How Many Anki Cards Should You Make From One Video?

Make as many Anki cards from a video as you can review reliably—not as many as you can extract. Before mining, choose a review-time budget, estimate the cost of each new card, and stop when the best remaining sentence is less useful than the reviews it will create.

For one video, that might mean a few carefully chosen cards or none at all. The useful number depends on the video, your level, the card type, and the review workload already waiting in Anki.

There is no magic number per video

A fixed rule such as “ten cards per video” sounds convenient, but it ignores the variables that determine whether those cards will help.

Start with the video itself. A short tutorial may contain three expressions that matter to you; a long interview may contain dozens. Duration does not tell you how many moments are clear, memorable, or relevant.

Then consider your learning goal. If you are practicing listening recognition, one useful clip can become one focused card. If you generate separate listening, reading, and production cards from the same note, a single saved sentence can create several scheduled cards. Count cards, not just sentences, when estimating the future workload.

Your level changes the calculation too. A beginner may need more explanation for one sentence, while an advanced learner can process a familiar structure quickly. Neither benefits from saving every unknown item.

Use three filters before a sentence enters your deck:

• Value: Do you expect to encounter or use this language again?
• Clarity: Can the card test one identifiable thing?
• Cost: Can you prepare and review it without displacing more important study?

If a candidate fails one of these tests, skip it. The point of sentence mining with Anki is selection, not transcript conversion.

Calculate a card budget from review capacity

Work backward from the time you are willing to spend reviewing. Do not treat the resulting number as a scientific constant; it is a planning estimate you will correct with your own review data.

Use this framework:

1. Choose a review-time budget for an ordinary day, not your most motivated day.
2. Reserve most of that time for cards already due.
3. Decide how much of the remaining capacity can go to reviews created by new cards.
4. Estimate how long one review attempt takes for the card format you use.
5. Add new cards gradually, then adjust from actual results.

You can express the estimate as:

new-card review capacity = available review seconds ÷ average seconds per review

That figure is not the number of new cards to add. Each can return for multiple reviews, and difficult cards may return more often. Use the calculation to compare plans. If a batch would consume nearly all available capacity under an optimistic estimate, reduce it.

Card design affects the estimate. A short recognition card with one target is usually faster to diagnose than a production card with several plausible answers. Media can make context clearer, but a long clip can also slow every attempt. Choose the card format that matches the learning job, then measure that format rather than borrowing somebody else’s pace.

Before mining a video, write down this checklist:

• Time available for reviews on a normal day
• Current due-card workload
• Number of card templates generated per sentence
• Maximum number of new cards you will trial
• Stop rule, such as “save only the best remaining example”

Anki’s deck options control limits and scheduling behavior, but a limit is a guardrail, not proof that filling it is wise.

A worked example: budgeting one 20-minute video

Suppose Maya watches a 20-minute interview in her target language. She has 15 minutes available for Anki on most weekdays. Existing reviews usually take about 10 minutes, so she does not plan as though all 15 minutes are free.

This is easier when you watch first and select cards in a second pass.

She marks 18 interesting moments while watching, then ranks them instead of turning all 18 into cards.

Maya’s answer is six cards. That is not a universal recommendation; it reflects her workload, one-card template, goal, and willingness to run a small trial.

If the six cards are quick and useful, she can return to the deferred candidates. If they are confusing or crowd out due reviews, she can stop without losing anything: the shortlist has already protected her from importing the weakest material.

Let review results set the next budget

Your first number is a hypothesis. After several review sessions, replace guesses with evidence from your own deck.

Pay attention to cards that take unusually long, repeatedly fail, or produce an “I sort of knew it” response. Those signals may point to an unclear target, a clip with poor boundaries, too many unknown words, or a prompt with multiple valid answers. Edit, simplify, or suspend the card instead of assuming that more repetition will repair its design.

Anki search can help you inspect patterns. The official search documentation describes filters such as deck, creation date, and card properties. You might isolate cards created from a particular video with a source tag, then compare their lapses and review experience with other mined material.

After reviewing a batch, ask:

• Did I finish due reviews without rushing?
• Which cards delivered useful recall rather than trivia?
• Did one note unexpectedly generate multiple cards?
• Which selection rule would have excluded the weakest card?
• Should the next video’s budget go up, stay level, or go down?

The sustainable target is not the largest batch you can produce. It is a repeatable loop in which watching supplies a small number of worthwhile prompts, and reviewing teaches you what deserves to be mined next.