How to Make Better Anki Flashcards: Principles for High-Quality Questions

Hi, this is Lesson 3 of 4 in the Anki Fundamentals free course. I hope you like it! Let me know if you have any questions or feedback — I'd like to hear what you think! 🙂

Hi, Al here.

In this post, you’re going to learn the most important principles I learned from experience that will help you create high-quality, memorable flashcards you can use in the long-term.

Other “tips” out there are just that — tiny pieces of advice that doesn’t encompass many other situations.

Meaning, they don’t always apply to your own situation.

With the principles I’m going to show you, on the other hand, you will be able to think of your own “tips” so you can ‘unstuck’ yourself whenever you feel like you’re not creating high-quality flashcards.

In addition, you’re also going to learn the anatomy of a high-quality flashcard.

It’s not what you might think, because in reality, you’re really not going to use a ton of add-ons and card types or styles for this — they’re supplementary at best.

Before anything else, though: If you’ve arrived here without taking the previous lessons, then I want to let you know that this isn’t your typical “top 4 tips” clickbait articles you see online.

What you’re going to learn are evergreen PRINCIPLES extracted from YEARS of combined experience and hour upon hour of testing. 


Please please please don’t expect you’ll be able to internalize them after 3 minutes of mere skimming.

That’s because it will take you a lot of WORK to practice these principles yourself and get them right the first time.

In any case, I made you a quick reference to see them in action: 17 Flashcard Do’s and Don’ts.

You can get the free guide by clicking the red button down below.

With that out of the way, let’s get started.

The Only Thing That Matters

There’s ONE thing in using Anki that if you didn’t respect it, even using Anki would STOP making sense.

Like I said in a previous lesson, using Anki effectively doesn’t mean using hundreds of “hacks”.

It’s about focusing on a few but vital concepts with tenacity; you guessed it— principles.

Learning hacks first is like drifting without even learning how to drive, or trying to learn flashy dribbling without learning the rules of basketball.

As you can imagine, that kind of thing only leads to trouble, and in our case, confusion.

When I was starting out, I figured I should just follow what the majority does —the “hacking” — because they’ve somehow “figured it out”.

You can even see this on YouTube, where productivity gurus are telling you to do “This Productivity Hack Will Stop Procrastination In 2 Minutes” when in reality, it’s just a repackaged version of “suck it up.”

Not really helpful, eh?

And let’s face it: It’s always more interesting to download tons of Anki add-ons to “increase productivity” or “improve memory”, right?

Oh, and don’t get me started with the “Fields…” and “Cards…” option on your Anki—it looks totally advanced, so it might just make you a “more effective Anki learner.”


Nothing matters other than card quality.

Hacks don’t matter if card quality sucks.

Improving how fast you create cards doesn’t matter if card quality sucks.

Even 100 best add-ons that “improve productivity” don’t matter if card quality sucks.

Custom fields don’t matter if your card quality sucks.

Formatting doesn’t matter if your card quality sucks.

Finally, using Anki doesn’t matter if your card quality sucks.

If your cards suck, using Anki will only make you suck — period.

Garbage in, garbage out, as the adage goes.

Figuratively, I think of it like this:

Cards are seeds that grow in the “memory forest”. These seeds eventually grow and root themselves deep in your brain — given they are cared for and are kept in good shape. But a bad seed will always turn into a bad plant. Put plenty of them in a forest and you’ll eventually have a bad forest. Put another way, once you use Anki as your primary learning tool, the quality of your cards will then determine the quality of your learning.

If we buy into this idea, the next question becomes this:

“How do I create high-quality flashcards?”

In the first lesson, I’ve given you some of the most important guidelines to create high quality cards fast without worrying about a lot of things.

But because you’re here, I’ll reveal to you where those pieces of advice are derived from.

It’s from Piotr Wozniak’s 20 Rules of Formulating Knowledge.

Read that article before moving on.

Read it, and make sense of it.

Go ahead, I’ll be waiting for you here.

Okay, you’re back…how did you find it?

Hard to remember?

Well, you can say that again.

That’s because they’re not meant to be memorized — you could, but they’re meant to be practiced and learned through experience rather than just remembered declaratively.

Does that make sense?

From my experience, though, some of the most critical rules in there can be condensed into four major principles.

I’m not saying you shouldn’t learn the 20 rules of formulating knowledge — but treat these 4 principles more as “hooks” at which you can attach some of the 20 rules so you can remember them better.

Got that?

This set of principles is NOT a substitute for the 20 rules, but rather a supplement to aid long-term learning.

Principle 1. Atomicity

The more ambiguity there is, the longer the time you’ll recall each card.

This principle builds upon the minimum information principle; it emphasizes you break down a question into sub-questions until they are, practically, irreducible.

You’ll have more cards that way, but you’ll answer them faster than 1 ambiguous card. It’s ironic, I know—but that’s how it works in practice.

The distinction between MIP and Atomicity is the addition of, well, atomizing your notes.

Breaking down questions reduces ambiguity, but atomizing multiple ideas into a single higher-level idea does that as well.

For example, in persuasive technology, I learned that Perceived Ability increases with number of brain cycles, money, time, physical effort, social deviance, and non-routineness1.

In psychology, these are well-known contributors of sunk costs2.

Because I already know the concepts behind sunk costs, this concept can then atomized into “Perceived ability increases with sunk costs”, and then turned into a question,

Perceived ability increases with what type of costs?

Answer: Sunk Costs

See how that works? Alright, let’s move on.

Principle 2. Avoid Collector’s Fallacy

Effective Anki learning means you have to understand before you commit them into memory. “Own before you collect”.

That also implies you should NOT use shared decks if your goal is to use Anki for learning. One exception, though, is vocabulary. (But even that is debatable.)

Back then, I’ve made the mistake of merely copying paragraphs in my textbooks and then using Cloze deletion to test myself on them.

It was a disaster.

When I got asked about any concept related to the paragraphs I’ve made cards on, I couldn’t even mentally conjure a vague answer.

That’s when I figured the hard way that acquiring knowledge comes first before retaining it.

Since then, everything made total sense.

Principle 3. Leverage Existing Knowledge

This principle might sound repetitive, but it’s too important to build upon existing knowledge because it actually helps you recall better. Two ways to do that:

  1. Create questions that treat older, better recalled cards as “common” knowledge. This is how you really build upon your past learning. The new associations only strengthen past knowledge (via association) such that you won’t have to review previously learned concepts anymore. As a concrete example, if you’re studying Quantum Mechanics, then you won’t need to review “What is the charge of an electron?” anymore. That’s just silly.
  2. Add visuals and personal touch. This forms what I call pseudocontext; Instead of relating the new card to older cards, adding pseudocontext relates them to items in your long-term memory. (Which technically is context itself.) But don’t get me wrong, you can do the same thing to your entire collection as well. My point is pseudocontext makes you remember the card better despite not having a related card to begin with, in the case of “orphaned notes”.

Principle 4. Future-proof

Using a tool to commit knowledge into long-term memory means you’re going to be using it for a long time.

“Long time” doesn’t mean long hours, just to be clear. It means you’re going to leverage what you’ve created again and again for years to come.

Since context per card is fleeting, we should make our cards future proof. The aforementioned principles do that very well, but so should the anatomy of the card itself. (More on this later.)

This isn’t a principle, but it’s highly important not to mention: Don’t use Anki for lists greater than 4 items. Flashcards are poor for memorizing lists—it’s better to use memory techniques instead.

My heuristic is if you’re memorizing items less than 7 items, use mnemonics. If you’re memorizing items greater than that, by all means use the most powerful memory technique of the ancients—the method of loci.

In sum, these four principles are what make a good card good.

You can’t create good cards without knowing (implicitly or explicitly) the fundamental principles that underlie it—it’s not even in the Anki manual.

However, these wouldn’t matter if you don’t know what a good card anatomy looks like.

Let’s talk about that next.

The Anatomy of a Good Anki Card

Having used a ton of flashcards as well as having re-created a couple of decks, I can say that there’s a repeatable process for creating an extremely well-crafted card.

Before that, though, let’s talk about card types.

There are only three card types that matter—you can do any variation of question and answers using these card types alone.

  • Basic. Basic cards are best for bulk uploads and the straightforward question-answer pair. Perhaps it’s the most convenient card type if you want to create hundreds of cards at once. But you won’t be doing that a lot, so I won’t cover it here.
  • Cloze Deletion. Cloze can create the same questions as a basic card type, but you can do “fill-in-the-blanks” stuff here. However, I don’t recommend that for two reasons: 1) It makes it easy to fall for the collector’s fallacy where you collect cards you don’t “own”. 2) I found that “fill-in-the-blanks” type of prompts train you to complete a statement rather than to commit what you’ve learned into memory. I’m not saying you shouldn’t do it at all, but rather you should do it conservatively.
  • Image Occlusion. This is just Cloze deletion applied to images. Of course, you might ask, “If fill-in-the-blanks style doesn’t work for learning, then is it the same for image occlusion?” Not necessarily. Visuals are far more memorable than words — it beats all senses combined. (Here is a humongous set of references to prove that.) Images can stand on their own — it’s kinda weird, but it’s like images already have “context” built-in.

Use either one of Basic and Cloze, in addition to Image Occlusion, and you already have everything you need. Seriously.

Virtually all types of questions can be made with those three.

Others are just variations that suit more specific uses (Vocabulary, for example.) but in no way the 3 basic types can’t do the same thing.

As for the add-ons, you only need 3 as well. Here they are.

  1. Image occlusion. Of course, I wouldn’t be telling this card type if you wouldn’t install it. (duh, Al) I bet you have this one already—it’s pretty popular. In case you don’t, here’s the link. Even if you’re a heavy reader, sometimes you’ll find custom visuals with labels help you recall in a more efficient way than words. That is the perfect time to use it.
  2. Review heatmap. Back then, I believed these heatmaps only served as “motivators” because of the visibility of progress. But it’s actually for a bigger purpose—it gives you objective feedback of your study habits. Feelings are unreliable — they fluctuate, and we can’t predict nor remember feelings accurately — but numbers don’t lie.

    When you fall off track, you’ll immediately know. When you’re not adding new cards per day, you’ll immediately know. Review heatmaps drastically decrease the delay of your feedback loop — making your learning system more “stable” i.e. less likely to fluctuate.
  3. Hierarchical tags. I told you not to use topics as decks in a previous lesson, and this is another reason why. If decks are huge umbrellas, then tags are the smaller “sub-umbrellas”. They don’t really separate what you can review, but tags organize them very well and prove to be extremely helpful for custom study sessions.

Surely, there are other add-ons, but the majority of them aren’t essential. They’re supplementary at best, and only provide miniscule benefits.

We’re trying to stay as lean as possible here so you can focus on what matters — creating high quality cards. (We literally just talked about that…)

Now then, if card quality is defined by the principles, and the cards differ in types, then the card’s anatomy puts the two together.

For your convenience, this is what we’ve previously discussed in the first lesson:

The Ultra-specific question

Just to be clear, statements aren’t questions.

It’s literally a huge tendency for Anki beginners to copy their notes into Anki and then treat them like an identification-type exam.

They literally turn their notes into cards.

It might work, but upon closer examination, you’ll realize it doesn’t test understanding — but rather it merely tests your ability to complete the missing pieces.

Questions, allow you to test your understanding of concepts — not to mention they’re more natural to deal with.

The Answer. Duh.


But again, make sure it’s not a list greater than 4 items. (The number was based on working memory slots, FYI.) If you’re going to do that, make sure you use either the Cloze Overlapper or Memory Techniques.

I prefer memory techniques because then I can create meta-mnemonic cards (a card that helps you remember your mnemonics) to remember them better.

The Excerpt

I got this one from Prerak Juthani back in 2019.

Essentially, you’d want to put a screenshot/excerpt of the material you got your card from.

That way, you won’t have to go back to the material itself to relearn the card in case of a total lapse — there’s the context embedded in an excerpt, after all.

(Obviously you won’t need these in Image Occlusion.)

You can say it’s also a future-proofing element.

The best part?

This allows you to totally ditch notebooks (eventually) and go straight to Anki to revise.

Prerak Juthani is living proof — he’s totally paperless when studying. However, if your use case is to develop ideas or enrich your knowledge, then ditching notes is quite debatable.

Just to be clear, I’m not telling you to ditch notebooks — it’s just one of the luxuries Anki can offer.

At this point, you’re already fully equipped with the essentials — with the principles you’ve learned including the anatomy of a good Anki flashcard, you’ll be able to think of tactics for yourself.

All that’s left now is to turn this knowledge into action.

Lastly, here’s your assignment for this lesson:

Practice the principles and make questions FROM this post so you can remember what you learned here.

Alright, is everything clear at this point?

I want to make sure it is, because I really want to help you study better using this thing. (Which mere tips don’t do for you)

If that’s the case, you can move on to the next lesson by clicking the button down below.

I’ll see you there!

P.S. I hope you’re enjoying this series! Btw, I’m making some more awesome stuff you’ll find at the end of this fundamentals course.

If you’re a college student or a lifelong learner who struggles with remembering what they read and creating good flashcards, (despite this post) I think you’re going to like what I’m going to share with you.


  1. Fogg, B. (2009). A behavior model for persuasive design. Proceedings of the 4th International Conference on Persuasive Technology – Persuasive ’09, 1.
  2. Arkes, H. R., & Blumer, C. (1985). The psychology of sunk cost. Organizational Behavior and Human Decision Processes, 35(1), 124–140.

Leave a Comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.