Where the blog lives is an ownership decision, not an SEO one

The subdomain question is almost always argued as a ranking question, which is why it usually gets decided badly. Google answered the ranking half on the record, and its answer has not moved since. What is left is a decision about ownership: where your archive accumulates its links, and whether you will still control that address in ten years.

Google already settled the ranking question

In its SEO Snippets series, John Mueller took the question directly — subdomain or subfolder, which is better for SEO — and answered that Google Search is fine with either, that URL structure changes take time to settle, and that you should therefore pick a setup based on your server and your longer-term plans. When SEOs pushed back hard on that answer, Danny Sullivan's reply was that the better option depends on the site, and that all things are not equal between one site and another. Search Engine Roundtable covered the exchange at the time.

Treat that as settled without treating it as unimportant. It is settled in one narrow sense: the shape of the URL is not the lever people believe it is. The folk theory that a subfolder ranks better because it inherits authority from the root domain has the same structure as the theory that Google demotes machine-written text. Both are widely repeated, both are intuitively satisfying, and both are contradicted by the company's own published guidance. We argued the second one at length.

So if rankings are not the reason to care where the blog lives, stop litigating rankings. There is a reason to care, and it is considerably more expensive to get wrong.

What you are actually choosing is who holds the address

A marketing subdomain is not one thing. A blog subdomain pointed at a server you control is a routing choice, reversible on a Tuesday afternoon. A URL on a publishing platform's own domain is a tenancy. In a browser address bar the two look almost identical. The difference appears the day you want to leave.

Every post you publish accumulates a set of things attached to its address rather than to its text. Inbound links from other people's articles. Citations inside AI answers that name the URL. Bookmarks, newsletter archives, a Slack message from two years ago, and the internal links your own archive has been quietly building. None of that is portable by copying markdown into a new CMS. It is attached to the string.

When you control the address, you can move the content and carry the accumulated signals with you. When someone else controls it, you can move the content and leave the signals behind.

Moving costs a year, and only if you can serve the redirect

Google's site move documentation is unusually concrete about the price. It tells you to keep redirects "for as long as possible, generally at least 1 year", because that window is what lets Google transfer signals to the new URLs, including recrawling and reassigning the links that other sites point at your old ones. It also warns that for small and medium sites it can take a few weeks or more before the new URLs start showing instead of the old, and longer for large ones.

Read that as a cost estimate rather than a procedure. A domain move is not a weekend of find-and-replace. Done correctly it is a year of maintained redirect infrastructure and a quarter of degraded search performance, and that is the good outcome.

Now notice the precondition sitting underneath the whole plan. Every step assumes you can serve a redirect from the old address. If the old address belongs to a platform you have stopped paying for, you cannot. There is no migration path to execute — only a fresh start, with an archive's worth of links pointing at a page that is gone, or at whatever the vendor decides to put there next.

There are good reasons to use a subdomain, and none of them are rankings

We are not arguing for subfolders. Plenty of teams have an honest engineering reason to put the blog on a subdomain: the marketing site is a static build and the blog needs a database, the main application cannot absorb another route, the security boundary between them is real. Those reasons are sound, and Google has told you on the record that they cost nothing in ranking terms.

The test is not subdomain versus subfolder. It is whether the DNS record is yours. A blog subdomain on a vendor's infrastructure, with a CNAME you control pointing at them, passes the test — you can repoint it and serve redirects from your own server the same afternoon. A URL on the vendor's own domain fails it at any price, because the thing you would need to change is not yours to change.

One question decides it. If this relationship ended badly tomorrow, could I serve a redirect from this address? It takes ten seconds to answer and it is the only genuinely irreversible part of the decision.

What this changes

Choose the address once, on the assumption that you will keep it for a decade, and then stop thinking about it. A blog is the rare marketing asset that gets more valuable the longer it sits still. The archive is the asset, and an archive is a collection of addresses at least as much as it is a collection of text.

If your blog already lives somewhere you do not own, the move gets more expensive every month you keep publishing into it. That is an argument for moving soon. It is not an argument for never moving.