Why Pocket shut down,
and what it teaches.

The timeline

Pocket (originally "Read It Later") launched in 2007 as a Mozilla Firefox add-on by Nate Weiner. It went mainstream as a standalone app in 2012, raised funding, and grew to ~30M registered users at peak. In 2017 Mozilla acquired the company and folded it into Firefox.

The shutdown announcement came on May 22, 2025, with the read-only mode beginning July 8. Full data removal was scheduled for September 8, 2025. Users had ~3 months to export.

Why Pocket closed (the official line)

Mozilla's statement framed the closure around "strategic focus": the company is consolidating on Firefox itself, AI products like Anonym, and privacy-focused tools. Pocket — at the consumer-app edge of the portfolio — didn't fit.

The non-official subtext, as best I can read it from public financials and ex-employee chatter: Pocket was a revenue laggard. Mozilla's Premium tier (which Pocket pioneered in their group) never grew large enough to subsidise the engineering investment. Free-tier ad revenue collapsed in 2023-2024 with the general display-ad downturn.

Read-later as a category has always been hard to monetise: the value to the user is real but episodic. Subscriptions in this space cap somewhere around 2-5% of the free user base (Readwise hit ~6% which is high). Multiply that across Pocket's user count and the math is tight.

What users lost (concretely)

The HTML export Mozilla provided is genuinely useful but incomplete. Here's what I personally got back from my 2,400-item library:

• URLs and titles: all of them.
• Tags: manual tags, yes. AI-generated Premium tags, no.
• Time-added: yes, as unix timestamp.
• Read state: yes (read / unread flag).
• Annotations / highlights: partial. The first version of the export tool truncated them; the patch released June 5, 2025 restored them but only for items saved in the previous 12 months.
• Cached article text: none. Pocket's offline reading store wasn't part of the export.

For someone who used Pocket lightly, the export is fine. For someone who annotated heavily over years, the loss is real. Reading lists are partly retrievable; reading itself is not.

What the closure teaches about the read-later category

1. App-as-storage is fragile.

When the app is your storage, the app's business model is your library's survival risk. Pocket's users were collateral damage to a portfolio re-shuffle they had zero control over. The lesson generalises: if your saved articles live in a cloud database controlled by an org with VC or M&A pressure, your library is one strategy review away from disappearing.

2. File-as-storage outlives the app.

A Markdown file in a folder you own doesn't need the app to survive. The app becomes a viewer / search interface, not the substrate. If the app dies, the files are still there, still searchable by any text tool, still openable in any editor. This is the obvious lesson but it took Pocket's closure for most users to feel it.

3. The consolidation is real.

The read-later space has thinned dramatically:

• Pocket — closed July 2025.
• Omnivore — closed Q4 2024 (acquired by ElevenLabs, absorbed).
• Diigo — alive but stagnant since ~2020.
• Pinboard — alive but bookmarks-first, not read-later.

The survivors are Instapaper (indie since 2018, now stable), Wallabag (self-hosted open source), Readwise Reader (premium subscription), and the file-first newcomers (Shelf and others).

What's next for the space

My read of the next 12-24 months:

• File-first tools rise. The Pocket closure accelerated the "your library lives in YOUR folder" pitch. Tools that ship as Chrome extensions writing to Markdown (or PDF, or any portable format) win the trust of burned users.
• Read-later merges with PKM. The line between "place where I save articles" and "place where I organise knowledge" was always thin. Tools like Obsidian Web Clipper, Notion Web Clipper, and Shelf collapse it entirely.
• AI moves from "tag for you" to "summarise for you." Pocket's AI tagging was useful but bounded. The next wave is on-device summarisation (Chrome's Prompt API, Apple Intelligence) — fast, private, free at inference. This shifts the value of read-later from "save and re-find" to "save, summarise, decide if I actually want to read."
• Subscription pressure stays. Free read-later with cloud sync is structurally hard to fund. Expect $3-10/mo subscriptions to be normal. The free tier strategy has to actually constrain (storage limits, feature gates) or the business fails.

If you're reading this looking for an alternative

See A Pocket alternative built after the shutdown — full comparison of the 7 surviving options as of 2026. Or jump directly to the migration guide for your platform of choice: Pocket → Notion or Pocket → Obsidian.

Disclosure

I built Shelf, a Chrome extension that's one of the post-Pocket alternatives. The opinions above stand on their own — the file-first observation isn't novel, just timely. The commercial bias is acknowledged.