SRS burnout rarely begins with reviews. It begins with lessons.
New cards feel like progress now and become scheduled work later. Because that future work is distributed across days, the queue can grow for a week before the decision that caused it becomes obvious.
The fix is not a heroic catch-up session. It is a review budget.
Use an ordinary-day budget
Choose the amount of review time you can protect on a normal workday—perhaps 10, 15, or 20 minutes. Do not use your most motivated Sunday as the forecast.
That budget has three parts, in this order:
- Due reviews: the material your system has already scheduled.
- Trouble items: a small amount of repair for confusing or repeatedly missed cards.
- New material: only what fits after the first two.
When the first two consume the budget, the correct number of new cards is zero.
Start below your apparent capacity
For a new deck or course, begin with roughly five new items per day. The exact number matters less than holding it steady long enough to see its delayed effect.
Wait ten to fourteen days before increasing. By then you have seen early failures, weekend disruption, and the collision between young cards and new lessons.
Increase by a small step. If your platform offers lessons in larger batches, you do not have to clear every available lesson just because the button is full.
Reviews first, but not reviews only
Clearing the queue can become a closed game. It feels measurable, so reading and listening get postponed.
Protect a second block—even ten minutes—for Japanese that is not shaped like a card. A short graded passage, a familiar podcast segment, or one scene with Japanese subtitles gives the remembered material somewhere to attach.
If review takes the entire study window for several days, reduce new material. The system is consuming the activity it was meant to support.
When you already have a backlog
Day 1: stop adding
Pause new lessons everywhere. If you use multiple SRS tools, choose one primary queue for the recovery week. A backlog spread across three apps feels worse because each app presents itself as the priority.
Days 2–4: cap the session
Review for your normal time budget and stop. Do not convert one missed week into a two-hour punishment. If the system lets you prioritize newer or lower-stage items, use that to prevent fragile material from decaying further.
Day 5: remove bad cards
Suspend or rewrite cards that are ambiguous, overloaded, or repeatedly wrong for reasons unrelated to the target memory. A card with three possible answers is not “hard”; it may simply be a poor test.
Days 6–7: accept forgetting
Some mature items will return as failures. That is not a broken streak. It is information. Grade honestly and let the scheduler rebuild the interval.
Resume new material only when the queue fits your ordinary-day budget again.
Do not press “Hard” when you forgot
In systems using FSRS, including current Anki, the answer buttons carry specific information. “Hard” means you successfully recalled the answer with effort. If you did not recall it, use the failure grade—usually “Again.”
Using “Hard” as a gentler failure can teach the scheduler that forgotten cards were remembered, producing intervals that do not match your history.
Fix leeches at the source
A card that fails repeatedly deserves diagnosis, not infinite repetition.
Ask:
- Is the prompt asking for one clear thing?
- Is the answer short enough to judge?
- Am I trying to learn a word with no sentence or situation?
- Are two similar items interfering with each other?
- Is this item useful enough to earn repeated attention?
Possible repairs include adding a contrast sentence, splitting one card into two, adding audio, changing recall direction, or suspending the item until you meet it in real Japanese.
Run one main queue
WaniKani for kanji plus Bunpro for grammar plus Anki for mined sentences can be a sensible long-term system. Starting all three aggressively in the same week is not.
Give one queue permission to create new work. Keep the others at maintenance load or paused. When the primary queue stabilizes, add the next layer slowly.
This rule is especially useful when trying a broad platform like MaruMori alongside a flexible system like Anki: decide which one owns new material for the month.
A weekly load check
Once a week, write down four numbers:
| Check | Useful signal |
|---|---|
| Average review minutes | Does the queue fit the time budget? |
| New items added | What decision created next week’s work? |
| Repeated failures | Which cards need repair or removal? |
| Non-SRS Japanese sessions | Is review supporting real input or replacing it? |
If review time rises for two weeks, cut new material before changing scheduler settings. The simplest lever is usually the correct one.
Sustainable SRS is deliberately a little boring: a small queue, honest grades, useful cards, and enough energy left to read or listen afterward.
The minimum viable review day
For travel, illness, or deadlines, define a version of the habit that takes five minutes. Review the most fragile items, add nothing new, and stop without guilt.
The objective is not to protect an app streak. It is to preserve the path back to ordinary study.