The Portability Audit: What Each Platform Actually Gives Back

Here is a test worth running before you commit a year of writing to any platform: find the export button, press it, and open what comes out. Most writers never do this until the day they need to, which is the worst possible day to discover that the file is useless.

This is the audit this desk runs on every platform it reviews, because content portability is the closest thing publishing has to a trust signal. A company that makes leaving easy has told you something about how it plans to behave while you stay. A company that makes leaving hard has told you something too.

The four grades of export

Not all exports are the same kind of object. These are the tiers, from best to worst.

Grade A — a standard file another platform can import. WordPress’s WXR format is the benchmark: XML containing posts, pages, comments and metadata, readable by more publishing software than anything else. Ghost’s JSON export and Blogger’s XML sit in the same bracket. You hand the file to a new platform and it reads it.

Grade B — structured data in an open format, no standard importer. You get real, machine-readable content and some work to do. Markdown archives from the minimalist platforms live here: nothing will import them automatically, but converting them is a scripting task rather than a rescue operation.

Grade C — your text, stripped of everything else. An archive of what you wrote, without the design, without reliable metadata, without the URLs. Medium sits here. The words are yours and that is genuinely something, but rebuilding a site from it is retyping with extra steps.

Grade D — a folder of rendered HTML. Pages with the styling baked in and the content tangled up in it. Technically an export. Practically a screenshot.

The difference between Grade A and Grade C is not a technicality. It is the difference between a move that costs you a weekend and a move that costs you your archive.

The audit

Ghost — Grade A. A complete JSON export of posts, tags, users and settings, plus the fact that the software is MIT-licensed and identical between the managed and self-hosted paths. Moving off Ghost(Pro) is genuinely just moving where the software runs. Strongest position on the list.

WordPress.com — Grade A. The WXR file is the most widely readable export format in publishing. Posts, pages, comments and metadata all come across, and if you have attached a custom domain the URLs are yours too. This single fact carries a lot of weight in our WordPress.com review.

Blogger — Grade A, with an asterisk. A complete XML export of posts, pages and comments. The asterisk is comment threading, which is where imports most often go wrong. Test it specifically.

Hashnode — Grade A+, uniquely. Hashnode will mirror your posts to a Git repository you control, which means you are not relying on an export at all. You already have the content, versioned, somewhere else. No other platform in the field guide offers anything structurally as good.

Substack — Grade B. You can export your posts and, importantly, your subscriber list. Getting the list out is more than most platforms with an audience will permit. Formatting and layout come across less cleanly, and there is no standard importer waiting for the result.

Bear Blog and the minimalist platforms — Grade B, trivially. Everything is Markdown already. There is nothing to extract because there was never anything wrapped around it.

Medium — Grade C. An archive of your text. No design, no URLs, no readers, no list. Medium’s own export is honest about being an archive rather than a migration, which is more than can be said for some, but it does not change what you are left holding. This is the central reservation in our Medium review.

What an export never contains

Even a Grade A export leaves four things behind, and it is worth being clear-eyed about each.

Your URLs. Unless you attached your own domain, every link anyone has ever made to your work points at a domain you do not control. This is the single most expensive thing about a platform move and the easiest to prevent: buy a domain on day one and point it at whatever you are using. It is the cheapest insurance in publishing.

Your readers. Followers on a platform are platform accounts. If the export does not include an email list, the relationship does not travel. Substack’s decision to let you take the list is genuinely unusual and deserves more credit than it gets.

Your design. Themes do not port between platforms. Budget for rebuilding the look, and treat it as an opportunity rather than a loss.

Your comments’ context. Threading, authorship and timestamps are where imports quietly fail. If discussion is a real part of what you have built, test this before you commit to anything.

Own the domain from the first post, and the worst migration becomes an inconvenience.

Run the test yourself, today

This takes fifteen minutes and it is the highest-value quarter hour available to anyone who blogs.

1. Find the export. If you cannot locate it in under five minutes, that is a finding in itself.
2. Run it, now, while nothing is wrong. Exports frequently arrive by email, sometimes hours later.
3. Open the file. Is it XML, JSON or Markdown — or is it a folder of HTML? Grade it against the list above.
4. Check what is missing. Are the comments there? The images, or just links to images still hosted on the platform? The publication dates?
5. Try importing it somewhere. Set up a free trial on another platform and actually feed it the file. This is the only step that proves anything, and almost nobody does it.
6. Diarise it. Twice a year, repeat. Export formats change and nobody announces it.

The three rules

After running this audit across every platform in the field guide, the advice compresses to three lines.

Own the domain. Before the first post if possible, after the fiftieth if not. Everything else about a move gets easier once the address is yours.

Keep a copy the platform cannot revoke. A downloaded export, refreshed twice a year, stored somewhere that is not the platform.

Grade the export before you commit, not after. Fifteen minutes now against a lost archive later is not a close trade — and the platforms that score well on this measure are, almost without exception, the platforms that score well on everything else.