Every abandoned SAT math plan dies the same way. Week one is heroic: three-hour sessions, color-coded notes, a spreadsheet. Week two collides with homework and practice. Week three never happens, and the plan is quietly replaced by guilt and one panicked practice test the weekend before the exam.
So this plan is built backwards from the reason plans fail. The weekday sessions are 30 to 40 minutes, small enough to survive a real school schedule. The heavy lifting happens on three weekends. And every session's content is chosen by your own error list rather than by a table of contents, because four weeks is not enough time to study everything, and it is more than enough time to study your things.
The frame: three tests, one error log
The skeleton of the month is three full-length practice tests, taken on the first, third, and fourth weekends. College Board's Bluebook app has free full-length adaptive practice tests that use the real interface, the real embedded Desmos, and the real adaptive module structure, so there is no reason to rehearse on anything else.
Between tests, everything runs off the error log: one page where every missed math question gets a line recording the topic and which of four kinds of miss it was, a slip, a method gap, a misread, or a time-out. That taxonomy and the redo routine that feeds the log come from the 20-minute redo, and it is the engine of the whole month: the log is what makes week three's studying different from week one's.
Week 1: diagnose, then triage
Weekend: full practice test in Bluebook, real conditions, then the redo session on every missed math question while the test is fresh. Your error log now exists, and it will name two or three topic clusters that account for most of the damage. Those clusters are your syllabus. Everything the log does not mention is officially not your problem this month.
Weekdays, 30 to 40 minutes each: one weak topic per day, but never as a marathon. A short focused set until the newest skill lands twice in a row without help, then stop and move on: spaced small sessions beat piled-up ones by roughly a factor of two in the retention research, and the gap only shows up after a week, which is exactly the delay between now and when it matters. End each session with two or three mixed problems pulled from different topics, because the SAT never labels which method a question wants, and mixed practice is where you train that recognition.
Week 2: tools and habits, not just topics
Content review continues from the log, but week two adds the layer most self-studiers skip entirely: the mechanics of taking this particular test.
- Two Desmos sessions. The digital SAT allows the calculator on every question, and the difference between the students who gain time from that and the students who lose time is a decision habit, not a skill. Drill the decision itself using the playbook in when Desmos is faster and when it costs you: for each practice question, call graph-it or work-it before touching anything, then check your call. Graph-reading fluency itself is trainable anywhere, including plotting a function and tracing its intercepts on a graphing calculator at home.
- One fill-in session. Do a set of student-produced-response questions and enter every answer as a fraction where possible. The entry-box rules, mixed numbers most of all, are free points for the prepared and a live grenade for everyone else.
- One habits session. Read through the seven SAT math mistakes that cost the most points with your error log open, and mark which ones your log already shows. Those go on an index card that sits next to every remaining practice session.
Week 3: the pacing week
Weekend: practice test two. Before it, set your bail-out rule: any question with no visible path after about 90 seconds gets a guess, a flag, and your absence. The math section runs 44 questions in 70 minutes, roughly 95 seconds each, there is no wrong-answer penalty, and the last question is worth the same as the first. During this test, pacing is the thing being practiced; the score is a byproduct. If you want the full evidence for why one stubborn question is the most expensive purchase on a timed test, the cognitive-lab findings in my ACT pacing breakdown apply to every timed math section ever printed, including this one.
Weekdays: the redo on test two's misses, then compare logs. The overlap between test one's log and test two's log is the honest list of what four weeks of your attention has not yet fixed, and it gets the remaining sessions. The topics that appeared once and vanished are done; stop paying them.
Week 4: taper, don't cram
Weekend: practice test three, one week before test day, same conditions, same bail-out rule. Redo its misses. After this test, no new topics enter the month, on principle: a new topic seven days out buys anxiety, not points.
Weekdays: short mixed sets, one rep from every topic in the log including the fixed ones, since mastered material still needs to stay in rotation to stay mastered. The index card of personal mistake habits gets one read before each session. Two nights before the test, stop entirely: the retention research is unambiguous that the final-48-hours marathon adds nothing durable, and sleep is doing memory work no worksheet can.
Test morning is logistics, not math: charged device, Bluebook installed and updated, admission ticket, watch the entry-box rules one last time in the car. You have done the actual preparation already, on twenty unremarkable weekday half-hours, which is the only place SAT preparation has ever really happened.
If you don't have four weeks
The plan compresses honestly, because its parts are ranked. With two weeks, keep the first and last weekends' tests, run one week of log-driven content and one of pacing and habits, and skip nothing else. With one week, take a single test on day one, redo everything, and spend the remaining days on the log's top cluster plus the entry-box rules and the bail-out habit, which pay off faster than any content review. What never gets cut, at any length: the diagnostic test, the redo, and the taper. A plan with those three is short but real. A plan without them is long but decorative.
The whole month on one page
- Week 1: test one, build the error log, triage to two or three weak clusters, short daily sessions ending in mixed problems.
- Week 2: content from the log, plus Desmos decision drills, one fill-in entry session, one mistake-habits session.
- Week 3: test two with the 90-second bail-out, compare logs, spend the survivors' list.
- Week 4: test three, taper with mixed one-rep sets, full stop two nights out.
Adjust the calendar to your test date, not the other way around; the structure holds for the October, November, or December sittings alike. And when a practice problem's solution refuses to make sense at 10 pm and no teacher is awake, MathSolver will walk it step by step for free, which is precisely the moment a solver earns its place in a study plan: after the attempt, in service of the log. Four weeks, three tests, one list of your own mistakes, shrinking. That is a plan you will actually finish.