Everyone tells you to keep an error log. Almost nobody tells you what goes in the columns, which is why most error logs are just lists of missed questions that change nothing. The column that matters is cause — because a question you got wrong from not knowing the rule and a question you got wrong from misreading the prompt need opposite fixes, and lumping them together is why the same mistakes keep coming back.
Rule one: redo it before you read the explanation
This is the step people skip, and skipping it destroys the diagnostic value of everything after.
Open the question again with the answer hidden. Redo it from blank, untimed, as long as you need. Then compare.
What you learn is the thing you can't get any other way: can you solve this when time and pressure are removed? If yes, you don't have a content problem — you have a timing or process problem, and studying the content will do nothing. If no, you have a genuine gap and now you know exactly where.
Reading the explanation first destroys this test forever. Once you've seen the worked solution, "I could have done that" is indistinguishable from "I could have done that." Every explanation is convincing after the fact.
The six causes, and the six different fixes
Every miss goes in exactly one bucket. Force yourself to choose — the discipline of picking one is what makes the tally meaningful.
| Cause | What it looks like | The fix |
|---|---|---|
| 1. Content gap | Couldn't do it untimed either. You don't know the rule, formula, or concept. | Learn the topic, then drill 15–20 of that exact type. The only category where more content study is the answer. |
| 2. Misread | Solved the wrong question. Found x when it asked for 2x; picked the choice that was true but wasn't what was asked. | Process, not content. Underline what's being asked before solving. Re-read the final clause of the prompt before selecting. |
| 3. Method error | Knew the content, chose a slow or fragile route. Algebra where Desmos was instant; reading the whole passage before the question. | Approach drills — same question, three different routes, compare times. Fluency with the tool, not more theory. |
| 4. Execution slip | Right method, broken arithmetic. Sign error, transcription error, mis-keyed value. | A verification habit, applied to a specific trigger. Not "be more careful" — a concrete check you perform every time. |
| 5. Timing casualty | Never really attempted it. Guessed in the last 30 seconds or ran out entirely. | Pacing, not content. Studying these questions is wasted effort — you were never given the chance to be wrong about them. |
| 6. Lucky right | You guessed. It was correct. Your score doesn't know the difference. | Treat it as a miss. Study it as if you'd got it wrong, because next time you probably will. |
Four answer choices means roughly one in four blind guesses comes back correct. Guess on twelve questions and about three of them land — three weaknesses that never appear in your review because the score report calls them right answers. Students who log lucky rights routinely find their real weak-topic count is 20–30% higher than their score report suggests. This is the single highest-value habit on this page, and it costs you nothing but a flag while you're testing.
The log template
Seven columns. A spreadsheet or a notebook both work — consistency beats tooling.
| Column | What goes in it | Example |
|---|---|---|
| Q# | Test and question number | PT4 · M2 · Q14 |
| Type | Domain and question type | Advanced Math — quadratics |
| Mine / Correct | Both answers | B / D |
| Cause | One of the six | 4 — execution slip |
| The trap | Why the wrong answer was tempting, in your words | "D is what you get if you forget the negative root" |
| Fix | The specific action, not "be careful" | "Always write both roots before choosing" |
| Re-test | Date, 10–14 days out | Aug 21 |
The trap column does more work than it looks like it should. Writing down why the wrong answer appealed to you is how you learn to recognise the same bait in a different costume — and the SAT reuses its bait constantly. "The answer you get if you compute percent change off the wrong base" will describe a dozen questions across your whole preparation.
Count before you fix
Once everything's logged, don't start at question one and work down. Tally the cause column first, then study in order of size.
The tally is usually a surprise. A student convinced they need more Advanced Math practice finds nine misreads and two content gaps — which means their next four weeks should go to prompt discipline, not quadratics. Another finds seven timing casualties, meaning their content is fine and their pacing plan is broken.
Those two students would follow identical study plans if they went by question number. Going by cause count, they do completely different work. That's the whole payoff of the taxonomy.
Getting the raw data out of Bluebook
Finish a full-length test in Bluebook, then sign in at College Board's My Practice to see question-level results. Score Details gives you every question, your answer, the correct answer, and an explanation.
Two limitations worth knowing about, because they shape what you can conclude:
- It shows you what, never why. The report can tell you that you missed four Advanced Math questions. It cannot tell you whether that was content, misreading, or the clock — and those need three different responses. The cause column is yours to fill in and it's the entire product.
- It doesn't flag which misses were adaptive artifacts. If you earned the harder second module, some of those questions are meant to be difficult for nearly everyone. Missing two of the hardest items on a hard module is a very different signal from missing two routine ones, and the report presents them identically.
For a fuller walkthrough of what to trust and what to ignore in that report, see how the digital SAT is scored.
The spacing rule
Here's the part that decides whether any of this sticks.
Re-test the same question types 10 to 14 days later, not the next day. Next-day accuracy proves nothing — the method is still sitting in working memory and you'd get them right by pattern-matching to yesterday. Two-week accuracy is the only evidence that something actually changed.
This is why the log has a re-test column. It's not a nice-to-have. A fix you never re-tested is a fix you're guessing about, and most score plateaus are just a pile of untested fixes stacked on top of each other.
How long this takes
Ninety minutes to two hours for a full-length test. That's long, and it's supposed to be — you're spending roughly as much time reviewing as you'd spend on half the test itself.
If your review takes twenty minutes, you're reading explanations and nodding. That feels like studying and produces nothing, which is exactly why students can take eight practice tests and stay flat. The tests aren't what raise the score. The review is. The test is just how you generate the material.
Everything above is a manual version of what the PrepGenix platform does automatically — it tags every miss by domain and type, tracks which questions you flagged versus solved cleanly, and schedules the spaced re-test for you. The method matters more than the tool, so run it on paper if you prefer. But the re-test scheduling is the part people abandon first, and it's the part that makes the rest work.
Frequently asked questions
How long should it take to review an SAT practice test?
Ninety minutes to two hours for a full-length test. A twenty-minute review means you're reading explanations rather than diagnosing causes.
Where do I see my Bluebook practice test answers?
In College Board's My Practice, after finishing a full-length test in Bluebook. Open Score Details for question-by-question results and explanations.
What should an SAT error log include?
Question number and type, your answer and the correct one, the cause category, the trap you fell for, the specific fix, and a re-test date. The cause column is what turns a record into a plan.
Should I review questions I got right?
Only the ones you guessed on. Those are weaknesses your score is hiding — about a quarter of blind guesses come back correct on a four-choice test.