SAT score plateau: why three months of Khan Academy didn't move you

You’ve done three months of Khan Academy, hundreds of questions, maybe a few full tests, and the score has moved 20 points in either direction and settled back where it started. An SAT score plateaus after months of self-study for five specific reasons, and every one of them is mechanical: a missing error log, practice that isn’t timed like the test, grammar answered by ear, a content gap hidden by the module you keep getting routed to, and a review method that confirms misses instead of correcting them.

None of those is about effort or belief, so this guide contains no pep talk. Each cause below comes with the mechanism and a fix you can run before next Saturday.

Cause 1: you’ve been practicing the test, not your misses

A plateau is a repeated set of misses. On College Board’s test, 54 Reading and Writing questions and 44 Math questions, scored 400 to 1600 (the structure is at satsuite.collegeboard.org), a student at 1350 is typically missing eighteen to twenty-two questions. The score is stable because the set is stable: the same eight to twelve question types, wrong in a different order each time. Practicing the whole test keeps the set intact, because every session spends most of its time on questions you already get right.

Practice banks make this worse by design. The software serves what you tap on, and students tap on what feels productive, which is usually the section they’re already better at. Nothing in the system knows that four of your Math misses this month were the same vertex-form gap wearing four different word problems. You have to know that, and you can only know it from a log.

This week’s fix

Build the error log from your last two full-length tests. One row per miss, five columns: College Board question type (Words in Context, Transitions, Boundaries, nonlinear functions, systems, and so on), module 1 or 2, the reason in one honest phrase (didn’t know, misread, rushed, changed a right answer), time spent if you know it, and a blank column for “fixed.” It takes about an hour. Then sort by type and count. Your plateau is those rows.

Cause 2: your practice isn’t timed like the test

Self-study practice runs in 20-minute chunks, on a weeknight, with a pause to check a phone. The real test is 2 hours 14 minutes, starting around 8:30 on a Saturday morning, with two 32-minute Reading and Writing modules and two 35-minute Math modules in a fixed order and a single ten-minute break. A score earned in chunks is not a prediction of a score earned in that format, and a student who has never sat the whole thing under those conditions has a practice number, not a practice score. Chunks don’t count.

The time pressure is the specific thing being skipped. At about 71 seconds a Reading and Writing question and 95 a Math question, the test asks you to decide quickly and move. A student who takes two minutes on a Words in Context question in practice and gets it right has learned nothing about whether they’d get it right at 71 seconds, which is the only version that counts.

This week’s fix

Saturday, 8 a.m., one full-length Bluebook practice test, both sections in order, no pauses except the built-in break, phone in another room. Log it the same way as Cause 1. If the score is lower than your chunked practice has suggested, that gap is your real starting point, and it’s good news in the sense that you’ve found something specific to work on: pace, not knowledge.

Cause 3: you answer grammar and transitions by sound

Standard English Conventions and Expression of Ideas together are roughly 46% of the Reading and Writing section by College Board’s published ranges, and a large share of those questions are written so that the wrong answer sounds fine. A comma splice sounds like a pause. “However” sounds reasonable at the start of almost any sentence. A student who reads well has been getting away with this since middle school, and the SAT is the first test built to catch it.

The tell is in the log. If your Boundaries and Transitions misses are steady at three to five per test and the reason column says “it sounded right,” you have this cause. The reading isn’t the problem. The method is.

This week’s fix

Learn the eight punctuation rules as rules you can say out loud: two complete sentences need a semicolon, a period or a comma plus a conjunction; nonessential phrases get a matching pair; a colon follows a complete sentence; and the rest. Our templates and rules guide has all eight with examples. Then do 20 Boundaries questions saying the rule before you answer, and 20 Transitions questions labeling the relationship between the sentences (contrast, cause, example, sequence) before you look at the choices. Twenty each is enough to feel the method replace the ear.

Cause 4: the module you keep getting is hiding a content gap

The second module in each section is chosen by how you did on the first. If your first module isn’t nearly clean, you’re routed to the easier second module, and the easier module doesn’t contain the questions that decide scores above the mid-600s in a section. College Board doesn’t publish the cutoff, but the effect is visible in Bluebook review: an easier module 2 is full of short passages and two-line linear equations.

The plateau mechanism here is quiet. A student routed down every time never sees a system with a parameter, a quadratic in vertex form, a two-text Cross-Text Connections passage or a hard Words in Context item with 19th-century prose. They can’t be missing those questions, so those questions never appear in the log, so the student never studies them. The content gap is real and invisible at the same time, and the score is stuck at the easier module’s ceiling.

Check your routing first. Then check the topics even if you were routed up, because the hard-module topics are also the ones a standard Algebra 2 class covered in a week two years ago. Our hard module guide lists them with examples.

This week’s fix

Open your last Bluebook test’s review and look at module 2 in each section. If you can’t find a parameter question, a vertex-form question or a two-text passage, you were routed down, and the fix has two parts in order. First, log the module 1 misses and count how many were avoidable; that’s usually the whole routing margin. Second, work one hard Math module and one hard Reading and Writing module as timed sets this week (35 and 32 minutes) so the hard-module topics start appearing in your log where you can see them. Pick the single topic that cost the most and learn it; for most students it’s vertex form or systems with a parameter, and each takes about a week.

Cause 5: your review confirms instead of corrects

The most common review method is to check the answer, read the explanation, say “oh, careless” or “oh, right,” and move on. It feels like learning and it isn’t, because it ends the moment the student understands why the right answer is right. Nothing in that method changes what the student will do next time they see the same stem, which is why the same stem keeps producing the same miss.

“Careless” is the word to watch for. Watch that word. On most plateaued students’ logs, a third of the Math misses are labeled careless, and when we press, “careless” turns out to mean three different things: misread what the question asked (answered x when it asked 2x), rushed the last step, or changed a right answer to a wrong one. Each has a different fix, and “careless” hides all three.

This week’s fix

Replace the review with three written sentences per miss. One: what the question actually asked, in your own words. Two: why the wrong answer you chose looked right. Three: what rule or habit would have caught it, stated as something you can do next time (“re-read the last line before submitting”; “label the relationship before the choices”). It takes about two minutes per miss, so a full test takes 40 minutes to review properly, which is roughly a third of the time it took to sit.

Then redo every missed question cold, three days later, without looking at the log. A question you get right on the redo is fixed. One you miss again goes back in the log with a new row, and that type moves up your list.

Rick’s note: when a student tells me a miss was careless, I ask them to show me the exact moment it went wrong. About half the time they can’t, which means they never actually reviewed it. The other half, it’s the same moment every time, and that’s not carelessness. That’s a habit with a fix.

What the five causes have in common

Each one is a loop with no measurement in it. Practice without a log, timing without the real clock, grammar without rules, a module that hides what you don’t know, and review that ends at understanding. Three months inside those loops produces exactly what you got: a lot of work and a flat line. Measurement is the repair.

The fixes above are free and they’re all this week’s work. One hour for the log. One Saturday morning for the proctored test. Forty questions with the method said aloud. Two timed hard modules. Forty minutes of three-sentence review and a cold redo on Tuesday. Run all five and the next test tells you something, which is more than the last three months did.

If you’d rather have someone read the two tests with you and name the four or five types in an hour, that’s what the free first hour is for. Bring the tests and the log.

What to do this week

1. Build the error log from your last two full-length tests: type, module, reason, time, fixed. Sort by type and count.
2. Saturday, 8 a.m.: one full-length Bluebook test, both sections, no pauses, phone in another room. Log it.
3. Say the rule aloud on 20 Boundaries questions and label the relationship on 20 Transitions questions before answering.
4. Check module 2 of your last test for routing, then work one hard Math and one hard Reading and Writing module as timed sets.
5. Review every miss with three written sentences and redo them cold on Tuesday.
6. Bring the tests and the log to a free first hour if you want the pattern named in sixty minutes.