Start with a review calendar, then change it according to what you can actually recall. For material learned on day 0, days 1, 3, 7, 14 and 30 are a possible set of return visits. These are dates after initial learning, not gaps to add one after another. They are also not a researched guarantee that five reviews will keep every card in memory until an exam.
The purpose of this guide is to make scheduling practical: know when a card is due, learn from a missed answer and keep the workload small enough to use. For the broader evidence, see spaced repetition.
An example calendar, with the arithmetic visible
Suppose you first learn a small set on 1 November. The dates below show one possible manual plan. Move difficult items earlier and reconsider the schedule if the assessment is sooner.
| Review | Day after first learning | Calendar date | Gap since previous session |
|---|---|---|---|
| First | 1 | 2 November | 1 day |
| Second | 3 | 4 November | 2 days |
| Third | 7 | 8 November | 4 days |
| Fourth | 14 | 15 November | 7 days |
| Fifth | 30 | 1 December | 16 days |
Answer before revealing the back, then compare your response with the expected answer. A partial answer should not be marked correct merely because it feels familiar. After the last date, choose another review based on the difficulty, the result and when you will need the knowledge. There is no automatic point at which the card becomes permanently safe.
What should happen after a missed answer?
A miss gives you information, but it does not identify the cause on its own. Work through three checks:
- Is the expected answer correct? Compare it with the original course source, especially for generated cards.
- Do you understand the idea? If not, revisit an explanation or worked example before testing again.
- Is the prompt clear? Split a question that asks for unrelated things, or add context if several answers would be reasonable.
After checking, try again following a short break and arrange a later return visit. The exact timing can vary. Anki's deck-options documentation describes learning, relearning and scheduler settings; pressing Again does not mean every app resets the card to an identical one-day sequence.
If you repeatedly fail a card, keep a note of why. “Confuses two definitions” suggests a comparison question. “Cannot follow the calculation” suggests worked practice. “Answer is ambiguous” suggests editing the card. Our card-writing guide gives examples of useful prompts.
How many new cards can you afford to add?
Count the future review work as well as the initial writing. Anki warns that consistently learning 20 new cards a day can produce roughly 200 daily reviews. That is an approximate workload warning from its manual, not a prediction for every deck or scheduler.
Try a small set and measure your own session. For example, if 20 cards take ten minutes including checking, 200 similar reviews would take about 100 minutes at that pace, before additional learning. A claim that every student can review 200 cards in fifteen minutes ignores the question type and the time needed to correct errors.
If the queue grows too large, pause new cards, remove irrelevant prompts and prioritise the knowledge you need. Do not make ten cards per module simply because a timetable says so. Definitions, proofs, diagrams and clinical reasoning need different kinds of work; cards will not cover every assessment skill.
What does the evidence establish?
Cepeda and colleagues' 2006 review covered 317 experiments on distributed practice. Their 2008 study varied the gap between learning and review and the later test delay. The useful gap depended on that delay. This supports thinking about the time horizon, but the experiments did not validate the five-review calendar above for every topic.
Kornell's 2009 flashcard experiments found better performance with a large spaced stack than small massed stacks for most participants. The often-quoted 72 percent preference figure applies to participants who judged one condition more effective, not to everyone in the experiments. Your feeling that a card is easy during an immediate repeat is therefore not the only evidence to use.
The 2015 Ebbinghaus replication studied one participant's relearning of nonsense syllables. It measured savings in relearning effort. It is not a timetable showing what fraction of your lecture you will forget by tomorrow, and it does not prove the first review must happen at a fixed hour.
How should the calendar change near an exam?
Start with the exam date and your current gaps. If you have two weeks, you might plan several visits within that period, with room between them to understand and practise the difficult material. If you have two days, an attempt today and another tomorrow can still be useful; short spacing is not automatically pointless.
Do not fill every available session with cards. For an essay exam, include argument and evidence practice. For calculations, solve unfamiliar problems. For an oral exam, practise explaining and responding to a follow-up question. The exam-preparation guide helps combine these tasks.
When you miss a planned day, look at the queue and resume with a feasible selection. Avoid compensating by adding tomorrow's new cards on top of every missed item without checking the workload. Reserve return visits in your study schedule and adjust from the results.
Paper dates or an app?
For paper cards, explicit next-review dates are easy to interpret. A box system can also work, but label whether a number means a gap after the latest successful review or a date counted from initial learning. Those are different systems.
An app can track individual cards and schedule them according to its algorithm and your responses. It still depends on honest grading and accurate questions. Notibo can generate draft cards from permitted recordings and documents; recording-generated cards and spaced repetition are Pro features included in the trial. Its schedule should not be assumed to match the manual example or Anki. Individual generated cards cannot currently be edited, so use a separate question list or an editable-card tool when revision of the prompts is essential.