Two weeks before the exam, most students do the same thing: they grind harder on the deck they have. More reviews, longer sessions, earlier mornings. It feels productive. But if the deck is missing 15 percent of the tested material, no amount of grinding covers it. You cannot mature a card that does not exist.
The highest-leverage study activity in the final weeks is not more reviews. It is a coverage audit: checking what your deck actually covers against what the exam will actually test, and closing the gaps while there is still time.
Start from the exam blueprint, not the deck
Every exam has a blueprint, explicit or implicit. For in-house exams it is the lecture schedule: the topics, the slide counts, the professor's emphasized points. For boards it is the published content outline. Write down the topic list first, then check the deck against it — not the other way around. Starting from the deck biases you toward what you already have.
The slide-by-slide audit
Take one lecture at a time. Open the slides and ask, for each major point: is there a card that would let me answer an exam question on this? Be strict. A card that mentions the topic in passing does not count. A card that tests the exact detail the professor bolded, underlined, or repeated — that counts.
Score each lecture roughly: what percentage of testable points have real cards? Anything under 85 percent is a gap worth fixing. Most students who do this honestly find at least two or three lectures per block sitting at 60 to 70 percent, usually the ones taught by guest lecturers or covering the professor's niche interests.
Prioritize by yield, not by anxiety
You will find more gaps than you have time to fix. Prioritize: high slide count plus professor emphasis plus historically tested equals fix first. A single slide mentioned once in passing can wait. The goal is not a perfect deck. It is eliminating the gaps most likely to cost you points.
Fix the gaps the right way
For each gap, make targeted cards — one fact per card, tested the way the exam tests it. Tag them with the exam name so you can cram them as a filtered deck in the final days. And check for partial coverage before making anything: if a card exists but misses the emphasized detail, edit it in place rather than duplicating it. Your scheduling history on that card is valuable; do not reset it a week before the test.
The fast way: audit the whole deck at once
The manual audit works, but it is slow enough that most students audit one lecture, get depressed, and stop. The bottleneck is the comparison itself: holding the slides in your head while searching the deck, fact by fact.
SlideToAnki's deck improvement automates the audit. Upload your .apkg and the lecture slides, and it compares them directly: material in the slides with no covering card gets flagged as a gap, with new cards proposed for each one. Existing cards that the slides contradict get flagged too. You work through the proposals in a review queue — accept, reject, or edit — and every card links to its source slide so you can verify coverage yourself.
Run it once per block, a few weeks before the exam, and the "cards that do not exist" stop being a surprise you discover on test day.
Make it a habit
Coverage audits feel unnecessary when you are keeping up with reviews, which is exactly when they matter most. A deck at 100 percent review completion and 70 percent topic coverage is a trap: the green checkmarks tell you that you are done while the exam tests the part you never carded. Audit early, fix the high-yield gaps, and walk in knowing your deck matches the test.