In May 2025, Mozilla announced that Pocket — the read-it-later service used by millions for nearly two decades — would shut down on the 8th of July. Users had three months to export their saved items, highlights, and lists. After the 8th of October, the export tool was disabled. By mid-November, the company confirmed that all remaining user data had been queued for permanent deletion.
Eighteen years of someone's reading list, gone.
Pocket wasn't a small operation. It launched in 2007, was acquired by Mozilla in 2017, and most of the people who used it had been adding to their archive since they were in their twenties. None of them did anything wrong. They paid for a service, then one day the service didn't exist any more.
This is subscription rot, and it isn't rare.
The three stages of a subscription dying
Most subscription apps don't die with a press release on the home page. They die in three stages, on a timetable most users never read.
First: announcement. A blog post, an email, a "we have some news" you nearly delete. The headline almost always thanks you for being part of the journey.
Second: export window. Usually 60 to 120 days. Long enough to look reasonable in a press release, short enough that most people never get around to it. Pocket gave 92 days. Some apps give 30.
Third: deletion. Your account becomes inaccessible, then your data goes to a deletion queue, then it's gone. The exact moment varies. The outcome doesn't.
If you missed the export window — because the email went to a folder you don't check, or because life got in the way that month — the file is no longer yours. It never really was.
What people actually lose
The lazy way to talk about subscription rot is to call it inconvenience. "Oh, you'll have to find a new app." That undersells it.
When Pocket shut down, people lost:
- Reading lists they'd been building since 2007
- Highlights — the bits of articles they'd thought were worth marking
- Tags and lists, which is to say, the structure they'd put on their own reading
- A searchable archive of how their interests had shifted across two decades
This wasn't data in the abstract. It was a record of someone's attention — the thing apps love to call your "second brain". It turns out it was someone else's brain, on someone else's server, on someone else's roadmap.
You don't notice you're renting your records until the rent is due in a currency you didn't agree to.
The simpler thing
There is a class of software that doesn't do this. It runs on your computer. It stores its data on your computer. If the company that made it disappears tomorrow, the file on your machine continues to work — because nothing on a server was holding the lights on in the first place.
People used to call this "buying software". Then we called it "owning a copy". Now we'd describe it as "self-hosted" or "offline-first" or "local-first", and the language has got more complicated than the thing it describes, which is just: a file you have, that does the thing.
It's the same shape as owning a book. You buy it once. It lives in your house. The publisher can go out of business and your copy of the book stays exactly where it was, doing exactly what it always did.
Subscription rot is what happens when you traded that for convenience, monthly, and quietly forgot you'd done it. Most people forget. The companies remember.
If you do nothing else
If you only ever do one thing about your records, do this: keep a copy. Even if you love the app you're using. Even if it's never going to shut down.
Export, monthly. Drop the file on your laptop or a hard drive that isn't plugged into anyone else's billing system. Or use software that stores its data locally to start with, so the export step disappears entirely.
The next Pocket shutdown will surprise the people who didn't keep copies. It won't surprise the people who did. That's the whole article.
OneFileClub makes powerful one-file tools for keeping your own records — habit trackers, training logs, food diaries, sleep logs — that run on your computer, store data on your computer, and don't have a deletion timetable.