Your First 30 Days With Spaced Repetition

The short answer

Everything below is elaboration on those three instructions. If you follow only them and read nothing else, you will do better than most beginners, because the standard failure is not doing the wrong thing on day 20 — it is having set the limit to 50 on day 1.

Day 0: four decisions

Make these before you add a single card. Each one takes about a minute and each one is a common way to fail.

1. Pick one subject

Not three. The temptation on day one is to card everything — the language, the certification, the book you're reading. Every subject you add multiplies the daily load and divides the attention available for building a habit. Add the second subject in month three, once the first is running without effort.

2. Set the new-card limit to 10

This will feel absurdly slow in week one. That feeling is the trap. Ten new cards a day becomes roughly 100 daily reviews within a month and 3,650 cards within a year, which is more than enough for almost any goal. The arithmetic is laid out in how many cards to review per day, and it is worth understanding before you override the default rather than after.

3. Attach the session to something you already do

Immediately after morning coffee. On the train. Before you open your laptop at work. A new habit anchored to an existing one survives dramatically better than one scheduled at a time of day. Pick the anchor now and keep it fixed for the month — the specific time matters much less than it being the same time.

4. Decide what a card looks like before you write 200 of them

One question, one answer, answerable in under ten seconds. If you write your first thirty cards badly you will spend the rest of the month reviewing cards you can't answer and concluding that spaced repetition doesn't work for your subject. Twenty minutes with how to make flashcards that actually work is the highest-return preparation available.

Week 1 — build the habit, not the deck

Expected load: 10 new + 5–40 reviews. Three to eight minutes a day.

Week one is easy and slightly boring, and that is the intended experience. Your only real job is to open the app seven times.

• Review at your anchor time, every day. Seven for seven. If you miss a day in week one, the habit hasn't formed yet and there's very little holding it together.
• Only card things you already understand. Cards are for retaining understanding, not producing it. If you can't explain the idea to someone else, it isn't ready to be a card — go back to the source, or work it out in conversation, and then card the result.
• Write cards at a different time from reviewing them. Making cards is a study activity; reviewing them is a memory activity. Blending the two turns a five-minute session into forty minutes and makes the habit fragile.
• Resist adding more. You will hit ten new cards in ninety seconds and want to raise the limit. Don't. The bill for week one's cards arrives in week three.

The week 1 quit-point is boredom: "this is too easy, it can't be doing anything." It is doing something; you just can't see it yet, because nothing has come back at a long enough interval to demonstrate that it stuck.

Week 2 — learn to grade honestly

Expected load: 10 new + 50–90 reviews. Ten to fifteen minutes a day.

Cards from week one now start returning, and you meet the material for a second and third time. This is when the method starts to feel real — you'll recall something you were sure you'd forgotten, and that's the first genuine hit of evidence.

The skill to develop this week is grading. Modern schedulers extract a lot of signal from which button you press, which means sloppy grading degrades everything downstream.

Two specific habits to avoid, both extremely common:

• Pressing Good on everything. This is the most common beginner pattern and it throws away most of the information the scheduler needs. If you genuinely find something instant, press Easy — that card leaves your queue for months and you get the time back.
• Pressing Hard when you actually failed. It feels less like a defeat. It also tells the scheduler you remembered, so the card comes back at a longer interval you're not ready for, fails again, and starts becoming a leech.

The other week-two job: rewrite any card you've now failed twice. Two failures in the first fortnight almost always means the card is broken rather than the fact being hard — usually two facts crammed into one, or a front that doesn't specify what kind of answer it wants. Fixing it now costs a minute. Leaving it costs you a review every few days for a year.

Week 3 — the spike

Expected load: 10 new + 90–140 reviews. Fifteen to twenty-five minutes a day, and it will feel like more.

This is the point at which most people quit, and they quit because it looks like evidence of an unsustainable trajectory. It isn't — the curve flattens within another two or three weeks, because mature cards start leaving the rotation faster than new ones enter it. But from inside week three you can't see that, which is why knowing about it in advance is most of the defence.

What to do:

1. Don't add cards. Not even a few. Keep the limit at 10.
2. Don't skip a day. A skipped day in week three doesn't halve tomorrow's work, it stacks onto it. The one week to be stubborn about is this one.
3. Split the session if the length is the problem. Fifty cards over coffee and the rest on the way home is easier than a hundred and forty in one sitting, and marginally better for retention.
4. If it's genuinely too much, lower the inputs — don't abandon the queue. Drop the new-card limit to 5 and desired retention to 0.85. Both reduce future load; skipping days doesn't, it just defers work and adds lapses.
5. Prune. If you're seeing cards you no longer care about, delete them. A card you don't want to know is pure cost.

What to expect emotionally: some resentment, a strong urge to reorganise your deck instead of reviewing it, and a suspicion that you're doing it wrong. All three are standard. The reorganising urge in particular is worth naming — restructuring decks feels productive and accomplishes nothing.

Week 4 — tune it and commit

Expected load: 10 new + 110–160 reviews. Fifteen to twenty-five minutes, and now stabilising.

The growth slows this week. Cards from week one are going out to three and four week intervals, so they stop contributing daily. This is the first week the deck feels like an asset rather than a debt.

Now — and only now — look at your numbers. Four weeks is the minimum for the data to mean anything.

Then do three things and stop:

• Prune ruthlessly. Delete anything you no longer need, anything duplicated, and anything you've failed five times and haven't fixed. A month-old deck typically has 10–15% dead weight.
• Set the limit you'll live with for the next few months and leave it. Changing it weekly means you never learn what any setting does.
• Write down where you are — card count, retention, minutes per day. In two months this is the only way you'll be able to tell whether things improved.

The seven beginner mistakes

1. Starting with 50 new cards a day. The single biggest cause of failure. Week one feels effortless at any limit, so people calibrate on the wrong week.
2. Downloading a 10,000-card shared deck. You get someone else's questions about material you haven't met, with none of the thinking that writing a card forces. Fine later for standardised subjects; a poor first month.
3. Making cards from material you haven't understood. Repetition cannot install comprehension. It reliably installs a memorised string you can recite and can't use.
4. Reviewing cards early because you feel unsure. Early reviews are easier reviews, which means weaker memories and a scheduler working from bad evidence. Trust the queue.
5. Changing settings weekly. The feedback loop is three weeks long. Change one thing, wait a month.
6. Treating a missed day as failure. A missed day costs almost nothing. Quitting because you missed a day costs everything, and it's the actual mechanism by which most decks die.
7. Reorganising instead of reviewing. Deck structure, tags, colour-coding, sub-decks — all of it feels like studying and none of it is retrieval.

What month two looks like

Briefly, so you know what you're aiming at:

• Daily load flattens. At 10 new/day it settles near 100 reviews and stays there, even as the deck keeps growing. This is the moment the method starts paying back.
• Long intervals appear. Cards start coming back at one, two, three months. Answering something correctly after nine weeks is the first properly convincing evidence, and it lands somewhere in month two or three.
• Leeches become visible. A handful of cards will have failed repeatedly. Now is the time to rewrite or delete them — they're a tiny fraction of the deck and a large fraction of the pain.
• You can add a second subject, if the first is genuinely automatic.

The honest timeline: individual facts stick within a week or two, and you'll notice. The systemic benefit — a large deck that costs twenty minutes a day — takes two to three months, because that's simply how long early cards need to reach long intervals. Anyone promising a transformation in ten days is describing the feeling of week one, which is not the same thing.

Frequently asked questions

Where this comes from