Why Flashcard Accuracy and Retention Statistics Disagree
Your flashcard accuracy and retention statistics can count different attempts. Successful retries can raise overall accuracy without changing your first-attempt result. Check what each percentage includes before changing your study routine.
This article is published by Miji’s maker.
What each percentage counts
Sources checked on October 17, 2026.
Anki’s daily review count includes each answer, including repeated answers to the same flashcard. Its True Retention table counts only each flashcard’s first review that day. Again means failure; Hard, Good, and Easy mean success in that table. Anki statistics manual
The denominator is the total used underneath a fraction. Change that total, and you change the question being answered.
Accuracy across all attempts asks: how often did your answers succeed? First-attempt retention asks: how often did you succeed before any retries today?
A fictional ten-flashcard review log
Imagine you review ten previously studied vocabulary flashcards today. Each asks you to recall one meaning. You retry both missed flashcards later that same day.
Here is the complete fictional log, grouped to keep it readable:
- Flashcards A–D: four initial successes; no further attempts.
- Flashcards E–H: another four correct answers on first presentation.
- Flashcard I: an initial failure, followed by a successful retry.
- Flashcard J: missed at first, then answered correctly on retry.
Count only the first presentation of each flashcard. You have eight successes from ten attempts: 8 ÷ 10 × 100 = 80% first-attempt retention.
Now count every answer, including the retries. You have ten successes from twelve attempts: 10 ÷ 12 × 100 = 83.3% overall accuracy, rounded.
Both calculations describe the same session. The higher figure includes your successful retries after seeing the answers. Those retries do not erase the earlier misses.
Neither percentage tells you how many words you can use in conversation. This log measures answers to specific prompts under specific conditions.
Try the counting yourself
Keep the same first attempts, but change the ending. Suppose flashcard J fails its retry and receives no further attempt today.
- What happens to first-attempt retention?
- What happens to accuracy across all attempts?
Answer key: First-attempt retention stays at 80%. Overall accuracy becomes 9 ÷ 12 × 100 = 75%. The extra failure changes only the measure that includes retries.
This also shows why overall accuracy need not be higher. Its direction depends on how those extra attempts go.
Check these details before changing your practice
Write down the labels beside the percentages you are comparing. Then check these details:
- Counting rule: Does the figure include retries or only first attempts?
- Date range: Are both figures covering the same period?
- Flashcard group: Are you comparing the same deck and material?
- Sample size: How many answers sit behind each percentage?
Anki supports statistics for a selected deck or the whole collection. Match that scope before comparing results. Its manual also warns that daily retention is noisy and recommends examining monthly data. Anki statistics manual
In our fictional log, one changed first answer moves retention by ten percentage points. That large swing comes from a small sample.
If the counting rules differ, explain the gap before treating it as a problem. If a longer decline remains, inspect the missed prompts. Look for unclear questions, changed material, or a break in practice.
A useful next step is to keep a short note of these changes. Review that note alongside your next monthly check. This is a suggested monitoring routine, not a prediction of future recall.
For broader practice guidance, see spaced repetition for vocabulary. For the underlying method, read what spaced repetition means.
Miji
As an optional vocabulary practice example, Miji supplies a Knew/Forgot workflow. It supports English learning on iPhone with iOS 26 or later.
Miji's free version keeps dictionary search and practice, with room for 10 words.
Comparable retention analytics are unverified here. Do not assume that similarly named percentages use identical counting rules. If detailed statistical inspection is your priority, consider Anki’s documented statistics tools. Anki statistics manual
Questions
Which percentage should I use?
Use first-attempt results to examine recall before same-day retries. Use all-attempt accuracy to describe the full session. Keep the definition consistent when comparing periods.
Does a successful retry cancel a forgotten answer?
No. In the fictional log, it adds another successful attempt. The original failure remains part of what happened.
Should one bad day make me switch apps?
One day gives you too little evidence for that decision. First check the counting rules and whether the difficulty persists across a longer period.
Does high retention prove I can use the vocabulary?
No. Recalling a flashcard answer and using a word appropriately are different tasks. Check your vocabulary in reading, writing, and conversation too. No percentage guarantees lasting retention.