HomeBlogStudy methods
Study methods

Spaced repetition: the review schedule that actually works

Reviewing at the right moment beats reviewing a lot: research shows the optimal gap depends on your exam date. In this article I walk through what the science of spaced repetition actually says — and how I apply it in SimulAI by generating a fresh practice test at every review.

By Douglas Oliveira8 min read

When I started building SimulAI Academy, I assumed the biggest problem students faced was a lack of good material. After watching closely how students and teachers actually use the platform, I changed my mind: the most common problem is the calendar. I see dedicated people with immaculate notes who pour all their effort into the three days before an exam — and who, a month later, have to relearn almost everything from scratch for the next one. Spaced repetition exists precisely to fix this, and it is one of the few study techniques backed by more than a century of accumulated experimental evidence. In this article I want to go beyond the “review at growing intervals” line that every post on the topic repeats: I will walk through what the research actually says about when to review, including a detail almost no generic article mentions — the optimal gap between reviews depends on the date of your exam.

Each practice test keeps attempts, scores and a redo button — the backbone of a spaced review cycle.
Each practice test keeps attempts, scores and a redo button — the backbone of a spaced review cycle.

What the forgetting curve really shows

Everything rests on an experiment from 1885. Hermann Ebbinghaus memorized lists of nonsense syllables and measured how much he could recall after delays ranging from twenty minutes to a month. The result became known as the forgetting curve: loss is steepest in the first hours and slows down over time. For decades that finding rested on a single old study, run by the author on himself — an easy target for skepticism. That is why the replication by Murre and Dros (2015), published in PLOS ONE, matters so much: redoing the original experiment with modern rigor, they found essentially the same curve. Without review, a large share of what we learn slips away within the first day; what survives the first week tends to last far longer.

The practical consequence is twofold. First: reviewing is not a luxury, it is a structural part of learning — studying something once and trusting your memory is a bet against biology. Second: every well-placed review “flattens” the curve. After you revisit a topic, the next round of forgetting is slower. That is why the timing of a review matters as much as the review itself.

The evidence that spacing beats cramming

This is neither a recent nor a fragile claim. Cepeda and colleagues (2006), in a meta-analysis published in Psychological Bulletin, reviewed hundreds of experimental comparisons between distributed practice (sessions spread over time) and massed practice (everything at once) in verbal memory tasks. The pattern is one of the most consistent in all of cognitive psychology: with the same total study time, people who spread their sessions out remember more, and for longer, than people who concentrate them. The “same total time” part deserves emphasis: spaced repetition does not ask for more hours of study, it asks for the same hours on different days.

The night-before marathon is not useless — it works for the next morning, which is exactly why the habit persists. What it does not do is build durable memory. If your goal ends on exam day, cramming gets you through; if you need the content for cumulative exams, board certifications, or your professional life, spacing is the only one of the two that delivers.

The optimal gap depends on your exam date

Here is the nuance generic articles almost never include: there is no universally ideal interval. It depends on how far away the day is when you need to remember. Cepeda and colleagues (2008) ran a large-scale study, published in Psychological Science, crossing dozens of combinations of gap between sessions (from hours to months) with retention deadlines (from one week to nearly a year). The results trace a kind of “ridgeline” of optimal retention that translates into practical rules:

  • Exam in about a week: do your main review roughly one day after first studying.
  • Exam in about a month: the optimal gap sits around one week.
  • Exam several months to a year away: gaps of roughly three weeks performed best.

A rule of thumb that falls out of the study: the ideal gap is roughly 10% to 20% of the time remaining until the exam. And there is a useful asymmetry the paper itself highlights: spacing too much costs little, while spacing too little wastes a lot. When in doubt between reviewing tomorrow or in four days for a distant exam, pick the longer gap.

Expanding intervals are not magic

Nearly every spaced repetition app uses expanding schedules: 1, 3, 7, 16, 30 days. I recommend that format too — but it is honest to say it works for a different reason than most people assume. Karpicke and Bauernschmidt (2011) directly compared expanding schedules with equal-interval schedules and found little difference between the patterns: what determined long-term retention was the absolute spacing — having gaps long enough between retrievals — not the shape of the growth. In practice I find this liberating: it is not worth stressing over the exact formula. Missing your “day 3” review and doing it on day 5 does not break the method. What breaks the method is piling every review into the same evening.

Where spaced repetition fails

No technique is proof against misuse, and I would rather be transparent about the failure modes I see most often:

  • Spacing out passive rereading. Spacing amplifies active retrieval; if each “review” is just skimming your summary, you keep the calendar and lose the engine. Every review should begin with an attempt to recall before checking.
  • Memorizing before understanding. Spacing consolidates what has already been understood. Scheduling reviews of content you have not yet grasped only spaces out the confusion.
  • Procrastination in disguise. “I will leave it for the day-7 review” without a real calendar is postponement with technical vocabulary. Spacing means setting dates, not pushing things back.
  • Trivial volume. Turning every sentence of the PDF into a review item drowns the important concepts in disposable detail. Fewer, better items.

A reference schedule to start with

For new content, with an exam about a month away, this sequence is a solid starting point:

  • Day 0: first contact with the material and an initial diagnostic test.
  • Day 1: first review, always trying to retrieve from memory before checking.
  • Day 3: second review, focused on what failed.
  • Day 7: third review.
  • Day 16 and Day 30: maintenance reviews, each one shorter.

If your exam is further away, stretch everything proportionally, following the 10–20% rule from Cepeda (2008). If you miss a lot on a return, shorten the next gap; if you sail through, stretch it. Performance is the best calibrator there is.

How I apply this in SimulAI

Much of SimulAI Academy’s design grew out of this research. While building the test generator, I realized the biggest practical obstacle to spaced repetition is not understanding the concept — it is having fresh material for each return. Reviewing by rereading the same PDF is too easy; redoing the same questions turns into memorizing the answer key. The flow I use myself: at each point in the schedule, I generate a new practice test on the same material, with different questions. That way every review is also an honest test.

  • Use performance as a thermometer: topics with many mistakes come back sooner on the calendar; solid topics can wait longer.
  • Mix old and new topics in the same session — interleaving subjects sharpens the ability to discriminate between similar concepts.
  • Split by tool: flashcards for facts and definitions that need frequent repetition; practice tests for applied reasoning; mind maps for rebuilding the structure of a subject from memory before checking.
  • Keep sessions short: 15 to 25 focused minutes per return is enough — the method saves time precisely because each minute yields more.

And the mistake I see most often among people using the platform: generating one giant practice test the night before. Technically that is active recall, but with no spacing — half the benefit stays on the table. Three short tests across two weeks are worth more than one long test the evening before.

The mistakes I see most

  • Cramming it all the night before: good for passing, not for learning.
  • Always reviewing too early: without the effort of nearly having forgotten, the gain is small.
  • Rereading instead of testing yourself: spacing without active retrieval loses half its power.
  • Weighting everything equally: let your mistakes decide what comes back first.

Start small: pick one topic from this week and lay out five spaced reviews across a month, each one opening with a test. The effort per session is low, and the accumulated difference is what separates remembering for one night from remembering for a semester. That difference is what convinced me to build an entire platform around the idea.

References

Tags:Spaced repetitionScheduleMemory

Keep reading