First-attempt accuracy is the number that matters
Every question bank shows you an accuracy percentage, and every candidate watches it climb. The problem is that most of that climb is not learning. It is recognition.
Why repeat accuracy flatters you
The second time you meet a question, you are not answering it the way the exam will ask you to. You remember the stem, or the shape of the options, or simply that the sneaky-looking answer was correct last time. Get it right and your overall accuracy ticks up, but nothing about your ability to answer an unseen question has changed.
Run enough repeats and the effect compounds. A candidate who has been through the bank twice can sit at 85 per cent overall while their performance on fresh material is stuck near 60. The dashboard says ready. The exam, which is all fresh material, will disagree.
What first-attempt accuracy tracks instead
First-attempt accuracy only counts the first time you ever see a question. It cannot be inflated by repetition, because repeats are excluded by definition. That makes it the closest thing a bank can give you to a rolling estimate of how you would perform on questions you have never met, which is the exam’s only condition.
It is a harsher number. It will sit well below your overall accuracy and it will move slowly. That is not a flaw. A number that moves slowly when your knowledge moves slowly is a number you can trust.
How to read it in MedQVault
MedQVault tracks first-attempt accuracy separately, overall and per topic, and the analytics screens lead with it. Two ways to use it:
- Trend, not snapshot. One rough session means nothing. Watch the direction over a few weeks of steady volume. If first-attempt accuracy is climbing while you work through new material, the preparation is working.
- Per-topic, not global. A respectable global number can hide a topic sitting twenty points below the rest. The per-topic view is where you find the categories that need a targeted block rather than more general volume.
Where repeats still earn their keep
None of this means repetition is wasted, it just belongs in a different column. Reworking your incorrect pool is one of the highest-yield things you can do, because it targets demonstrated gaps. MedQVault keeps every question you have got wrong in a pool you can build a session from directly. The point is to learn from repeats while measuring yourself on first attempts, and to never let the second number borrow credibility from the first.
Try a set of unseen questions in the free samples and see where your first attempt actually lands.