Test a Flashcard Backup Before You Delete or Move a Deck

List what must survive the move

Write down the deck name, export date and number of question-answer pairs. Add any tags, images, audio, reverse cards or review history that matter. A pair of text columns may preserve the questions and answers while omitting everything else.

For example, a language deck might contain 120 notes, with two card directions per note. That can produce 240 review cards. Comparing the exported row count with 240 would be misleading if the format stores one row per note. Record both counts when the application distinguishes notes from cards.

Choose a format for the job

Use a text export when your destination needs plain question-answer content. Use a native backup when you need the application's richer state. Anki's export documentation distinguishes plain-text notes from packaged decks and collection backups; check the documented options for media and scheduling before exporting.

Do not infer content from the extension alone. A CSV file may contain only visible text, and an image filename inside a field does not necessarily include the image. Preserve the corresponding media through the originating application's supported export process.

Create a small set of revealing examples

Choose a few real cards that expose likely problems: an accented word, a comma in an answer, quotation marks, a long response and a card with more than one line. If the deck contains non-Latin text, include it. A test containing only short English words can miss an encoding problem.

Record the expected front and back for each chosen card. In the test destination, inspect the actual pairing, not merely whether some text appeared. A reversed column mapping can produce a full-looking deck with every question on the wrong side.

Restore into a separate test destination

Use a temporary deck or separate application profile where appropriate. Read the import behavior before continuing: some imports update existing notes, while a full collection restore can replace existing data. Never use your only working collection as an experiment.

For the flashcards.website tools, retain the editable source text as well as the downloaded output. The deck builder and CSV converter prepare content; they do not preserve another application's study history. Use the CSV-to-Anki guide below for the detailed field-mapping process.

Compare content and behavior

Count the imported notes or pairs, then inspect the sample cards from both directions where relevant. Check that punctuation, line breaks, accents and mathematical symbols are intact. If formatting was represented as HTML, verify whether the destination renders it or shows the tags literally.

Open included images and play included audio in the destination. Test a card without relying on a network connection when offline access is part of the requirement. A successful text import does not prove that separate media files were included or that their references still work.

Keep a recoverable record

Name the export with the deck and date, and store a second copy somewhere independent of the browser's temporary state. Keep a short note describing the application version, export format, options selected and restore result. Preserve the source until you have completed the checks.

If a test fails, keep the failed output for comparison and export again with one relevant setting changed. Avoid deleting the original to force a clean import. The goal is a repeatable recovery process, not simply a folder containing many similarly named files.

Run a miniature restore with known outcomes

Before testing a large deck, make a separate five-note sample whose contents you can check completely. Give each front a distinct prompt and each back a distinct answer. Include one answer containing a comma, one quotation mark, one accented word, one line break and one optional tag. The point is to exercise the format, not to create useful study material. Keep a plain written list of the five expected pairs beside the import preview.

After export, open the file in an appropriate text editor as well as the intended importer. Check the delimiter and encoding settings in the importer rather than assuming a spreadsheet's display proves the file is correct. Count five notes after import and reveal every back. Then export the imported sample again and compare the meaning of each pair. Different applications may change formatting conventions, so the comparison should distinguish harmless representation changes from missing content or a changed answer. Record the successful settings before using the real deck.

Investigate a mismatch without overwriting evidence

If accents become garbled, first check the selected text encoding and return to the original export. Repeatedly saving the damaged display can make recovery harder because the wrong characters become the new source. If answers split into extra fields, inspect quoting and delimiter choices. If a single card becomes several rows, inspect how embedded line breaks are represented and whether the destination supports them.

If imported notes disappear or update unexpectedly, read the application's duplicate-matching rule. Two notes with the same first field may not be treated as two independent additions. Keep a record of the imported, skipped and updated counts where the importer provides them. Do not change every front merely to bypass matching: that can create duplicate study content and make later updates harder. Test the intended update behavior with one disposable note first, and use the application's documented identifier strategy when repeated synchronization is required.

Separate a shared deck from your private backup

A file for a classmate and a file for your own recovery have different purposes. A shared deck may deliberately exclude personal review history, comments or unrelated decks. A recovery copy may need those details. Give the two outputs distinct filenames and inspect the shared version before sending it. Removing private material from a text export does not prove that another attached package excludes it too.

For a team deck, agree on who maintains the editable master and how corrections are identified. Record the revision date in the handoff and keep a short change list for corrected answers. Before a future replacement, compare one changed card, one unchanged card and one newly added card in the destination. This checks whether the update preserves the intended relationships. A documented small restore test can then be repeated whenever the application, export format or deck structure changes.

Copyable planning template

Deck and export date:
Original note count:
Original review-card count, if different:
Required content and media:
Export format and options:
Separate test destination:
Sample cards checked:
Count and pairing result:
Media result:
Backup storage locations:
Original retained until:

Frequently asked questions

Is CSV a complete backup of my study history?

Usually it is a content-transfer format. Check which fields the exporter actually includes. Use the originating application's native backup when scheduling, templates or review history must be preserved.

Should I test with a full collection import?

Read the application's restore behavior first. A collection import may replace existing data. Use a separate test profile or another documented safe destination and keep the original backup.

Why is my card count different after import?

The source may distinguish notes from cards or generate multiple directions from one note. Compare equivalent counts and inspect the card templates before assuming rows were lost.