How to Make Flashcards That Actually Work (Not Just Feel Productive)

You've made 200 flashcards. You flip through them every night. And somehow, in the exam, half the answers won't come.

This happens because most flashcards test the wrong thing. They test whether you recognize an answer when you see it — not whether you can produce it from nothing. Recognition is easy. Recall is what the exam actually demands. If your cards only ever ask you to recognize, you're training for a skill you won't be tested on.

Here's how to fix the cards themselves, not just how often you review them.

Cut definition-only cards

"What is a perceptron?" with a one-line definition on the back feels like a flashcard, but it barely tests anything. You read the term, your brain half-remembers the shape of the definition, you flip the card, and it looks close enough. You mark it "easy." You didn't actually retrieve it — you recognized it.

Better cards force you to produce something specific:

• Ask for an example, not just a definition ("Give an example of a linear vs non-linear activation function")
• Ask you to explain why, not just what ("Why does ReLU avoid the vanishing gradient problem that sigmoid has?")
• Ask you to apply the concept to a new case you haven't seen written down before

If the answer on the back is a single word or phrase you could guess from the shape of the question, rewrite the card.

One fact per card, not five

Cramming a whole paragraph onto one card feels efficient — fewer cards to manage. But it means you can "half-know" a card and still mark it correct, because getting 3 of 5 facts right feels close enough.

Split it. One card, one fact, one clear right or wrong. This also makes it obvious which specific piece of knowledge is shaky, instead of a vague sense that "that topic" needs more work.

Rate honestly, not hopefully

Most people mark a card "easy" the moment they recognize the answer, not the moment they actually produced it unprompted. Be stricter with yourself:

• If you hesitated more than a couple of seconds, mark it "hard" — not "okay"
• If you needed a hint from the wording of the question, it wasn't a real recall
• Cards you keep marking "easy" that you still get wrong on a real quiz need to be rewritten, not just reviewed more

The rating only works as a signal if it's honest. Inflated ratings just hide the gaps until the exam finds them for you.

Space them out, don't binge them

Reviewing the same 50 cards five times in one sitting feels thorough, but most of that repetition is wasted — you're recalling the answer from thirty seconds ago, not from long-term memory. Spreading review across several days, with growing gaps between repeats, is what actually moves a fact into long-term memory.

• Review new cards the next day, then a few days later, then a week later
• Cards you consistently get right can wait longer between reviews
• Cards you keep missing should come back sooner, not less often

If you're doing this by hand, a simple three-pile system (daily, weekly, mastered) gets you most of the benefit without needing software.

Recap

Flashcards fail when they test recognition instead of recall. Fix that by writing cards that demand a real answer, keeping one fact per card, rating yourself honestly, and spacing reviews out over days instead of cramming them into one sitting.

Kora generates study cards straight from your course PDFs and automatically resurfaces the ones you rate "hard" — free to start.