Back to blog
8 min readMay 3

Evernote search broke on a 12-year archive, what to do before you lose context

If your Evernote archive has grown over the years and search has stopped returning notes reliably, this is a known pattern on long archives. What users report, why it surfaces at scale, and what to do before you migrate.

Gregory PotemkinProduct Engineering @ dEssence
Evernote search broke on a 12-year archive, what to do before you lose context

Evernote Search Broke on a 12-Year Archive, What to Do Before You Lose Context

Evernote search broken on a long archive is a pattern users describe across community threads. The index lags, old notes drop out of results, and OCR inside images stops returning hits. Export a clean .enex backup before you migrate anywhere, and stop adding new captures to a system you've stopped trusting.

You type a word into Evernote that you know is in there. A client name. A book title. The brand of paint from the bedroom. Nothing comes back. You try again. You try a tag. You try a notebook. Still nothing.

What is actually broken in Evernote search?

This isn't paranoia. Search degradation on long Evernote archives is a pattern users describe across community threads. The shape of it:

  • Queries that used to return everything now return a subset.
  • Recently edited notes sometimes don't appear in results for hours or days.
  • Notes inside PDFs and images stop showing up because OCR indexing falls behind or fails silently.
  • Tag filters and full-text search disagree with each other on the same note.
  • Search results differ between desktop, web, and mobile, because each client maintains its own index.

Users describe the root cause as the index, not the data. On community forums, long-archive users report that the same query returns different results across desktop, web, and mobile clients of the same account, that recently edited notes can take hours or days to appear in results, and that rebuilding the local index on desktop can take hours without always finishing cleanly. Whatever the underlying architecture, the user-observable behavior is that retrieval gets less reliable as the archive grows.

You're not doing anything wrong. You just hit the scale where the tool stops behaving like the tool it was when you started.

What do most people try first?

Rebuild the search index. On desktop, you can delete the local cache and force a re-download and re-index. It takes hours. Sometimes it helps. For some users the same notes that were missing before are still missing after.

Upgrade the plan. Some users assume the limit is a paid-tier issue. Forum threads from paid users describe the same symptoms at scale, so upgrading should not be treated as a guaranteed index fix.

Split notebooks. You move half the archive into a second account or a separate Archive stack. This reduces the size of the active set but does not solve the search problem. It hides notes you might still need.

Try a different client. Legacy Evernote, the third-party clients, the web app. Each behaves slightly differently. None of them addresses the underlying index behavior.

Why is migration its own trap?

The instinct is to migrate. Notion. Obsidian. Apple Notes. UpNote. Bear. Capacities. Reflect. The list is long, and each promises something different than what you have.

But migration carries its own tool-dependent risks. The actual outcome depends on which importer or conversion path you use, and you should check each one before committing:

  • Formatting may break. Tables, embedded files, web clips, and handwritten notes from Penultimate can degrade in transit, depending on the importer. You may end up with notes that are technically present but practically unusable.
  • OCR data may be lost. The text inside images and PDFs that Evernote spent years indexing does not always come along, depending on the destination tool. You may be back to having unsearchable scans.
  • Tags and notebooks can translate poorly. Some tools do not have nested notebooks. Some do not have tags at all. Your organizational layer can get flattened, depending on the target.
  • You inherit a new tool's failure modes. The new app's search may feel better today. In a few years it could behave the way Evernote does now.

Before migrating, run a small test export and import into your candidate tool, and confirm OCR text, tags, and timestamps survive the round trip.

What does an archive actually need to do?

Step back. What were you ever trying to do with Evernote?

You were trying to make sure that the things you noticed, clipped, photographed, scanned, or scribbled would still be available when future-you needed them. The notebook structure, the tags, the stacks, those were never the point. The point was retrieval.

  1. You can dump things in fast, in whatever format the moment gives you, without deciding where they go.
  2. The system reads what's inside, not just where it's filed. A receipt photo, a recipe screenshot, a forwarded email, a voice note, all of it should be readable later by meaning, not just by exact keyword.
  3. You can find things by describing them the way you would to a friend, not by remembering the exact phrase you used when you saved them.

How does dEssence reshape the same job?

dEssence is memory you don't have to maintain, built around the retrieval side of the problem. Save it, forget it, ask for it later. Save through the Chrome extension, the Telegram bot, or the web app at dessence.ai, whichever is closest at the moment. No folders, no tags, no organizing. You drop things in, and you find them later by describing them in your own words.

The save layer is the same speed as a screenshot or a forwarded message. You don't pick a notebook. You don't decide on a tag. You don't title the note. A receipt from a hardware store: forward it to the Telegram bot. A page from a book: snap it. An article on your laptop: one click in the Chrome extension. A voice memo about the contractor your sister recommended: send it.

The retrieval layer is the part that changes. You ask in your own words. The hardware store receipt from when we redid the kitchen. The contractor my sister mentioned. The article that made me think about leaving the job. dEssence reads the content of what you saved, including text inside images and PDFs, and brings back what matches the meaning, not just the literal characters.

What should you do before you migrate?

Whether you stay on Evernote, switch tools, or run a memory layer alongside it, do these first:

Stop adding to a system you've stopped trusting. New captures shouldn't go into the place you've lost confidence in. Pick a new save destination today and start fresh from there.

Don't delete the old archive. Even if you migrate, keep the Evernote account around as a cold store. Search may feel unreliable, but the data is still there. You may need a specific old note someday and you'll be glad you didn't burn it down.

Lower the stakes of the migration. If the new tool has to be perfect on day one, you'll be paralyzed. If it just has to hold what you save from this point forward, you can move at your own pace.

Honest about dEssence

Where it's still rough: dEssence is in beta. The paid tier (Pro (price not finalized)) isn't finalized yet. There's no native iOS or Android app; capture works through the Chrome extension, the Telegram bot, or the web app at dessence.ai. There's no team or shared list feature. Search quality grows with what you've put in, a near-empty account won't feel like much.

Frequently Asked Questions

Why did Evernote search stop working on my large archive?

Users with long archives report that search becomes inconsistent once accounts hold many years of notes, attachments, and OCR'd images: queries miss notes that are clearly there, recent edits take time to appear, and results differ across desktop, web, and mobile clients. The data is still there; the retrieval is what users describe as unreliable at scale.

Will rebuilding the index fix it permanently?

Sometimes for a while, often not. Rebuilds can take hours and don't always complete cleanly. On very large archives, the index drifts out of sync again within weeks.

Is migrating to Notion or Obsidian a fix?

It can be, but migration loses OCR data, formatting, timestamps, and the structure of attachments. New tools also have their own failure modes at scale. Migrate only when you have a clear picture of what you actually need to keep and search later.

Can I keep using Evernote and just add a memory layer for new stuff?

Yes. Many people keep the old archive as cold storage and route new captures into a memory app like dEssence. Over time, the active set of what you actually need lives in the new place; Evernote becomes the basement.

Does dEssence handle PDFs, images, and voice notes?

Yes. It reads text inside images and PDFs, transcribes voice notes, and indexes them by meaning. You don't have to remember the exact words you wrote when you saved something.