On which Blogging Platform to Post?

Since I heavily researched blogging, I thought I’d share a bit of the research. I thought I’d see how Hairy would do writing. Not the coding Hairy. It doesn’t do that great, but it told me why, and I get it. I don’t think I’ll share that insight though. There are domain specific, I guess you’d call it, AI writers, but I just used the general purpose. I probably wouldn’t do it like this again (I made three posts in this way), just because I don’t think it was time efficient. There’s a ratio between time spent and control over the outcome, and it felt like that ratio wasn’t in balance.

There’s another issue though, and that’s the amount of ideas I generate when interacting with Hairy. It’s too much. Literally too much. The ratio of output to ideas was too low. Look at that. Two ratios. Maybe I’m just having one of those creative periods. Also, I argue with General Airy.

I never can just focus on one thing, mainly because the connection speed is so slow. So I open up another session and start writing German, then we get into this argument about evolution and politics. I was trying to point out that the mutation mechanism might have a prime mover conundrum and Hairy just about boxed me about the ears denying that. Then I used one example from a certain political wing to emphasise a point about people operating on instinct not conscious thought and Hairy seemingly summoned Chuck Norris to roundhouse kick me into oblivion. Honestly, the users who have found their AI agreeing with them so much that they develop an attachment disorder must hold all of the ‘permitted opinions’, lining up completely with the guard-railed AI.

I cannot imagine – I mean this in all earnestness – existing like that. I can imagine a lot, but not that. At any rate, I hope this is useful information. It was a pain in the ass for me to bring about. I added table of contents fuctionality for this post that’ll be useful generally. I haven’t used headers much, but as this is a sort of reference post, it seemed correct to have a toc to navigate the post.

There are already a thousand posts comparing blogging platforms. This is one more. The difference is I actually went and tried to make this decision for months, got stuck, and came out the other side with a clearer sense of what the decision actually is – which is smaller than it looks.

tl;dr: the platform is not your bottleneck. Pick one from the right category for how you actually want to spend your time, and go write. Everything below is just working out what ‘the right category’ means for different kinds of people.

The real question isn’t ‘which platform?’, it’s ‘What type of Mensch are you?’↑

Every blogging tool bundles three separate jobs:

Writing – the interface you type into
Storage – where the content lives
Publishing – how it becomes a website other people can read

WordPress, Ghost, Hugo, Substack, and Beehiiv are five platforms picked as clear examples, not the full field, that take different approaches addressing these concerns. There are dozens more platforms, and almost all of them are a variation on one of these four, such is the nature of categorisation.

Do you want to be a writer, or are you also willing to be a systems administrator?

That’s not a rhetorical question. Some people genuinely enjoy the tinkering – config files, deploy pipelines, themes as a hobby unto themselves. If that’s you, one whole category of tools opens up and is genuinely great. If it’s not you, that same category will quietly eat months of your life without producing a single sentence of actual writing. Be honest with yourself here before you read another comparison chart, including this one.

There’s a second question worth asking alongside that one, and it’s often the more decisive of the two: are you writing to build something – an audience, a mailing list, eventually maybe income – or writing as a pastime, a place to think out loud with no particular endpoint in mind?

The answers to these questions can provide guidance in selecting a blogging platform. Someone building toward a genuine sales funnel – an audience that eventually subscribes, converts, or buys something – is better served by Beehiiv, Ghost, WordPress, or Substack, all four of which are built on the assumption that growth and monetization are the point. Someone writing purely as a pastime is often paying for exactly that infrastructure and never using it – which is where category three’s quieter platforms, and Hugo/Eleventy’s own-it-forever approach, tend to actually fit better.

The three real categories (and one bridge too far)↑

Ignore the marketing names – and ignore the specific platform names too. There are three main directions, and almost everything else, all forty-plus entries deep into any directory you find, is a variation of one of the following.

1. Static site generators (Hugo, Eleventy, Jekyll, Astro)↑

You write Markdown files locally, a tool converts them to HTML, you push the result somewhere it can be served – GitHub Pages, Cloudflare Pages, Netlify.

What you get: total ownership of your files, zero recurring cost beyond a domain, a workflow that fits neatly into git commit && git push, and content that will still open in a plain text editor in 2040, regardless of what happens to any company.

What it actually costs you: you become responsible for hosting, deployment, theming, and (if you’re not careful) an afternoon lost to YAML frontmatter instead of writing. None of it is hard exactly. All of it is there, sitting between you and publishing, every single time, until you’ve automated it away.

Hugo vs. Eleventy vs. Jekyll, briefly: Hugo is a single fast binary with almost no dependencies and strong built-in taxonomies – the ‘buy once, run forever’ option. Eleventy is more hackable and Node-based, better if you like shaping your own structure but drags in a node_modules folder and more moving parts. Jekyll is the oldest and most GitHub-Pages-native, but Ruby dependency management is a genuine headache in 2026 if you don’t already live in that ecosystem. If you’re choosing blind, Hugo is the safer default.

Right for: people who already think in files and folders, who want a body of writing that outlives any single company, and who don’t mind that ‘publish’ means ‘run a build step’ instead of ‘click a button.’

You log into a browser, write in a rich editor, click publish. Hosting, email delivery, and often monetization are bundled in.

What you get: essentially zero setup, a real newsletter engine (which none of the static generators give you natively), and, critically, you can be publishing your first post sooner, rather than later.

What it costs you: you’re renting, not owning. Ghost can be self-hosted (which reintroduces some of the sysadmin burden from category one) or run as a paid managed service starting. Beehiiv and Substack are free to start – you only pay if you turn on paid subscriptions (Substack takes a 10% cut of that revenue; Beehiiv charges a flat monthly fee once your list passes a certain size instead of taking a cut) – and design customization on both is minimal. None of these three make you think about servers.

Right for: people whose priority is getting words in front of readers and building an email list, who don’t especially care about owning the underlying infrastructure, and who’d rather pay a small monthly fee than spend a weekend on deployment.

The WordPress option deserves elaboration, because it’s not really competing with the others so much as looming over the whole category. It runs something like 40% of the web, in two very different flavours: WordPress.com (hosted, closer to Ghost’s deal) and self-hosted WordPress.org (your own server, your own plugin sprawl, closer to category one’s sysadmin burden but with twenty years of accumulated cruft instead of a clean build). The plugin ecosystem is the whole pitch – there’s a plugin for everything – and also the whole cost, since ‘everything’ includes conflicting updates, abandoned plugins, and security patching that never really ends.

The update mechanism itself is trivial – one click in the dashboard, or wp plugin update --all if you’re comfortable with WP-CLI. But ‘just keep things updated’ undersells the actual exposure. Patchstack’s 2026 security white paper¹ found 11,334 new vulnerabilities across the WordPress ecosystem in 2025, a 42% increase on the year before, with 91% found in plugins and 9% in themes – WordPress core itself had only six all year, none of them serious. That part sounds like a solvable discipline problem: keep the plugins current, you’re fine. It isn’t. 46% of those vulnerabilities had no fix available from the developer by the time they were publicly disclosed. Adopting plugins and templates as a matter of course, which is basically how anyone actually uses WordPress, is kind of like walking through no-man’s land.

The problem is so bad that there’s a regulatory fix for this: as of this year (2026), every commercial WordPress plugin sold to European users is required by law to run a vulnerability disclosure program. The report’s own doubt is pointed:

Will they all have the resources to deal with the insane amount of incoming security reports? Probably not.

It’s not that WordPress is unsafe by design, or that plugin developers are careless. It’s that a decentralized ecosystem of mostly small, independent developers is being asked to carry a security burden that scales with the platform’s popularity. In addition, the white paper points out that website (and by extension blogging) activity is increasingly using AI optimised technology, but that comes at a price. WordPress core itself is free, but the plugins doing the real work rarely are – Jetpack AI at over $20/month, a premium page builder, a security plugin, a backup service – and a fully-featured setup can surpass the costs for Ghost or Beehiiv.

Setting up WordPress↑

I thought I should at least sign up and look around WordPress to get a feel for it. Getting a WordPress.com site launched, in practice, is a fairly straightforward linear path – but it’s not a short one. There are prompts to dodge along the way (an ‘upgrade or miss out’ modal appears before you’re even allowed to continue with the free plan), decisions to make about which tier and how much to pay, and once you get access you are presented with an admin dashboard with a full sidebar of its own: Stats, Hosting, Jetpack, Posts, Media, Pages, Comments, Feedback, Appearance, Plugins, Users, Tools, Settings. None of it is confusing exactly, but it’s a lot of surface area to get comfortable with before ‘write and publish’ – there’s a genuine on-ramp of reading and clicking to do first, the kind of thing reflected in the existence of entire sites dedicated to helping people get oriented in the WordPress ecosystem.²

This is presumably not unique to WordPress – every hosted CMS with a GUI-first admin panel probably asks for some version of the same click-through orientation, though this wasn’t tested across platforms here, so take that as an assumption rather than a finding. What does seem safe to say is that anyone who’d rather work from a command line than a dashboard sidebar (probably the minority of bloggers), will either go looking for a different category of platform entirely (categories one and four above) or look for whatever command-line-adjacent access a given CMS happens to expose – WP-CLI, in WordPress’s case, though how much of the admin experience that actually replaces is its own question this post didn’t chase down.

3. Zero-maintenance simple blogs (Blogger, Write.as, Bear Blog, Pagecord, Mataroa)↑

The quiet middle category most comparison posts skip because it isn’t exciting. No newsletter empire, no build pipeline – just a place to put words online with almost no ceremony.

Blogger has been running, basically unchanged in spirit, since 1999. It’s free, straightforward, and it’s outlasted platforms that were supposed to kill it. If longevity and ‘I will never think about this again’ are the priority, it’s a genuinely underrated pick, not a punchline.

Write.as is the closest thing to a spiritual successor to a static-site workflow without the static-site labour: Markdown-native, has a real command-line publishing option, supports a custom domain, includes a basic newsletter, and is run by a small founder-owned company rather than a VC-backed one.

Bear Blog is the one built almost entirely around speed and restraint: no JavaScript by default, minimal CSS, and a free tier generous enough that a lot of users never leave it. The paid tier exists mainly to unlock a custom domain and basic analytics – it’s a donation-shaped upgrade more than a feature unlock.

Pagecord takes a different angle on the same philosophy: instead of a web editor, you write in whatever app you already use – Obsidian, iA Writer, even email – and publishing is closer to hitting ‘send’ than logging into a dashboard. If the barrier between ‘I had a thought’ and ‘it’s published’ is the thing that’s been stopping you, this is the platform built specifically to remove it.

Mataroa goes furthest in the opposite direction from WordPress’s everything-is-a-plugin sprawl: no themes, no styling options, nothing to configure, by design rather than limitation. It’s built and priced by a single developer close to actual running cost rather than for growth or profit, and it’s refreshingly honest about what it can’t guarantee: its own site states plainly that it’s currently a one-person operation, a ‘bus factor of 1,’ and that bigger, more profitable platforms like WordPress and Substack have better odds of still existing in ten years for exactly that reason. Its own FAQ takes that candour further, recommending Substack instead if you publish more than a couple of times a week or actually want to get paid by readers.

What makes Mataroa unusual is that it hedges its own risk the same way WordPress splits into two products: mataroa.blog is the hosted version – zero setup, someone else’s server, the ‘I don’t want to think about infrastructure’ option – and the entire platform is also open-sourced on GitHub, so anyone willing to run their own server gets the identical minimal, no-themes app instead of paying someone else to host it. That’s the WordPress.com/WordPress.org split in miniature, minus the plugin sprawl that makes self-hosted WordPress a genuine burden.

It’s also, in effect, a pre-built version of category four below: someone who likes the idea of ‘own the whole stack’ but doesn’t want to actually design and debug one from scratch can just run Mataroa’s code themselves – a roll-your-own with the rolling already done. Whether anyone would actually do that if the one-person team behind it stepped away is an open question, but the door being built at all, by a company this candid about its own mortality, is unusual.

Right for: almost anyone who’s spent more time researching blogging infrastructure than actually blogging. Which, statistically, is most people who research blogging infrastructure.

4. Roll your own, if you’re insane³ (like me), and have spare LLM tokens↑

There’s a fourth category the tidy comparison charts leave out, because it isn’t really a platform. It’s building the platform yourself – or, per Mataroa above, running someone else’s already-built one on your own server instead of designing your own from scratch.

That’s possible now in a way it wasn’t a few years ago, because an LLM can pair-program the whole thing with you: SvelteKit with the static adapter, Tailwind for styling, Markdown files with frontmatter as the content store, Cloudflare Pages for hosting, and a handful of small Cloudflare Workers for anything that needs to be stateful (a live subscriber count, a contact form, comments and likes, a magic-link login that gates certain posts behind a ‘prove you’re a subscriber’ wall).

That last part is the actual reason to do this instead of just using Hugo. A hand-rolled stack gets you things no static-site generator or hosted CMS offers out of the box: a real paywall (preview shown to everyone, full post served only to verified subscribers via a signed cookie), your own comment system, your own auth, full-text search built as a static index instead of a paid third-party service. If you have a specific, unusual feature in mind, this is the only category where you’re not waiting on someone else’s roadmap for it.

The honest cost, though, is not ‘a weekend,’ no matter how clean the plan sounds in a chat window. A real build like this racks up real production incidents. A git history rewrite went wrong in exactly the way history rewrites go wrong, silently gutting a day’s commits with no remote and no second backup to fall back on. It got recovered, in the end, only by grepping an AI coding tool’s own local file-edit cache for old versions of the lost files – a strange sentence to have needed to write.

One afternoon, a syntax-highlighting theme change appeared to do nothing, and it turned out to be three unrelated bugs stacked on top of each other: a paywalled post serving a frozen snapshot, a malformed code fence silently swallowing half the article into itself, and a missing default-language setting dropping the rendering entirely. Three plausible wrong explanations before the real one turned up. Separately, and more seriously, a framework serialization quirk quietly exposed gated content in a place it was never meant to be visible. Fixed once it was found, it turned up again a month later in a different corner of the same codebase before the pattern was finally stamped out.

None of that is a reason not to do it. It’s a reason to be honest about what ‘I’ll just build it myself’ actually costs, when a chatbot makes the architecture sound like an afternoon’s work and the debugging turns out to be where the real time goes.

Right for: people who have a specific stateful feature they actually need (a paywall, a live-updating widget, custom auth) that no off-the-shelf platform will give them, who are comfortable debugging production incidents alone, and who, frankly, have something close to unlimited patience – and LLM tokens to spend on problems a hosted platform would never have let them encounter in the first place.

One to skip entirely↑

Tumblr deserves a specific mention because it comes up in every ‘what about– conversation. It’s real, it’s active, it has tens of millions of users – but it’s a social/reblog platform whose native unit is a short post in a fast-moving feed, aimed overwhelmingly at a young, casual audience. Long-form essay writing goes there to die quietly, unread, buried under the next reblog. Fine for fandom art and quick asides. Not a blog in the sense this post means; Tumblr seems more like Instagram for words.

The comparison nobody puts in these posts: how hard is it to leave?↑

This is the actual useful question, because most people underestimate that they’ll eventually want to switch, and overestimate how bad that will be if they pick a Markdown-based tool from the start.

Platform	Migration difficulty if you leave
Hugo / Eleventy	Lowest – it’s already just Markdown files
Mataroa	Lowest – built-in automatic Markdown export to GitHub; leaving is closer to a feature than a chore
Write.as / Bear Blog / Pagecord	Low – clean exports, Markdown-native
Ghost	Low-medium – structured JSON/HTML export
Blogger	Low-medium – XML export, surprisingly clean for its age
Beehiiv	Medium – HTML export, needs clean-up to convert
Substack	Medium-high – content moves, your audience and network don’t
WordPress	Highest – plugin- and theme-specific formatting turns export into a small archaeology project

If you genuinely don’t know which category you’re in yet, this table is a reasonable tiebreaker: start with whatever has the lowest cost of being wrong.

A word on SEO, because someone always asks↑

Whichever platform you pick, don’t spend real time optimizing for search engines. That advice would have sounded strange five years ago; it doesn’t now. Organic search traffic to small, independent sites has fallen sharply as AI summaries and AI answer engines absorb more of what used to be a click.

The ‘answer engine optimization’ industry currently selling dashboards for this has no reliable mechanism and no consistent measurement – vendors publish numbers that disagree with each other by a factor of two in the same quarter. Sitemap, clean HTML, and RSS (which every platform above gives you for free) is all the ‘SEO’ that’s worth doing. Being found now runs through people linking to you, not algorithms ranking you, which happens to be the same old boring advice for building an audience the durable way.⁴

Three years paying month to month tallied↑

Every platform’s homepage leads with its cheapest number. None of them lead with the number you’ll actually pay in month 13. The pattern shows up in some form on almost every platform: quote the annual-billing price like it’s the monthly price, throw in something ‘free’ that expires after year one, or attach a feature you actually want (monetization, a real custom domain, plugins) to a marketing upsell. None of this is unique to blogging platforms – it’s the same playbook web hosts and software subscriptions use everywhere – but it means ‘which one is cheapest’ isn’t a number you can read off a pricing page.

The table below compares the paid platforms priced over three years, using month-to-month billing of the cheapest paid tier, not the annual-prepay rate most pricing pages lead with. That’s a deliberate choice: annual billing requires having a chunk of cash on hand up front, which is exactly the thing a genuinely budget-conscious reader – a student, someone between jobs, anyone for whom $150 upfront is a real number – often doesn’t have. Monthly is the version of each platform’s pricing that a tight budget actually has to live with, and it turns out to tell a much less flattering story than the annual numbers do.

Reminder before the table: WordPress.com, Write.as, Mataroa and Pagecord all have free tiers. None charge anything as long as the blogger is okay with a subdomain (yourname.wordpress.com / yourname.write.as) instead of a self-owned domain. Everything below is the cost of the paid tier, i.e. the cost of wanting your own address. Also note that none of Mataroa, Pagecord, or Blogger are set up for the customer to commercialise blog posts.

Assumptions, stated so you can swap in your own numbers:

Domain registration/renewal assumed at $15/year (a plain .com) – except where a platform locks you into a specific TLD for its free year.
No add-ons (premium themes, extra storage, custom email) on any platform.

Platform (paid tier, monthly billing)	3-year total	Details
Blogger + own domain	$45	No paid tier exists at all — only cost is the domain.
Beehiiv Launch (free) + own domain	$45	Free up to 2,500 subscribers, no monetization.⁵
Mataroa Premium + own domain	$72	$9/year flat — includes analytics, image hosting, API.⁶
Pagecord Premium + own domain	$189	$4/month flat.⁷
WordPress.com Personal (monthly) + own domain	$369	$9/month — no free domain or support on monthly billing.⁸
Write.as Pro (monthly) + own domain	$369	$9/month, capped at 500 email subscribers.⁹
WordPress.org + Bluehost (monthly, no term)	$621	$15.99/month — no free-domain perk without an annual term.¹⁰
Ghost(Pro) Starter (monthly)	$678	$18/month flat, 1,000 subscribers before scaling kicks in.¹¹

A note on the numbers above: prices were checked directly against each platform’s own pricing page where possible (Ghost, WordPress.com, Beehiiv, Write.as, Bluehost, Pagecord, Mataroa), supplemented by third-party pricing trackers only where a platform’s own page didn’t spell out a detail plainly (e.g. Bluehost’s exact renewal-rate table). Checked September 2026. Three assumptions were held constant to make the comparison fair: month-to-month billing throughout, deliberately, since that’s the version of each price a genuinely budget-constrained reader actually has to live with rather than the discounted annual-prepay rate most pricing pages lead with; a $15/year domain unless a platform locks you into a specific promotional TLD or explicitly withholds a free domain from monthly billing; and no premium themes, plugins, or add-ons, since those vary too much per person to generalize. Prices, terms, and even which tier includes what change often enough on these platforms that some of the specifics above shifted more than once while this post was being written – treat this table as a snapshot, not a guarantee, and re-check before you commit real money to anything on it.

The pattern, and it’s a genuinely striking one: paying monthly instead of annually doesn’t just cost more per month – on WordPress.com, Write.as, and Bluehost, it strips out every sweetener at once. No free domain. No discount. On WordPress.com, no support. The customers least able to prepay a year upfront are, structurally, the ones paying full freight for the least. Mataroa and Pagecord are the exceptions – small, flat-priced, and honest about it – and it’s the same reason they showed up cheapest in every version of this table: no growth-stage pricing means no one to penalize for paying as they go.

Blogger and free-tier Beehiiv (≈$45) < Mataroa (≈$72) < Pagecord (≈$189) < WordPress.com Personal = Write.as Pro (tied, ≈$369) < WordPress.org + Bluehost (≈$621) < Ghost Starter (≈$678).

What does ‘cheap’ mean for blogging?↑

A pattern emerges: the cheapest platforms in dollar terms are also, not coincidentally, the ones with the least reason to trick you. Blogger has no growth model to fund. Mataroa and Pagecord are solo-built, priced near their running cost, and have no promotional-rate game to play because they never had a ‘regular’ rate to discount from in the first place. The platforms with the biggest gap between headline price and three-year reality – Ghost, WordPress.com, Bluehost – are also the ones with investors, sales funnels, or both, and pricing pages built by people whose job is to get you past the ‘get started’ button before the fine print registers.

Now that the prices are clearer, what do they mean? Ghost’s $18/month buys you a genuinely serious publishing tool, and $678 over three years when you’re making bank or have a staff budget is just another line item. – the point here is that a clear comparison cuts through the marketing haze to allow the potential customer to make objective decisions using his or her own budget constraints. Hmm, isn’t there a story about a brilliant apples marketing campaign? Jerry? Let’s stay focused.

An important note in the research: there are two different business models in play here, which affects how a potential user is priced. Ghost and Beehiiv pricing scales with your subscriber count past a low cap (1,000 and 2,500 respectively) – while WordPress.com’s pricing doesn’t care how many people show up; ‘unlimited visitors’ is stated plainly on its own pricing page. This makes a blow-by-blow comparison nigh on impossible and the extra features offered used as marketing sweeteners to entice the potential customer. There’s no easy way to precisely compare the offers without extensive testing. Write.as sits in between the business models with its Team tier unlimited subscribers and micropayments for monetisation.¹²

What should you choose?↑

If none of the above resolved it for you, here’s the blunt version:

You want full ownership and don’t mind a little setup once: Hugo.
You want a newsletter and a professional feel with minimal fuss: Ghost (managed) or Beehiiv.
You want to be publishing today and don’t want to think about any of this again: Blogger or Write.as.
You need a specific feature no platform offers, and you’re willing to pay for it in debugging time: build it yourself. Just budget in months, not a weekend, and keep backups you actually trust.
You’re not sure: pick the one with the lowest cost of being wrong (see the table above) and start writing. You can always move later – the friction of switching platforms is almost always lower than the friction of never having started.

The platform will not be the reason your writing succeeds or fails. It never was. That’s the one conclusion that every hour of this research kept confirming, and the one I kept avoiding believing until I realised I had spent far too much time – and energy – on the subject.

One more variable worth naming, separate from price: platforms with both a free and a paid tier – Beehiiv, WordPress.com, Write.as, Pagecord, Mataroa – give you somewhere to grow without moving. Outgrow the free plan and you upgrade in place; nothing migrates, nothing breaks. Blogger doesn’t offer that, because it doesn’t have a paid tier to grow into – but that’s arguably the point of Blogger, not a gap in it. If a free, unchanging blog is genuinely all you want, having no upgrade path could be a feature. The upgrade path only matters to the version of you that expects to eventually need more than free – and it’s worth asking honestly, before you pick a platform, which version of you that actually is.

Worth sitting with, too: not every blog is trying to reach a stranger. Someone writing posts meant for their own kids to read one day, years from now, has a genuinely different set of priorities than someone building a public audience; subscriber counts, SEO, discovery, monetization – all the machinery this post has been comparing – are close to irrelevant to that kind of writing. What matters instead is more likely to be permanence and control: will this still exist and still be readable in twenty years, and who else can see it in the meantime?

That second question opens onto an entirely separate set of concerns – Christopher Robin Milne¹³ spent much of his adult life unhappy about a childhood turned into a famous character by a father who, by every account, loved him and meant well – and it’s worth remembering that ‘written with love’ and ‘the subject is glad it was written’ turned out to be two very different questions, decades apart. What’s safe to publish about a child (or anyone at all) in public, whether they get any say later in what was written about them before they could consent to it, whether ‘public’ was ever the right setting to begin with – that deserves its own post rather than a paragraph bolted onto this one.

Bonus information: A word on domains, because it’s the one cost the free tiers don’t mention up front↑

Every ‘free’ platform above is free right up until you want yourname.com instead of yourname.substack.com or yourname.wordpress.com. A domain itself is cheap in absolute terms, but it’s not nothing, and ‘not much money’ and ‘no money’ are different categories if you’re a student, or a dependent. WordPress.com makes the tradeoff explicit: its free tier keeps the subdomain, and a custom domain only shows up once you’re on a paid plan (see the pricing table above for exactly what that costs, and when). Substack and Blogger will let you attach a domain you bought elsewhere without upgrading anything, but you’re still paying a registrar for it either way.

Here’s the case for getting a domain at the outset: a domain you own is portable in a way a subdomain never is. If you ever move platforms, your links, your backlinks, and anything Google has already indexed under that name move with you. A subdomain ties your identity to the platform’s, permanently, until the day you leave and every link you ever built starts pointing at a dead address.

Here’s the case for waiting: none of that matters if there’s nothing to point the domain at yet. SEO compounds off a body of work, not off a URL – a domain with three posts on it ranks for approximately nothing, regardless of how ‘canonical’ it looks. The actual sequence that works is write first, get a handful of people reading, then worry about whether the address is portable. Buying a domain before you’ve published consistently for a few months is a classic way to feel like you’ve made progress on a blog without having written one.

It also matters less than it used to because some platforms bring their own audience with them. Substack, Ghost(Pro), and WordPress.com all have some form of built-in discovery – Substack’s recommendation network and app, Ghost’s newsletter and syndication tools, WordPress.com’s Reader – so you’re not purely dependent on a domain and search rankings to be found at all. Blogger and the fully static options (Hugo, Write.as, Bear Blog) have none of that; a domain there is just an address, not a source of traffic on its own.

So: if money’s tight, or inspiration hasn’t yet struck, skip the domain for now and use whatever subdomain your platform gives you for free. Buy the domain later, once you know you’re actually going to keep writing – which, going back to the very first line of this post, was always the only decision that mattered.

Honourable mentions: the long tail↑

If you search around, you’ll find a list like Manuel Moreale’s directory of blog platforms – over 40 entries at last count. Almost none of them are a fourth category. They’re variations on the three above, and most differ from their closest sibling in aesthetic or community vibe more than function. A few worth naming, because people do search for them by name:

Publii – category one, but with a GUI instead of a config file. It’s a free, open-source desktop CMS: you write locally, and it builds a static site you publish to GitHub Pages, Netlify, S3, or your own hosting via FTP – no database, no PHP, no admin panel exposed on the public internet. Basically Hugo’s ownership model for people who’d rather click than type YAML.

Kirby – technically self-hosted like WordPress, but flat-file instead of database-driven: content lives in plain text files, and folders are pages. It’s a paid licence, PHP-based, and squarely aimed at developers/designers building client sites – overkill for a personal blog unless you’re already fluent in that world.

Textpattern – a PHP CMS from the early-2000s blogging wave (same era as Movable Type), self-hosted and free. Still actively developed, with infrastructure support from DigitalOcean, BrowserStack, and 1Password – not an abandoned relic, just a smaller, quieter community than WordPress’s. Category two’s sysadmin burden, category three’s minimalism, if you want a CMS with two decades of stability behind it rather than the newest thing.

Nicheless – closer to Twitter/X than a blog: a micro-blogging platform for short, unfiltered posts rather than long-form pieces. Worth knowing it exists mainly so you don’t confuse it with the blogging platforms above – it’s solving a different problem.