Note Apps Compared by What Survives When the Company Doesn't

Most "Best note app 2026" listicles score tools by features and price. This one scores them by one question: if the vendor were gone today, what plain-readable artifact would you still have? Call it the File Format Precedence Test — three vendor-observable axes that split twelve current tools into four survivability tiers before any subjective judgement enters the room.

The test is not a new idea. Steph Ango, Obsidian's CEO, argued the frame in his 2023 essay File over app: "Apps are ephemeral, but your files have a chance to last."¹ The four Ink & Switch researchers behind the local-first paper put the same point in engineering terms: "'Old-fashioned' apps continue to work forever, as long as you have a copy of the data and some way of running the software. Even if the software author goes bust, you can continue running the last released version of the software."²

What follows is that principle applied — as a buyer's rubric, before the disaster — to the apps you can download this week. The companion piece Export before you're forced out is the emergency drill during the shutdown; this is the filter before the sign-up.

Why the file format matters more than the feature list

The unit of ownership is the file, not the app, and the format decides whether the app is disposable or load-bearing. Plain Markdown in a folder: any editor from 1985 to 2085 opens it. Rows in a proprietary block-database: the vendor's server is the only reader. The former is a substrate; the latter is a hostage situation.

The graveyard makes the point without embellishment. Mozilla announced that "Pocket has shut down", with the API disabled on November 12 2025 and the export window closed the same day — the announcement had gone out on May 22 2025, giving users under six months from notice to complete data-deletion³. Skiff (Notes, Pages, Mail, Calendar) posted the same shape: "On Feb 9th, we announced that Skiff was acquired by Notion and that services would be sunset on August 9th with email forwarding enabled for those who chose to be extended to Feb 9th 2025."⁴ Athens, the open-source Roam clone that promised the file-first alternative, froze at "UPDATE: Athens is no longer being actively maintained"⁵.

Killed by Google — Cody Ogden's tracker — lists 306 discontinued Google products as of today⁶. Google Keep is not on that list. Neither were Google Reader, Google Answers, or Google Bookmarks the day before their announcements.

The pattern is not a vendetta against any single company. It is the shape a modern SaaS note-tool takes at end-of-life, and the axis that predicts what you get to keep is the file format. This is why the same essayists who care about reading your notes in 10 years care about how they are stored today.

The File Format Precedence Test: three axes, all vendor-observable

Every note app can be scored on three questions whose answers live in the vendor's own documentation. First: what does the on-disk data look like — plain Markdown, an SQLite database, or nothing local at all? Second: what does the vendor's own export produce, and does it round-trip? Third: can you write new notes offline today, without the vendor's server?

Each axis is factual, not aesthetic. On-disk format is what ls returns when you point it at the app's data directory. Export completeness is what the vendor's own help doc says the export contains. Offline authoring is either supported or it is not.

The test is deliberately conservative: an app can be beautiful, fast, and useful and still fail all three axes; it can be ugly and win on all of them. The test is not "which app is best" — it is "which app leaves you with a file you can still read."

The reason to score every app the same way is that the graveyard already scored some of them for you. What remains is deciding, honestly, whether the tools still alive today would leave a plain artifact behind.

The twelve-app matrix

The matrix below rates each app on the three axes, using the vendor's own primary documentation as the source for the on-disk-format and export-format columns. A paid tier does not change whether the file format survives, so pricing is left out of the ranking on purpose. All claims are dated 2026-07-07.

Reading the matrix: four survivability tiers

The matrix sorts itself into four tiers by the first-axis question — what is on disk — and the tiers correspond to what actually happens the day after a shutdown. This section reads the matrix in tier order, not brand order, so the reasoning stays about the substrate rather than the logo.

Tier 1 — Plain Markdown in a folder

Highest survivability. Obsidian, Logseq, Files.md, and MNMNOTE all store notes as .md files in a directory the user chose. Obsidian's own help doc puts the sentence in the plainest form: "Because notes are plain text files, you can use other text editors and file managers to edit and manage notes."⁷ If Obsidian, Logseq, or Files.md announced a sunset tomorrow, cat, grep, VS Code, and every other Markdown reader ever shipped would still open the files.

The engineering argument for this tier is the one Ink & Switch made most cleanly. "An important aspect of data ownership is that you can continue accessing the data for a long time in the future. When you do some work with local-first software, your work should continue to be accessible indefinitely, even after the company that produced the software is gone."² The file format is the accessibility guarantee, and every app in this tier delivers it by default rather than as a paid tier or as an export step.

The point of a matrix that puts four separate apps into the same tier is that the class of "plain .md in a folder" is a class, not a brand; every team that ships to that shape leaves the same artifact behind. Choosing between them is a preference question about the editor; the survivability question was already answered by the format.

Tier 2 — Local database (recoverable but not portable)

Middle survivability. Bear stores notes in an SQLite file at ~/Library/Group Containers/9K33E3U3T4.net.shinyfrog.bear/Application Data/database.sqlite¹¹. Standard Notes stores them client-side under an encryption protocol whose spec names "Argon2id" for the KDF and "XChaCha20+Poly1305" for authenticated encryption¹⁰. Apple Notes uses CoreData.

An SQLite database is not a hostage situation. It is a well-documented, open format; every language has a driver; you can read it with the sqlite3 shell that ships on macOS. What is missing is the "just open the folder" story — a non-technical user has no ergonomic way to browse their notes without the app. The Bear team correctly points at where the file lives, but pointing is not the same as portability. The tier survives an app shutdown; it does not survive a user losing familiarity with the tool that reads its substrate.

Standard Notes deserves a specific caveat. Its encryption spec is public and precise: the server is treated as "a dumb data-store" and cannot decrypt user data¹⁰. That is a strong privacy design, but the file-format question is separate from the encryption question. If you lose your passphrase, the encryption stops the world from reading your notes and also stops you.

Tier 3 — Cloud-native, lossy export

Low survivability. Notion, Google Keep, and Evernote share the same property: no plain artifact exists on your disk during normal use; the vendor's export exists, and the vendor's export is lossy.

Notion's own export doc says it in one paragraph: "Markdown file. Full page databases will be exports as a CSV file, with Markdown files for each subpage. Callout blocks will be exported as HTML, as there is no Markdown equivalent."¹² The vendor typo is preserved verbatim — the point is not the typo but the shape of the export: a Notion database becomes a CSV file, which means the formulas, the relations, and the rollups do not round-trip. Callouts become HTML islands inside your Markdown. Whatever workspace you designed lives on Notion's servers or not at all.

Evernote's ENML XML is more thoroughly documented but no more portable to a plain-text workflow; every migration guide (including sibling piece Cleaning up a vendor export into plain Markdown) reads like an obituary for round-tripping. Google Keep's Takeout produces .html per note, not .md. In every case the export exists — the vendor is not withholding your data — but the export costs a conversion, and the conversion loses shape.

The trap in this tier is that the vendor can be entirely alive and your notes still trapped. Notion is not shutting down. That is not the point. The point is that the substrate is the vendor's schema, not a filesystem, and if the schema changes tomorrow — or if you simply stop paying for the product — the shape of what you wrote is inseparable from the shape of the vendor.

Tier 4 — Frozen

Zero survivability. Skiff, Pocket, and Athens are here for evidence, not as attacks on the teams behind them. Skiff's own root page states: "On Feb 9th, we announced that Skiff was acquired by Notion and that services would be sunset on August 9th."⁴ Mozilla's support KB confirms "Pocket has shut down" and that the "Pocket API was disabled on November 12, 2025, concurrent with the end of the user export window and the beginning of data deletion."³

Athens's GitHub README opens with "UPDATE: Athens is no longer being actively maintained", and points users to a community athens-export tool that converts pages to "markdown and a logseq-compatible directory"⁵ — a hand-off that only works because someone volunteered to build the bridge.

The Athens case is the honest one. An open-source project can still leave you with plain files, because the shape of the export was Markdown all along; the community wrote the tool anyway. A closed-source cloud shutdown, by contrast, is a hard cliff — Skiff users had six months to migrate everything, then a further six months of forwarding, then nothing.

Who should not use the File Format Precedence Test as their only filter

The test is deliberately narrow. A team collaborating in real time on structured data — the reason Notion exists — is not choosing between "plain Markdown" and "cloud database" on equal terms. A knowledge base of relational databases and rollups is a database; porting it to .md files is not a lossless move.

The test also does not measure editor quality. Bear is a lovely editor whose typography and rendering people miss when they leave; the fact that its substrate is SQLite does not make the writing experience worse. Standard Notes' security model is stronger than most of the tier-1 tools; some users choose that trade-off over portability, and reasonable people can.

The 306 dead Google products on the Killed by Google list⁶ are a risk signal, not a moral judgement — Google Keep may well outlive most of the apps in this piece, and a lot of people will be happy with it if it does.

The point of the test is to make the trade-off legible before you sign up, not to end the argument. If you know that your notes are worth reading in 10 years, use it. If you know they are worth reading only until you finish your current project, ignore it.

Frequently asked questions

Which note app will still work if the company shuts down?

The specific app matters less than the on-disk format. Any note tool that stores notes as plain Markdown files in a folder — Obsidian, Logseq, Files.md, MNMNOTE — will still work in the sense that matters: the files remain openable in any text editor. Any cloud-native tool without a plain-file substrate becomes unreachable the day the vendor's servers stop responding, regardless of whether an export existed while it was alive.

Can I still open my notes if a specific note app shuts down?

Only if the on-disk format is one you can read without the app. If the app stored notes as .md files in a folder, yes — any text editor works. If the app stored notes in a proprietary database or on the vendor's servers, you can open them only if you exported before the shutdown, and only in the shape the vendor's export produced. The Pocket shutdown gave users under six months from announcement to the end of exports³.

Which note app can I still open in 10 years?

Steph Ango's version of the answer, paraphrased from his 2023 essay: choose files whose format was already readable on a computer from 1975, and the software of 2035 will still open them¹. Plain Markdown, plain .txt, and plain HTML meet that bar. Proprietary block models, vendor-specific XML schemas, and cloud databases do not; they inherit the lifetime of the vendor.

Does Obsidian really store notes as plain files?

Yes, per Obsidian's own help documentation: "Obsidian stores your notes as Markdown-formatted plain text files in a vault. A vault is a folder on your local file system, including any subfolders."⁷ The same doc adds "Because notes are plain text files, you can use other text editors and file managers to edit and manage notes." This is the vendor's own architectural claim, not a third-party assertion.

How lossy is a Notion export?

The vendor's own export doc lists the losses. Databases become a CSV file with Markdown per subpage, which means formulas, relations, and rollups do not survive¹². Callout blocks become HTML islands. Bulleted lists and headings round-trip cleanly; anything database-shaped does not. The export is a snapshot of your text, not a snapshot of your workspace.

Is Bear's SQLite database safe if the vendor shuts down?

Yes, in the sense that SQLite is a well-documented open format any language can read. The database file lives at ~/Library/Group Containers/9K33E3U3T4.net.shinyfrog.bear/Application Data/database.sqlite¹¹. A future user could open it with the sqlite3 shell that ships on every macOS. That is different from "plain Markdown in a folder" — recovery is possible; browsing is not.

What about Google Keep?

Google Keep is not on Killed by Google as of today. Killed by Google lists 306 discontinued Google products⁶. That number is the honest framing: Google has a track record of retiring consumer products, and Google Keep's users are betting on Google's continued interest. The Skiff users who lost Notes/Pages/Mail/Calendar had six months' notice from a company that had just acquired their vendor⁴.

What is the best note app for data durability?

The tightest answer the File Format Precedence Test produces is not a brand but a shape: plain Markdown files, in a folder on your own disk, with an app that can read and write them offline. Multiple apps ship this shape today — Obsidian, Logseq, Files.md, MNMNOTE. Choosing between them is an editor-preference question. Choosing this class over a cloud-database class is the durability question.

The shape you write in outlives the app that reads it

The comparison is not about picking a favourite. Four separate teams have shipped the same architectural class — plain .md in a folder — and three shutdowns in the last two years have shown what happens to the other classes. The buyer's filter is the same as Ango's emergency drill: choose files you can control, in formats easy to read.¹

If you already made a different choice, export before you're forced out is the drill for making the migration less painful. If you have not chosen yet, run the test.

For a browser-based, local-first take on the plain-.md-in-a-folder class, mnmnote.com is one of the tools built to the same shape.