Skip to main content

Why Anki Has Become Painful (and How to Fix It): A Diagnostic Triage

When Anki becomes painful, the cause is usually configuration, not proof that spaced repetition is the wrong tool for you. The dread, the swelling backlog, and the cards that keep failing are usually symptoms of a few fixable settings, not a verdict on your discipline.

Overview​

This is a diagnostic triage, not a theory lesson. The goal is to take the specific pain you feel right now, trace it to its likeliest root cause, and give you an ordered list of fixes. They run from gentle and reversible to nuclear.

The new cards entering your deck each day, the maximum reviews Anki shows you, your desired retention target, and how leeches are handled are deck options you control. They are not fixed properties of the algorithm.1 When the load becomes unbearable, the usual culprit is one of those dials set wrong, not a flaw in spaced repetition itself.

Is it Anki, or is it your config?​

The central thesis of this article is that most Anki pain is a settings or deck-quality problem. The manual itself frames overload as a predictable outcome of intake settings: "More than one Anki user has excitedly studied hundreds of new cards over their first few days of using the program, and then has become overwhelmed by the reviews required."2

That language is worth sitting with. The manual describes overload as a configuration outcome, not a willpower failing. This triage takes the same stance.

The dials are yours to turn

New-card intake, maximum reviews per day, desired retention, and leech handling are all user-controlled deck options.1 Before you conclude that spaced repetition is broken for you, check whether one of these is set somewhere you never intended.

A deeper question sits underneath all of this: whether spaced repetition is genuinely the wrong tool for you right now, rather than just badly tuned. That question has its own home in J-Compass's article on when SRS becomes counterproductive, and this triage routes you there rather than answering it inline.

Triage: name your symptom​

Pain tends to show up in a few recognizable forms. Find the one that matches what you feel when you open Anki. Then follow it to its root cause in the next section.

Symptom: the review count keeps climbing (backlog dread)​

A review count that keeps growing, no matter how many cards you clear, is the most common reason people quit. The mechanism is the same forgetting-curve scheduling that makes reviewing on a schedule work. The manual explains: "Studying new cards will temporarily increase the number of reviews you need to do a day, as newly-learned material needs to be repeated a number of times before the delay between repetitions can increase appreciably."2

Anki does not silently drop the cards you miss. "When you fall behind in your reviews, Anki by default prioritizes cards that have been waiting the longest."3

The scheduler is also built to absorb gaps gracefully. "When you answer cards that have been waiting for a while, Anki factors in that delay when determining the next time a card should be shown."3

A climbing count is a load problem, and its root causes are Cause 1 and Cause 2 below.

Symptom: the same cards keep failing (leeches)​

A leech is a defined, measured state, not a vague feeling that a card is annoying. "Each time a review card 'lapses' (is failed while it is in review mode), a counter increases."4

The threshold is fixed by default. "When this counter reaches 8, Anki tags the note as a leech and suspends the card."4 After that, the alerts continue at half the threshold interval. With the default of 8, you are warned again at 12, 16, and every 4 lapses thereafter.4

A card that crosses 8 lapses is signalling that it is malformed or overloaded, not that you lack discipline. Its root cause is Cause 3, deck quality.

Symptom: reviews feel joyless even when the count is fine​

Sometimes the count is perfectly manageable and the sessions still feel like a chore. That points at how the cards are built rather than how many there are.

The Minimum Information Principle says material "must be formulated in as simple way as it is" possible. In practice, complex cards leave inconsistent memory traces and are harder to refresh uniformly at review.5 Cards that are too easy, too hard, or stripped of context produce disengagement even at a healthy daily count, and that construction problem is addressed under Cause 3.

Diagnostics: find the root cause​

Three fixable causes sit underneath nearly all Anki pain. Each symptom above maps to one or more of them. The diagram below shows the routing at a glance.

Cause 1: desired retention set too high (FSRS)​

Desired retention is the FSRS (Free Spaced Repetition Scheduler) dial that tells Anki how often you want to remember a card at review time. "The default is 90%, which offers a good balance of retention and workload."1

Push it higher, and the cost climbs out of proportion to the benefit. "Above 90% the workload increases very quickly, and above 97% the workload can be overwhelming."1 The manual adds its own guardrail: "we suggest you be conservative when adjusting this number, and recommend you keep it lower than 97%."1

If your retention target sits well above 0.90, you are multiplying review volume for shrinking real-world gain. The practical fix is to bring it back toward the default. For the mechanics of how that target maps to interval length, see J-Compass's article on how Anki's scheduling algorithms work.

Cause 2: the new-card ceiling outran your sustainable load​

New cards temporarily inflate daily reviews until their intervals lengthen.2 If new cards keep entering faster than reviews stabilize, the backlog never drains. The count climbs no matter how diligently you study.

The manual gives a yardstick for judging whether your intake is sustainable: "If you are consistently learning 20 new cards a day, you can expect your daily reviews to be roughly about 200 cards/day."2 Weigh that steady-state ratio, near 10:1, against the time you actually have.

The dial to turn is New cards/day in deck options.1 For how to compute a personal ceiling from your own available minutes, see J-Compass's article on finding your sustainable new-card load. Here, the diagnosis is simply that intake outran absorption.

Cause 3: the deck itself is low quality​

Overloaded, complex cards violate the Minimum Information Principle, which says material "must be formulated in as simple way as it is" possible. Simple items refresh uniformly at each repetition, while complex ones do not.5

Leeches are the measurable downstream symptom of that problem. A card that lapses to the threshold of 8 and gets auto-suspended is, in the common case, carrying too much or too little to recall reliably.4

For Japanese learners this failure class is especially common. Kanji homographs, polysemy (one word having multiple meanings), and isolated context-free word cards are all high-risk. The remedy follows from the same principle: atomize the card or add a sentence of context.5

Isolated single-word cards are leech factories

A card that asks you to recall one Japanese word with no sentence around it gives your memory nothing to grab onto, which is why these cards lapse repeatedly.5 Adding a single line of context is often the difference between a card that sticks and one that fights you to the leech threshold.

The fixes, from gentle to nuclear​

Work this list top to bottom. Try the reversible fixes before the destructive ones, and stop as soon as the pain eases.

Gentle: pause new cards and cap review time​

New cards/day and Maximum reviews/day are user-set deck options. Setting New cards/day to 0 stops the inflow without touching any existing scheduling.1 This is fully reversible; you are pausing intake, not deleting anything.

Capping the daily review count then lets you clear a small, predictable batch instead of facing the whole backlog at once. The scheduler keeps prioritizing the oldest-waiting cards as you chip away, so nothing you miss is lost.3

Moderate: lower desired retention and rebuild leech cards​

Lowering desired retention from a too-high target toward the 0.90 default directly reduces review volume. The manual ties the workload curve to exactly this knob: "Above 90% the workload increases very quickly."1

Rebuilding leech cards means reformulating them under the Minimum Information Principle. You can atomize them into simpler cards or add context so they stop lapsing.5 Both of these fixes are reversible.

Nuclear option 1: delete the leeches​

Anki's default leech action already suspends the card at 8 lapses, pulling a chronically failing card from review automatically.4 Deleting a true leech is simply the permanent, user-chosen extension of what the program already does on your behalf.

Deletion here is triage, not failure. The system itself treats a leech as something to remove from circulation, so you are agreeing with its design rather than giving up.

Nuclear option 2: suspend everything old and start fresh​

Suspending means Anki "Hides a card or all of the note's cards from review until they are manually unsuspended (by clicking the suspend button in the browser)."3 Mass-suspending your mature cards lets you escape the backlog without deleting the deck. You can reintroduce them gradually by unsuspending in batches.

Reintroduction is safe because the scheduler "factors in that delay when determining the next time a card should be shown." The cards you bring back are not all dumped on you at once at minimum interval.3

Nuclear option 3: restart the deck​

A genuine restart is warranted when the deck's construction is the root cause, not its scheduling. Before deleting everything, consider the reversible alternative: a filtered deck. It lets you step outside the normal daily limits by temporarily moving matching cards out of their existing decks for study. This is the manual's mechanism for working a backlog down at a chosen pace without growing it.6

Prefer the filtered-deck rescue or a fresh, better-built deck over salvaging a structurally broken one. If the cards themselves are the problem (Cause 3), starting over with well-formed cards beats nursing a deck that will keep generating leeches.

When the problem is not Anki at all​

Sometimes no dial is set wrong, the cards are fine, and the friction is coming from somewhere else entirely. This is the honest off-ramp. It routes back into J-Compass's broader diagnostic guide to stalled progress, where Anki overload is just one of several plateau causes.

When SRS itself is the wrong tool right now​

The deeper decision of whether to stop using spaced repetition for a while belongs in J-Compass's article on when SRS becomes counterproductive. The tool-by-tool survey of alternatives lives in its comparison of SRS tools beyond Anki. This subsection points you to both rather than re-deriving them.

There is one case where an alternative tool is the genuine fix. When the pain is specifically the overhead of tuning retention, intake, and leech thresholds, J-Compass recommends Amenokori. It is a mobile FSRS app with a fixed daily cap that removes the knobs the rest of this article spends its length tuning.

Amenokori's level counts are vendor leveling, not official JLPT

The per-level word and card counts Amenokori shows are the vendor's own leveling, not official JLPT figures. The JLPT has published no official vocabulary list since 2010. Treat the broader landscape of alternatives through J-Compass's dedicated comparison rather than any single product's numbers.

Good to know​

The "every card is sacred" sunk-cost trap​

Anki's default leech behavior is to suspend the card at 8 lapses. In other words, it removes a chronically failing card from circulation automatically.4 Treating every card as undeletable fights the program's own design.

Deleting a true leech is the sanctioned move, not a failure.4

Chasing a 100% retention target​

A near-perfect retention goal creates a self-inflicted workload spike. The manual is explicit: "Above 90% the workload increases very quickly, and above 97% the workload can be overwhelming." It recommends keeping the target below 97%.1

Aiming for total recall multiplies your reviews for negligible real gain.

Reading a post-vacation backlog as moral failure​

The scheduler is built to handle delay. "When you answer cards that have been waiting for a while, Anki factors in that delay when determining the next time a card should be shown." Falling behind simply makes Anki "prioritize cards that have been waiting the longest."3

A backlog after a break is expected behavior, not a verdict on you as a learner.

Confusing a leech with a genuinely hard card​

A leech is a defined state: the lapse counter reaches the threshold, default 8, and the card is auto-tagged and suspended.4 A card that is merely hard but still being learned has not crossed that counter.

The threshold draws the objective line between "hard, keep going" and "leech, intervene."4

Reviews are about ten times the new count​

The manual's own figure, 20 new cards to roughly 200 reviews per day, gives a memorable steady-state ratio.2 Use it to sanity-check whether a configured intake is survivable before you commit to it: multiply your daily new-card number by about ten and ask whether you have time for that many reviews.

See also​

References​

Footnotes​

  1. Anki Manual. "Deck Options." Anki Manual (docs.ankiweb.net). https://docs.ankiweb.net/deck-options.html ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7 ↩8 ↩9

  2. Anki Manual. "Deck Options" (Daily Limits, New Cards/Day). Anki Manual (docs.ankiweb.net). https://docs.ankiweb.net/deck-options.html ↩ ↩2 ↩3 ↩4 ↩5

  3. Anki Manual. "Studying." Anki Manual (docs.ankiweb.net). https://docs.ankiweb.net/studying.html ↩ ↩2 ↩3 ↩4 ↩5 ↩6

  4. Anki Manual. "Leeches." Anki Manual (docs.ankiweb.net). https://docs.ankiweb.net/leeches.html ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7 ↩8 ↩9

  5. Wozniak, Piotr. "Effective learning: Twenty rules of formulating knowledge." SuperMemo (supermemo.com), 1999. Rule 4, "Stick to the minimum information principle." https://www.supermemo.com/en/blog/twenty-rules-of-formulating-knowledge ↩ ↩2 ↩3 ↩4 ↩5

  6. Anki Manual. "Filtered Decks." Anki Manual (docs.ankiweb.net). https://docs.ankiweb.net/filtered-decks.html ↩