What's actually in a bookmarks.html export

Chrome, Edge, Brave, Firefox, and Safari all still speak a format Netscape shipped in the 1990s. In Chrome it is Bookmarks manager → ⋮ → Export bookmarks. The file is HTML. It is not JSON. It is not your Google account. It is a local snapshot of one profile.

Most "import your bookmarks" pages skip the file and jump to a signup. That is backwards. You should be able to see what you are about to import.

The bookmark file analyzer does that in this tab. Nothing is uploaded. This post is what the file actually contains.

The shape

A Netscape bookmark file is a nested definition list:

<H3> starts a folder
<DL> is the folder body
<A HREF="..."> is a bookmark
ADD_DATE is a Unix timestamp (seconds, or milliseconds if someone used a tool that multiplied by 1000)
TAGS appears on some Firefox/Raindrop exports
ICON can be a huge data URL. Ignore it. It is a favicon, not the page.

Folders have no IDs. The path is the nest: Work / Docs / 2024. Empty folders are easy to lose on a round-trip because a flat list of <A> tags has nothing to hang them on.

What "duplicate" should mean

Chrome will happily save https://example.com/a?utm_source=twitter and https://www.example.com/a/ as two bookmarks. They are one page.

A useful deduper lowercases the host, drops www, trailing slashes, and tracking parameters (utm_*, fbclid, gclid, and the usual click IDs), then groups. The duplicate bookmark finder does that locally and lets you download a cleaner Netscape file.

Do not trust a tool that "checks if the URL still loads" by sending your list to a server. That is a copy of your reading history.

Dates, depth, domains

ADD_DATE is how you notice the 2019 pile. Folder depth is how you notice the taxonomy that looked smart in 2021. Domain counts are how you notice YouTube ate the bar.

None of that requires an account. The analyzer shows it. Import into Markwise only if you want a library you can ask later — Pro bulk import, 3-day trial, card required. Free still lets you save new links one at a time.

If the file is a Pocket CSV instead

Pocket's export was a ZIP of CSVs, not Netscape HTML. Pocket shut down in July 2025; the export endpoint died on 8 October 2025. If you still have the ZIP, unzip it and use the Pocket export converter for a bookmarks.html, or import the CSV into Markwise directly.

Practical order

Export from Chrome. Keep the original.
Open it in the analyzer. Note duplicate groups and tracking URLs.
Download a clean HTML if you want. Import that, not the noisy original.
Only then decide whether the destination is Markwise, Raindrop, or a self-hosted archive.

The file is yours. A bookmark product that will not show you the file before it eats it is not a librarian. It is a hopper.

What's actually in a bookmarks.html export