How to write flashcards that actually stick

A spaced-repetition schedule is only as good as the cards it's scheduling. A vague or overloaded card gets graded inconsistently — sometimes Good, sometimes Again, for no reason related to how well you know it — and an inconsistent grade means the scheduler can never settle on the right interval. Most of what makes a card "good" comes down to a handful of habits.

The minimum information principle

The single biggest improvement you can make to any deck: ask one thing per card. A card that asks for three facts at once forces you to decide which grade to give when you got two of three right, and that decision is where consistency breaks down. Split it into three cards instead, and every review has one clean pass-or-fail answer.

This feels like it makes more work up front — more cards to write, more reviews to do. It doesn't, in practice: the total amount you need to review to know three facts is the same either way, but three clean cards get graded correctly and scheduled at three different, appropriate intervals, while one overloaded card gets scheduled wrong for all three facts at once.

Before and after

Cloze deletions: fill-in-the-blank without extra cards

A cloze card takes a sentence you'd otherwise have to write as a Q&A pair and blanks out the part you want tested, in place. "The powerhouse of the cell is the {{c1::mitochondria}}" shows as "The powerhouse of the cell is the [...]" on the front and reveals the answer on the back — same sentence, no rewriting into question form. Flashcards supports cloze notes directly, including cards with more than one blank in the same sentence, and Anki decks that already use cloze notes import with the formatting intact.

Cloze cards are especially good for material that's naturally a sentence rather than a fact pair — definitions, formulas embedded in context, steps in a sequence — where forcing it into "question / answer" would be more awkward than the fill-in-the-blank already is.

Use examples, not just definitions

"What is the minimum information principle?" answered with an abstract definition is harder to hold onto than a concrete instance. Where you can, phrase the back — or a second card — around a specific example: not just what a term means, but what it looks like when applied. A worked example gives your memory something more anchored than a phrase you memorized without really using.

Two-way facts: make a reversed card, not a harder one

Some material is genuinely symmetric — a word and its translation, a country and its capital, a formula and its name — and you want to recall it in both directions. Resist the urge to cram both directions onto one card ("Capital of France, and what country has Paris as its capital?"). Instead make a second, reversed card: front and back swapped, scheduled completely independently. It's common to know a fact well in one direction and shakily in the other — reversed cards let the scheduler notice that and treat the two directions as the different skills they actually are, rather than averaging them into one card that's graded confusingly either way.

When to suspend a card instead of deleting it

Not every card that's causing trouble should be rewritten immediately. Suspending a card removes it from review without deleting it or losing its history — useful for a card you've realized is ambiguous but don't have time to fix right now, one that duplicates another card you just noticed, or material you've decided to set aside for later. The deck editor lists suspended cards separately so you can come back and either fix or unsuspend them in a batch, rather than losing track of which ones you meant to revisit.

Frequently asked questions

How long should the front of a card be?

As short as it can be while still being unambiguous. If you find yourself adding qualifiers to rule out other interpretations, that's usually a sign the card is really testing two things and should split.

Should I write my own cards or use someone else's deck?

Writing your own forces a first pass of understanding the material, which helps on its own. Someone else's deck (or a starter deck) is faster to start with and fine for facts that don't need personal framing, like times tables or vocabulary.

Is it bad to have very easy cards in a deck?

No — an easy card just settles into a long interval and stops costing you review time. The Easy button exists precisely so trivial cards get out of your way quickly.