Turning a 50-slide lecture into 20 cards

If you end up with 100 flashcards from a single 50-slide lecture, you aren't studying. You are transcribing. Within three weeks, that habit will lead to a review queue of 400 cards a day, and that is usually when people quit spaced repetition entirely.

The goal is to find the information density of the lecture. For most STEM and humanities courses, a healthy ratio is roughly one card per two or three slides. If you have more than one card per slide, you are likely testing facts instead of concepts.

I work on Nebulearn. I built it because I was tired of the "transcription trap" where I spent two hours making cards for a one-hour lecture. I have a conflict of interest here, but the math of card bloat applies no matter which app you use.

Why the 1:1 ratio fails

When we see a bullet point on a slide, our reflex is to turn it into a question. If a slide has six bullet points, we make six cards.

The problem is that slides are designed for a presentation, not for your long-term memory. A professor might use three slides to build up to a single conclusion. If you make cards for the setup, the middle, and the conclusion, you are memorizing the "scaffolding" rather than the actual building.

This ratio changes based on what you study. A math lecture might have 20 slides of derivations that result in just two cards: the final formula and the conditions for using it. A history lecture might be denser with names and dates, but even then, the narrative usually clusters around a few key "why" questions.

Major	Typical Slides	Target Card Count	Ratio
Mathematics	40	5-10	1:4
Biology / Med	60	30-40	1:1.5
Engineering	50	20	1:2.5
History / Psych	40	15-20	1:2

Identifying filler slides

To hit these ratios, you have to learn what to ignore. In a typical 50-slide deck, about 15 slides are functionally invisible for flashcards.

First, look for the "Intro and Logistics" slides. The title slide, the learning objectives, and the "any questions?" slide at the end do not need cards. Next, find the "Example" slides. If a professor spends four slides walking through a single chemistry problem, you don't need a card for every step. You need one card that asks for the core principle used to solve that specific class of problem.

Finally, watch out for the "Visual Aid" slides. These are images or diagrams meant to provide context. If the diagram is the thing you need to label, make one Image Occlusion card. If the diagram is just there to look nice while the professor talks, skip it.

If you find yourself stuck, check out my guide on how to make flashcards from lecture notes for a deeper look at the filtering process.

The One Concept per Card rule

The biggest cause of card bloat is breaking a single concept into too many tiny pieces. This is often called the "minimum information principle," but people take it too far.

If you are learning about the Mitochondria, you don't need a card for "What is the powerhouse of the cell?" and another for "Where is ATP produced?" and another for "What organelle has a double membrane?"

When you have three cards for one concept, you aren't learning the concept better. You are just giving yourself three times as much work every morning. One card with a few well-placed cloze deletions is often more effective than four separate Q&A cards.

I wrote about this in stop making easy flashcards. Easy cards feel good in the moment because you get them right, but they don't actually build the mental models you need for an exam.

How your tool affects the ratio

If you use Anki, the friction of creating a card is high enough that it naturally discourages bloat. However, if you use a shared deck like AnKing, you might be inheriting someone else's 1:1 ratio. In that case, the "Suspend" button is your best friend.

If you use Quizlet, the "Learn" mode is session-based. It focuses on getting you through a set in one sitting. Because it doesn't use a true per-card due calendar like FSRS, having 100 cards from one lecture feels manageable on Friday but becomes a nightmare by finals week when you have 1000 cards to "master" in one go.

I work on Nebulearn to solve the "creation bottleneck" without creating "review bloat." When you upload a PDF of your slides, the AI doesn't just copy every bullet point. It looks for the underlying concepts. It tries to group related facts into a single, cohesive card so you hit that 1:2 ratio automatically.

If you are currently spending more time making cards than actually studying them, you might need a workflow that filters the noise for you. This is the core of the time-saving study app bottleneck: the faster you create, the more you have to review.