How to Back Up Your Bookmarks (Not Just Chrome)

Somebody told you to back up your bookmarks, or an app you rely on announced it was closing, or you just looked at a folder you have been filling since 2019 and felt how little would have to go wrong for it to vanish. So you went looking for instructions. The instructions are easy, they take two minutes, and every guide on the first page of Google gives you the same ones.

Here is what none of them tell you. That file you are about to export covers your browser bookmarks and nothing else. If you are a normal person in 2026, your browser bookmarks are the small pile. The big pile is the X threads you saved mid-scroll, the Reddit posts you tapped to keep, the LinkedIn updates you meant to reread, and whatever is sitting in a read-later app. No bookmark export on earth touches any of that.

There is a second problem, and it is the one that actually bites. Even the backup you can make is a list of links. Not the pages. Links. So the day you open it looking for something you saved four years ago, the question is not whether you have the backup. It is whether the internet still has the page.

Everything below was checked against browser and platform documentation on August 8, 2026. The two-minute version comes first, because you should do it, and then the honest part about what it does and does not protect.

How do I back up my bookmarks?

Every desktop browser can write your bookmarks to a single HTML file. The steps differ only in where the menu hides.

Save the file somewhere that is itself backed up, which means a synced folder or an external drive, not your desktop. Then do the step almost nobody does, which is open it. Double-click it. Your browser renders it as a plain clickable list. If it looks right, you have a backup. If it opens empty, you exported from a profile that was not the one holding your bookmarks, which is a more common mistake than it sounds.

Firefox is worth one extra note. Its JSON backup keeps things the HTML file drops, including tags and the separate structure of your toolbar, so if you are staying on Firefox, take that one instead. It will not import anywhere else. We walk through the whole cross-browser transfer, including the duplicates trap, in move bookmarks between browsers.

Where the backup file comes from, and why it is so old

The Netscape Bookmark File Format is the HTML file every browser writes when you export bookmarks. It stores each bookmark as an anchor tag with a URL, a title and a date, wrapped in nested definition lists that represent your folders.

It dates from Netscape Navigator in the mid-1990s. Microsoft documented it because Internet Explorer’s Favorites used exactly the same format, and thirty years later Chrome, Firefox, Safari, Edge and every bookmark app worth using still read and write it. It is, quietly, one of the most successful file formats in computing history, and nobody has ever bothered to give it a real specification.

Read the spec and you can see the whole problem in what the fields are. There is HREF for the link, ADD_DATE for when you saved it, LAST_VISIT, LAST_MODIFIED, an icon. There is no field for the text of the page. There never was, because in 1996 the page was going to be there when you clicked. That assumption is baked into your backup file, and it is the assumption that no longer holds.

What each of your backups actually contains

This is the audit worth doing before you trust any of it. Go through every place you save things and ask what its official export hands you.

Look down the “what it contains” column. Four of the five rows say links. The one platform where people save the most, X, does not appear in the column at all, because X has no bookmark export and its full account data archive gives you your posts, your DMs and your follower list while leaving bookmarks out. The workarounds are real but all third-party, and we test them in how to export Twitter bookmarks.

Reddit and LinkedIn are more generous and land in the same place. Reddit’s data request reaches saves older than the roughly 1,000 its live feed shows you, which makes it the only route to your oldest Reddit save, and it hands you bare permalinks. LinkedIn’s Saved Items.csv is a column of URLs and a column of dates. Both are honest exports of exactly what those platforms stored, which was never the post.

Why a list of links is not a backup

Here is the number that should change how you think about this. The Pew Research Center studied a decade of web content and found that 38% of webpages that existed in 2013 were no longer available by late 2023. A quarter of all pages from that ten-year span had gone.

The social numbers in the same study are worse and much faster. Nearly one in five tweets is no longer publicly visible within months of being posted. One percent vanish within an hour. Three percent within a day. Ten percent within a week.

Now put those two facts together with your backup file. If you exported your X bookmarks through some third-party tool and got a spreadsheet of links, roughly a fifth of that spreadsheet is already pointing at nothing, and you will not find out until you click. The backup succeeded. The archive failed. Those are different things, and only one of them is what you wanted. The mechanism behind that erosion, platform by platform, is in why social bookmarks disappear.

What about the app shutting down?

This is the scenario people picture, and it is the less likely one. It is worth planning for anyway, because the two recent examples show how differently it can go.

Mozilla announced Pocket’s closure in May 2025. The app stopped working on July 8, 2025, then ran in export-only mode until October 8, and Mozilla extended the export window again before deleting data. Whatever else you think of it, users got months of warning and a CSV.

Omnivore went the other way. ElevenLabs acquired the team, the closure was announced on October 29, 2024, and everything was deleted on November 15. That is seventeen days.

Seventeen days is the number to plan around, not six months. An annual export habit survives a Pocket. It does not survive an Omnivore, because your one chance to act falls inside a window you did not know was open, probably while you were not reading that app’s blog. The lesson is not “export more often.” It is that a copy you have to remember to make is a copy that depends on you having been paying attention on the right fortnight. That is the same reason people lose bookmarks they thought were safe.

Does a backup you never tested count?

Run one test, once, and you will know more about your setup than any guide can tell you.

Pick five bookmarks from your oldest folder. Open them. Count how many still load the thing you saved rather than a 404, a paywall, a redirect to a homepage, or a page that has been rewritten into something else. Then pick five saved posts from X or Reddit and do the same.

Most people are surprised twice. They are surprised how many old links are dead, and they are surprised that the survivors are mostly big sites, because small blogs and personal sites are the ones that go offline. The bookmarks you lose are disproportionately the unusual ones, which are the ones you saved because they were unusual.

What a real backup of your saves looks like

The fix is not a better export. It is moving the copy earlier, to the moment you save.

A backup made later can only ever contain what still exists later. A copy made at save time contains what you actually saw. That one change in timing is the entire difference, and it is why archivists talk about capture rather than export. Everything else in this post is downstream of it.

For browser bookmarks, the HTML export is genuinely fine. They are links to pages you chose to keep, the format has worked for thirty years, and there is no content to lose that you did not already accept losing. Export it once a year, keep the file somewhere real, and stop worrying about it.

For social saves there is no equivalent, which is the gap I built ContextBolt to fill, so weigh that as you read this part. It is a Chrome extension that captures what you save on X, Reddit and LinkedIn into one collection, and it stores the content of each post rather than a link to it. So when the author deletes the thread eight months from now, your copy still has the words. AI tags each save by topic as it arrives, and search works by meaning, so “that thread about pricing psychology” finds it without you remembering a single phrase it used.

Basic is free and covers 150 saves with AI tagging and semantic search. Pro at $6 a month makes it unlimited, adds encrypted cloud sync, and gives you an MCP endpoint so Claude or Cursor can read your saves inside a conversation. It does nothing for the bookmarks already sitting in your browser, and it cannot go back and rescue the X bookmarks you saved in 2022. Nothing can. It stops the next four years from going the same way.

How often should you back up your bookmarks?

Schedule is the wrong frame, and it is worth saying why rather than giving you a number.

Browser bookmarks barely need a schedule. Once a year, or whenever you switch machines, covers it. The file is a list of links and the list does not decay just because you left it alone.

Social saves cannot be fixed by any schedule, because the loss happens between exports. If a post is deleted in March and you export in April, your April export is missing it and always will be. No cadence closes that gap. Only capturing at save time does.

So the answer is one yearly export for the browser, and a change in how you save everywhere else. If you want a routine to hang it on, bookmark hygiene has a ten-minute weekly version.

The honest take

The bookmark backup advice everybody gives is a ritual, not a backup.

Exporting bookmarks.html once a year feels responsible, and it protects you against roughly one thing, which is losing the browser profile itself. It does not protect you against the thing that actually empties your collection, which is the slow erosion of the pages behind the links. A file full of dead URLs is a receipt for a library you no longer have.

None of that is an argument against doing it. Go and do it, it takes two minutes, and a receipt beats nothing. Just be clear that you have made a copy of your pointers, not your things, and that the pile you have not backed up at all is probably the bigger one. The next save you make is the only one you can still do something about.