The 20-minute rule: why cramming before a lesson backfires

Here's the ritual. Your lesson's at 6. At 5, you sit down with the deck and go through all 90 cards, front to back, twice. By 5:55 you feel ready.

You walk in, the tutor asks how your week was, and three of the words you drilled forty minutes ago have already gone soft.

The pre-lesson hour wasn't wasted, exactly. But you paid full price for maybe a quarter of the memory it should have bought. The reason is one of the most reliable findings in a century of memory research, and once you see it you can't unsee it.

01 / 05

Cramming buys performance that's mostly gone by Thursday

Psychologists have a dry name for cramming: massed practice. You pack all your repetitions of a thing into one tight block. The opposite is distributed practice, where the same repetitions are spread across days.

Massed or distributed, it's the same total effort. Wildly different results. When you mass your practice, each repetition lands on a memory that's still fresh from the last one, so the brain shrugs: already know this, no need to work. Retrieval feels easy, and easy retrieval barely strengthens anything.

Spread the same reps out, and each time you come back the memory has faded a little. Pulling it up takes effort. That effort is the workout. The cognitive-science term for this is "desirable difficulty," which is a very academic way of saying: if it feels too easy, it isn't working. We wrote a whole post on it.

There's a useful distinction hiding under all this, and Soderstrom and Bjork's 2015 review spells it out. Performance is how well you do during practice. Learning is what's still there next week.

The two come apart constantly. Cramming buys performance. It looks great at 5:55 and it's mostly gone by Thursday.

Soderstrom and Bjork's review walks through decades of experiments where the group that looked worse during practice kept more afterward. Language vocab, motor skills, math procedures. The pattern holds almost everywhere anyone has checked.

And the reason spaced reps punch harder is that each one is a genuine retrieval event. Pulling a word back across a 3-day gap is a small act of reconstruction (the same mechanism behind the testing effect). The 5th repetition inside a single hour is closer to reading.

Spacing upgrades the reps you already had, without adding any.

02 / 05

Split the same 30 minutes across three days, and French learners kept about 35% more

The spacing numbers aren't subtle. A 2006 review by Cepeda and colleagues pulled together 317 experiments on distributed practice and found the same thing almost every time: spacing wins, and it wins by a lot. The gap between crammed and spaced retention is the whole reason a scheduler exists.

One study applies almost rudely well to language learners. Bloom and Shuell (1981) had high-school students learn French vocabulary for a total of 30 minutes. One group studied in a single half-hour block, the other in three 10-minute sessions on consecutive days.

On the immediate test, the groups tied. Four days later, the spaced group remembered about 35% more words.

Spread the reviews out and far more of them survive the week Same cards, same six reps, same total minutes. The only thing that changes is when they happen. The spaced schedule costs no extra time.

35%

recalled a week later after cramming all six reviews in one night

80%

recalled after spacing the same six reviews across two weeks

0 min

of extra study time the spaced version required

Illustrative, not measured. The figures sit in the range the spacing literature reports (Cepeda et al. 2006, 317 experiments). Your real numbers live on your home screen.

The catch that makes spacing hard to do by hand: the best gap between reviews grows as a memory gets stronger. A word you learned yesterday wants to be seen tomorrow; a word you've known for a month can wait three weeks. Track that across 600 cards and you've got a spreadsheet nobody wants to keep.

The right gap gets longer every time a word holds A word you learned yesterday wants to be seen tomorrow; one you have known a month can wait three weeks. That is why no fixed interval works, and why doing this by hand across 600 cards is a spreadsheet nobody keeps.

day 0 +1d +3d +1wk +3wk +2mo

Schematic, not measured. It is one card's schedule, drawn to show the shape. Expanding gaps follow Cepeda et al. 2008 (1,354 participants). Verbamor computes the real ones per card.

There's even a paper mapping how long the gaps should be for a given goal. Cepeda and colleagues (2008) ran 1,354 people through gaps ranging from minutes to months and found the best gap scales with how far away the test is. Want to remember something for a year? The reviews need to sit weeks apart, which no amount of pre-lesson drilling can fake.

03 / 05

Why does cramming feel like it works?

Cramming feels like it works because during the cram, recall is fast and smooth. And your brain reads smooth as learned. Psychologists call this the fluency illusion: we judge our own learning by how easy something feels right now. That's roughly the worst available signal. Koriat & Bjork (2005) put it memorably as "illusions of competence".

Nate Kornell tested the fluency illusion directly with flashcards in 2009. Students learned GRE-type vocabulary (the kind on a US graduate-school entrance exam) either massed or spaced, total study time held equal. Spacing won for about 90% of them.

Most predicted cramming would win before they started, and a good chunk still believed it afterward, with their own results sitting in front of them.

So the feeling of a productive pre-lesson hour is real. It's just measuring the wrong thing.

04 / 05

Twenty minutes a day beats two hours the night before

So here's the rule I give every learner who asks why their pre-lesson cram isn't sticking:

Twenty minutes a day, most days, beats two hours the night before. Every time.

Twenty minutes is enough to clear the cards that are actually due without burning out. Most days means the spacing does its job in the background. And because you're reviewing when things are on the edge of forgotten, every rep counts double.

The arithmetic works out kindly, too. A deck growing by 10 or 15 words per lesson settles into roughly 60 to 90 due cards a day once the scheduler spreads them out. That's 15 to 20 minutes at a relaxed pace. Miss a day and the pile bends; miss a week and it stacks. The habit is the whole product.

I'd also say "most days" on purpose. A missed Sunday costs you almost nothing; the scheduler absorbs it. Chasing a perfect streak is how people burn out in month 2. And a learner who quits in month 2 loses to a learner who reviews 5 days a week for a year. It's not close.

If you can pick your 20 minutes, evening is a nice slot. Words reviewed close to bedtime go straight into the night's consolidation window, which quietly adds a second effect on top of the spacing. Morning works too. The time of day matters far less than the most-days part.

The fill-in-the-blank format matters here too. Producing the word yourself, rather than flipping a card and nodding, is a second free boost to memory. Here's the same card, drilled two ways:

Flipping a card and nodding is not the same rep as producing the word The same card, drilled two ways. Only one of them makes you retrieve anything.

Flip & nod el bosque → the forest. "Yeah, I knew that."

Fill the blank el      "the forest" → you have to pull bosque out yourself.

Not data. These are two card formats, shown as the app renders them. The effect behind it is the testing effect; more in our post on it.

05 / 05

Keep the hour, and spend it on the lesson instead of the deck

Keep the hour. Just spend it differently:

• Do your normal 20-minute review, the same one you'd do any day. Let the scheduler pick the cards.
• Skim your notes from last week's lesson once, to warm up the context rather than memorize it.
• Write down two or three things you actually want to say this session. Real sentences, not vocab.

Then close the laptop and let the words stay a little uncertain. That uncertainty is exactly the state your tutor's hour is best at fixing. Walk in planning to use the deck, out loud, on a real person.

And after the lesson, resist the urge to re-drill whatever went badly. Flag it, let it come back tomorrow morning when the retrieval will actually cost you something, and go eat dinner. Tomorrow's rep is worth more than tonight's three.

---

Sources