Two students use the same flashcard app, review the same number of times, and get completely different results. The difference is not the app and not willpower; it is how the cards themselves are written. A badly written card never grows past "looks familiar," no matter how many hundred times you flip it. Every review session run on bad cards is study time being quietly thrown away.
The short version: a card worth reviewing does three things. It carries exactly one fact, it forces you to produce the answer from memory, and it asks "why," not just "what." Miss those, and no algorithm can save the card.
Why the card decides the outcome
A flashcard is a personal test item. In Dunlosky's 2013 review, the highest-rated study technique was practice testing, and cards are how you run that test daily. So if the card fails as a test, everything downstream fails: a card whose answer can be guessed from its shape, or whose question is ambiguous, is not a test at all.
The gap between recognition (knowing it when you see it) and recall (producing it from nothing) is the whole game, as we covered in Why You Forget What You Read. A good card is simply a card that forces recall.
The six rules of cards that stick
- One card, one fact (minimum information): split knowledge down to its smallest unit. Big cards create "partially recalled" states that wreck your grading and your schedule.
- Zero ambiguity in the question: reading the front should determine exactly one expected answer.
- Demand recall, not recognition: no multiple choice. Make yourself say it.
- Back every "what" card with a "why" card: definitions plus mechanisms. Understanding-linked memory resists interference.
- Break up list cards: "name the five X" becomes five cards plus one overview card.
- Attach one context or example: concrete examples multiply the retrieval hooks.
Before and after: fixing bad cards
| Bad card (front) | Problem | Fixed card (front) |
|---|---|---|
| "About the sympathetic nervous system" | No defined answer | "When the sympathetic nervous system dominates, what happens to heart rate?" |
| "Key facts of the French Revolution (with timeline)" | Too many facts on one card | "In what year did the storming of the Bastille happen?" and one card per fact |
| "The cooling-off period is 14 days: true or false" | Answerable by recognition | "How long is the cooling-off period?" plus "Why is it set at that length?" |
| "sustain: all meanings" | Partial recall, fuzzy grading | "sustain: what does it mean in 'sustain growth'?" |
Notice the pattern: every fixed card has moved closer to the form an exam will use. Cards are not a storage box for knowledge; they are a rehearsal of the real performance.
The AI era: your job becomes the one-minute check
Honestly, enforcing these rules by hand across hundreds of cards is brutal. At one to two minutes per card, a semester's material costs entire days. That is why the emerging default is to photograph textbook pages or upload PDFs and let AI generate the cards: extraction and question-forming are exactly what it is good at, so the output starts far closer to these rules than a deck written in a hurry.
Your role shifts from writing to a one-minute inspection. Three checks:
- Does any card carry two or more facts? Split it.
- Is a point your professor emphasized missing? Add it as a "why" card.
- Can you answer the question in your own words? If the phrasing is a verbatim copy of the source, reword it.
Good cards plus spaced repetition completes the loop
Even perfect cards fade if you review them at the wrong times. Cepeda et al.'s 2006 meta-analysis shows spaced reviews beat massed ones. With Memly, AI generates cards that follow these rules from your photos, PDFs, or pasted text, and the FSRS algorithm sets each card's optimal review date automatically. It runs on Web, iOS, and Android, so the commute becomes review time on day one. The right moment to review is a science of its own; see The Best Time to Review: Just Before You Forget.
For the full picture of AI-assisted memorization, see the pillar guide AI-Assisted Memorization: How It Works and the Best Tools.
Fix exactly one card today
Most readers will nod, close this tab, and keep flipping the same ambiguous cards tomorrow. It feels like reviewing. The exam score will say otherwise.
One action: take the biggest card in your deck and split it into one-fact cards today, or photograph one textbook page and compare the AI's cards against your own. Memly is free to try with 120 credits, no credit card required.
