberties, voted to exclude end-to-end encrypted messages from detection orders entirely. They also limited scanning to known CSAM only, not AI-based grooming detection, and required reasonable suspicion before any scanning could happen.

The European Data Protection Supervisor issued a critical opinion. So did the European Data Protection Board. Both said the proposal, as written, was incompatible with fundamental rights.

The encryption question

There is no way to scan end-to-end encrypted messages without breaking end-to-end encryption. This is not a political opinion.

If you build a system that scans content before encryption, you’ve created a vulnerability. A backdoor. It doesn’t matter what you call it. If it exists, it can be exploited. By hackers. By governments. By anyone who finds it.

Signal’s president Meredith Whittaker said publicly that Signal would leave the EU rather than comply. WhatsApp’s head Will Cathcart said Meta would not weaken WhatsApp’s encryption. Threema, the Swiss messenger, took the same position.

So the EU has a choice. Pass this law and watch every serious encrypted messenger either leave or get banned. Or don’t pass it.

It keeps coming back

The Belgian presidency tried in early 2024. Failed. The Hungarian presidency tried in late 2024. Pushed hard, circulated multiple compromise texts, tried to carve out exemptions for audio calls while keeping text and image scanning. Still failed. The blocking minority held.

But it didn’t die.

There’s a temporary regulation that allows platforms to voluntarily scan for CSAM. It keeps getting extended. Every time it’s about to expire, it creates political pressure: “If we don’t pass the permanent regulation, platforms will stop scanning entirely.” It’s a ratchet. The temporary measure justifies the permanent one.

And every new Council presidency picks it up again. Tweaks the language. Rebrands the scanning. Adds a carve-out here, removes a safeguard there. Tries to find the exact combination of words that gets past the blocking minority.

The push for European digital sovereignty makes this more complicated. The EU simultaneously argues for data protection as a competitive advantage AND for mandatory mass scanning of private communications. Those two positions are not really compatible.

But that hasn’t stopped anyone.

Targets won’t be affected

People who distribute CSAM professionally use custom servers, steganography, VPNs routed through non-EU jurisdictions, or encrypted channels that no EU regulation can touch. They will not be using WhatsApp. They are not sending unencrypted messages through Gmail.

The only people affected by this regulation would be regular citizens. Most likely.

The Court of Justice of the EU has repeatedly ruled against general, indiscriminate surveillance of communications. The current proposal is essentially that, with extra steps. If it passes, it will almost certainly face legal challenges.

But in the meantime, the infrastructure gets built.

Options

The EU could invest in better-funded law enforcement with targeted investigations. It could require platforms to improve reporting mechanisms. It could fund victim support organizations. It could go after the hosting infrastructure where CSAM lives, not the messages of 450 million Europeans.

All of those would be more effective than scanning every message on the continent, right? I’d say so.

But none of them make for a good press release that says “we’re protecting children.”

The Bottom Line

The EU is the same institution that gave us GDPR. The same one that fined Big Tech billions for privacy violations. The same one I’ve written about positively when it comes to data protection.

And now it wants to read your messages. Like other countries do, if I may add that.

But for the children, of course. I have children. I want them to be safe. I don’t think this is the solution. Do you?

Bloomberg confirmed it. Apple’s first foldable iPhone is on track for September. Book-style design, 7.6-inch display when open, no visible crease, starting around $2,300.

Samsung did this seven years ago. Basically.

The timeline

Samsung shipped the Galaxy Fold in September 2019. It was not great. Review units broke. They recalled them, redesigned the hinge, and shipped again.

But they shipped.

• By 2021, foldables were water-resistant.

• By 2022, they were mainstream enough that Samsung was selling millions.

• By 2025, the Galaxy Z Fold was in its sixth generation. Thinner, lighter, more polished every year.

Apple watched all of that happen. For seven years.

The Apple argument

I know the counterargument. Everyone does.

Apple doesn’t do things first. Apple does things right. They wait until the technology is mature, then they ship something better than everyone else.

That used to be true.

The original iPhone in 2007 wasn’t the first smartphone. But it redefined what a smartphone was. The iPad wasn’t the first tablet. But it killed every tablet that came before it. Apple Silicon wasn’t the first ARM chip. But it made x86 laptops feel slow overnight.

Those were moments where Apple changed the rules. But… the foldable iPhone rumors don’t feel like that.

What’s new

The crease. Or rather, the lack of one. That was supposed to be the “new” thing. But opposition did that just now. Basically no crease.

Apple reportedly spent years eliminating the visible crease (really?), “regardless of cost.” But Android foldable do that now too.

So, what’s left?

The iPhone brand. Not much else, right?

The rest of the spec sheet reads like trade-offs. Two cameras instead of three. No Face ID, back to Touch ID. Battery around 5,000 mAh. Starting at $2,300, going up to nearly $2,900 for the 1TB model.

You’re paying more and getting less, it seems like. For a phone that folds. Nothing more.

The problem

The foldable iPhone won’t be bad.

It’ll probably be well-built, smooth, tightly integrated with iOS. Apple will make it feel premium. That’s what Apple does.

The problem is the signal.

Apple used to define categories. Now it’s entering categories that Samsung defined almost a decade ago. The M-series chips were the last time Apple genuinely surprised the industry. That was already half a decade ago.

The MacBook Neo is interesting. It’s hyped. Not because it’s a new category or Apple changing the game. It’s because it’s cheap. For Apple.

A foldable phone in 2026 is not really innovation anymore. It’s a product line extension.

But that’s what Apple does right now, isn’t it? Product line extension. Not much else.

Who this is for

Well, people who want a foldable phone but won’t buy Samsung or Google or Oppo. That’s about it.

That’s a market. Probably. Apple’s ecosystem lock-in is strong, and many iPhone users would never switch to Android just to get a foldable.

So they’ll sell millions. Of course they will. They always do.

But selling millions and being innovative are two different things. The iPhone 18 Pro will outsell the foldable. The foldable is a flex purchase. A $2,300 conversation starter.

What I’d rather see

A real Siri replacement. Real personal AI with privacy embedded.

Not the incremental updates they’ve been shipping, but something that actually competes with what Google and OpenAI are doing in conversational AI.

A Mac that runs local AI models well to be useful for developers. Out of the box, ideally.

And many updates for existing apps. Bug fixes, cosmetic upgrades, stability improvements, and more. I need that.

But none of those are as photogenic as a folding phone.

The price question

$2,300 to $2,900.

For context, that’s a MacBook Pro. More than an iPad Pro. More than 3 great phones that don’t fold.

Samsung’s Galaxy Z Fold is around $1,800. Still expensive. But $500 to $1,100 cheaper for a phone that already folds, already has three cameras, already has six generations of refinement.

Apple’s pitch will be: no crease, better build quality, iOS. Samsung’s pitch will be: we’ve been doing this since you were still making flat phones.

Both are fair.

The Bottom Line

Apple will ship a foldable iPhone in September. It’ll be beautiful. It’ll be expensive. It’ll sell.

But it won’t be really anything special, I feel like. Could prove me wrong. We’ll see.

For now, I don’t see it redefining a category. It won’t make you think differently about what a phone can do.

It’ll just be an iPhone that folds.

Maybe that’s enough.