Loading
The reasoning test that AI summaries can't fake.
The LSAT is the purest reasoning test in the standardized-testing world. It has no content — there's nothing to memorize, no formulas to recall, no facts to know. What it tests is your ability to hold an argument in working memory, identify its structure, and evaluate what would strengthen or weaken it, all under time pressure. That's it. That's the entire test.
This is why generic 'study harder' advice fails on the LSAT and why AI summaries make it worse. There is nothing to summarize. The skill is the reasoning act itself, and the only way to build it is to perform it — attempt after attempt, with feedback on what your reasoning specifically missed.
Brainback is the practice layer for that. Every LR question, every RC passage, every LG game (if you're on a games-still-included administration) runs through Learn mode with the attempt-gate on. You commit to an answer choice + a reason before you can see any feedback. The tutor then names the flaw class you're closest to, gives you one guiding question, and hands the argument back to you.
Over a 3-6 month cycle, the effect compounds — or it doesn't, and you can see which. The drills come from your own preptests: upload them and the tutor works the questions you actually sat, tagging each miss by question type and logical form. We publish no score-gain figure, because we have run no study that would support one.
Total: ~3 hours (including experimental section). Below is every section, with its duration, question count, and scoring role.
| section | duration | questions | scoring |
|---|---|---|---|
| Logical Reasoning · Section 1 | 35 min | 24-26 | counts |
| Logical Reasoning · Section 2 | 35 min | 24-26 | counts |
| Reading Comprehension | 35 min | 26-28 | counts |
| Experimental (variable) | 35 min | varies | doesn't count |
| Writing sample (unscored) | 35 min | 1 prompt | reviewed by schools |
Scaled score 120–180. Median 152. T14 competitive: 168+. Full-ride threshold: 172+.
Rehearsable. Predictable. Adjustable for your calendar. Each week has one focus and one specific Brainback move.
| when | focus | brainback move |
|---|---|---|
| Month 1 | Fundamentals: LR question types + diagramming | One LR question type per week in Learn mode. Explain-back on the flaw class definitions. |
| Month 2 | Reading Comp technique + LR mixed practice | RC passages daily. Recall queue at 100+ cards on flaws and RC structures. |
| Month 3 | Timed sections | Timed sections 3x/week. Untimed error review daily. Brain score breaks accuracy out by question type. |
| Month 4 | Full-length preptest weekly | Every Saturday: a full preptest. Every Sunday: Learn mode on every wrong answer. |
| Month 5 | Weakness kill + timing polish | Any question type with < 80% accuracy gets an extra 30 min daily. |
| Month 6 | Rest week + AAMC-equivalent official prep | One preptest early in the week. No new content. Recall + sleep dominate the last 5 days. |
Timed section from PrepTest 90 as a baseline. Score it. That's your starting number.
A representative exchange: the problem, your attempt, the tutor’s response, your follow-up, the downstream card. Notice the tutor never solves.
LR: 'The mayor's proposed transit plan will reduce congestion because it will discourage private-car use. Therefore, the plan will succeed.' What flaw does this argument make?
It assumes discouraging cars will actually reduce congestion.
That's a real assumption in the argument, but it's not the FLAW being tested here. Look at the conclusion — 'the plan will succeed.' What does 'succeed' mean here? The premise talks about congestion reduction; the conclusion says success. Are those the same thing?
Oh — the argument conflates 'reducing congestion' with 'succeeding.' Success might mean adoption rates, budget, ridership, political viability. The premise only supports one narrow reading of success.
All seven product surfaces are always on. For this exam specifically, these three carry the load. Each has its own dedicated page.
Every practice question in attempt-first mode. The tutor won't diagram it for you.
Explain the flaw class in your own words. Rubric-graded. Card generated for anything under 4/5.
Logical flaws + argument patterns retrieved on schedule, not the night before.
And the specific Brainback mechanism that catches each one. This isn’t a promise — it’s what the workflow is designed to do.
Reading answer explanations without re-attempting the question
Every wrong answer becomes a Learn session; you attempt it a second time before the explanation appears.
Untimed practice never transitions to timed
The weekly cadence forces the transition on a schedule; you can't hide in untimed forever.
'I've done a million questions but my score isn't moving'
Volume without reasoning practice plateaus. Brain score decomposes accuracy by question type so you can see which specific type is holding you back.
Common approaches that feel productive but produce nothing. Skipping these is worth as much as adding a technique.
because · You'll read them and take zero real preptests. Two prep books + PrepTest volumes is the sane stack.
because · If you're on a test date that still includes games, they're 25% of your score and the most learnable section. Learn them.
because · You'll score low, feel demoralized, and gain nothing. Preptests are a diagnostic once you're at the accuracy floor.
because · The LSAT rewards well-rested reasoning. Cramming produces the opposite.
We publish no expected point gain — no study, no number. The LSAT has no content to memorize, so the only thing that moves is the reasoning act itself, and the only way to train it is to perform it. What the cycle guarantees is volume of your own production: you commit to an answer and a reason before any feedback appears, and every miss returns as a card tagged by question type and logical form.
Student annual — $59/yr. LSAT prep is longer than most exams. Annual is right for a 3–6 month cycle. See all plans →
Yes — the tutor walks you through diagramming without producing the diagram. Since games were dropped from the LSAT for many test-takers, this is less critical than before, but still available for those still testing on games.
Different products. Those are content + video-lecture heavy. Brainback is the practice + retrieval + recall layer around them. Use both.
No. We have run no study and no cohort, so any number we printed would be made up. When we have real data collected properly, we will publish it with the method attached.
Not scored, but reviewed by schools. Efficiency mode can help you draft the sample (it's not a graded exam once submitted); Learn mode isn't the right tool.
Modal successful test-taker takes it 1-2 times. More than 3 sittings starts to look like a red flag on your application. Brainback helps compress the number of sittings needed by getting your prep to the ceiling faster.
Personal statement + application essays live in Efficiency mode (not Learn mode). See /for/law-school for the full 1L-through-bar workflow.
The bar is a reasoning exam wearing a memory costume.
→LSAT prep
Student annual — $59/yr · cancel anytime.