Your Practice Scores Stopped Improving. Here's How to Review Wrong Answers So They Move Again
· 4 min read
You've done hundreds of practice questions. A month ago your scores were climbing. Now they've been stuck in the same range for weeks, and some days they go down. The obvious response is to do more questions, so you do, and nothing changes.
This is one of the most common complaints on bar, CPA, CFA and medical-board forums, and the answer from people who broke through is remarkably consistent: the questions weren't the problem. The review was.
What usually happens after a wrong answer
You answer, you see the red X, you read the explanation, you think "oh, right," and you click next.
That last step feels like learning, but it's mostly the same illusion as rereading: the explanation makes sense while you're reading it. Nothing makes you come back. A week later, the same idea comes up in a different question and you miss it again — often for the same reason.
So the misses keep recurring, the score stays flat, and doing more questions only produces more misses that get the same quick read.
Step 1: Log every miss, with a reason
Keep an error log. A spreadsheet is enough. For every question you got wrong — and every one you got right by guessing — write one row:
| Date | Topic | What I chose | Why I missed it | One-line rule |
|---|
The last two columns are the point. "Why I missed it" forces you to diagnose. "One-line rule" forces you to state, in your own words, what you'd need to know or do to get it right next time.
Step 2: Sort the misses into four kinds
When you fill in "why", almost every miss falls into one of these:
1. Didn't know it. The fact, rule or concept wasn't in your head. Fix: learn it, then test it again later — not just reread it.
2. Knew it, mixed it up. You confused two similar things: two standards, two drugs, two legal tests. Fix: practise telling them apart directly, with questions that differ by one detail.
3. Misread the question. You missed "NOT", "except", "first", a date, a condition in the stem. Fix: this is a process problem, not a knowledge one. Slow down on the stem; underline the qualifiers; check what is actually being asked before looking at options.
4. Ran out of time or rushed. Fix: timed sets, and a rule for when to flag and move on.
After a couple of weeks, count. Most people find one category dominates. If 40% of your misses are misreads, more content review won't help at all. If most are mix-ups, you need comparison drills, not more reading.
Step 3: Don't just read the explanation — answer again later
A miss is only fixed when you get that idea right again, on a later day, ideally on a different question.
So each miss goes into a review queue:
- Next day: try the same idea again with your notes closed.
- About a week later: again.
- A couple of weeks after that: again, ideally on a different question about the same point.
- Miss it at any stage → back to the next day.
This is spaced repetition applied to your mistakes instead of to flashcards. It's slower per question than clicking "next", and much faster overall, because each idea stops costing you points.
Step 4: Review the questions you got right, too
Read the explanation for every question you got right but weren't sure about. A lucky guess counts as correct in your score and is still a gap. Log those as misses.
A useful test for any right answer: could you say why each wrong option is wrong? If not, you were partly guessing.
Step 5: Do fewer new questions until the queue is under control
If your review queue keeps growing, that's a sign you're taking on new questions faster than you're fixing old ones. For a week or two, cut back on new sets and work the queue down. Scores often jump after this, because you're finally collecting the points you kept losing.
Why scores can drop for a while
Two common, harmless reasons:
- You moved to harder questions or a new bank. A dip is expected.
- You started reviewing properly, and harder topics are now coming up more often instead of being skipped.
What matters is the trend over a few weeks on fresh questions, not any single day.
Doing this with less admin
The log and the review schedule are where most people give up — it's bookkeeping, and it competes with actual studying. You can make it lighter: an AI chat can write a new question on the same idea when you paste in a missed one, and a spreadsheet with a "next review" date column handles the schedule.
Tali does both automatically: questions you miss go into a review book that brings them back on a spaced schedule, and you can ask for a new question on the same point instead of seeing the one you remember. But the method works on paper. Log the miss, name why, and come back to it — that's what turns a plateau back into a slope.