How to Start a Blog That's Still Running in 10 Years

Starting a blog takes an afternoon. Still having one in ten years is a different problem, and almost none of the difficulty is technical.

Why most blogs stop

Abandoned blogs usually share a small set of causes, and none of them is "chose the wrong platform":

• The subject was too narrow. Six posts in, everything worth saying has been said.
• The subject was too broad. No particular reason to write any given post, so none gets written.
• The pace was unsustainable. "Twice a week" survives about a month against a full-time job.
• Nobody read it, and that was the whole point. If the only motivation was audience, the first quiet year ends it.
• Publishing became a chore. Any workflow with enough friction eventually loses to doing nothing.

The decisions below are aimed at those five failure modes specifically.

Decide what it's about

Narrow enough that someone can tell what they'll get; broad enough that you can write a hundred posts. That balance is the single highest-leverage decision, and it deserves more than an afternoon — Choosing a Blog Subject You Can Sustain covers how to test one before committing.

The quick version: write down ten post titles you could start today. If you can't reach ten, it's too narrow. If the ten have nothing in common, it's too broad.

Pick a platform honestly

The right platform is the one matching your actual technical comfort — not the most powerful, and not the most minimal.

• Hosted (WordPress.com, Ghost, Bear, Substack) — someone else handles servers, updates and security. You trade control and pay monthly. Correct choice if you want to write rather than administer.
• Self-hosted WordPress — enormous ecosystem, complete control, and a real maintenance obligation: updates, backups, and the plugin surface area that comes with it.
• Static site generators (Eleventy, Hugo, Astro) — fast, cheap, minimal attack surface, and files you fully own. Requires comfort with a build step and version control.
• Hand-written HTML/PHP — entirely viable for a small site. This one runs on a handful of PHP templates and plain data files, with no CMS or database.

Starting without paying

You can start at zero cost, and for testing whether you'll actually keep writing, that's often the sensible move. Most hosted platforms have a free tier, and static sites can be hosted free on several services.

What free tiers typically cost you instead: a subdomain rather than your own domain, limits on customisation, and sometimes ads you don't control. The first is the one that matters — see below.

Free is a good way to answer "will I still be doing this in three months?" before spending anything. It's a poor place to stay once the answer is yes.

Own the address

If you're at all serious about the archive lasting, buy a domain early. It's the cheapest insurance in publishing.

The reason is portability. On yourname.someplatform.com, every link and search result you accumulate belongs to that subdomain. When the platform shuts down, changes terms, or you outgrow it, that value doesn't move with you. On your own domain, you point it somewhere else and everything follows.

You don't need hosting to own a domain — see What Is a Website? for how the pieces relate.

Write before you launch

Write three to five posts before the site is public. Not for search engines — for you.

Writing five posts is the honest test of whether the subject has depth and whether you enjoy the work. It's much better to discover in week two that you have nothing to say than in month four with a domain, a theme and an announcement behind you.

It also means a first visitor arrives to something worth reading rather than a single "welcome to my blog" post.

Pick a pace you can hold

Choose the frequency you can maintain in your worst month, not your most enthusiastic one. Once a month, consistently, beats twice a week for five weeks and then silence.

There's no SEO penalty for publishing monthly. There is a real cost to a blog whose most recent post is eighteen months old — it reads as abandoned, which is worse than a small archive that's obviously still tended.

What to ignore

Things that consume early enthusiasm and return nothing — several of which the 2020 version of this article recommended:

• Link-building schemes. Guest-post swaps, directory submissions, forum-signature links, comment links. These were common advice years ago; they now range from ineffective to actively penalised.
• Post-count thresholds for ad networks. "You need 20–30 posts for AdSense approval" is folklore. Approval assesses content quality and policy compliance, not a count.
• Perfecting the design first. A default theme with readable text is fine. Nobody left because of your typography; they left because there was nothing to read.
• Choosing analytics on day one. Traffic will be near zero for months. Set it up when there's something to measure.
• A launch announcement. Publish, keep going, tell people when there's a body of work worth pointing at.

The blogs still running in ten years are, almost without exception, the ones whose authors found the writing itself worthwhile. Everything above is in service of protecting that.