Method · How to study for this exam
How to get more out of every hour you study.
Two people can put in the same hours and walk out of this exam with very different scores.
This is the method our team built But For on, written out in enough detail that you could use it with materials you already own. The last section is what no method can do for you, which matters just as much when you’re deciding where your hours go.
But For · practice for the NextGen bar exam
Why getting through the material isn’t the same as being ready
Most bar courses measure your progress by how much you’ve gotten through: lectures watched, outlines finished, subjects checked off. That’s easy to count and it feels like progress. The problem is that the exam doesn’t ask you to watch or read. It asks you to answer, and those are different skills.
There’s a real difference between recognizing a rule and being able to produce one. When you reread your outline the rule is right there on the page, it looks familiar, and you feel like you know it. On exam day nothing is in front of you but the facts and six answer choices, two of them right and four of them written to look right. You have to pull the rule out of your own head, and the only way to get better at that is to practice doing it.
The NextGen exam makes this harder to avoid. It grades you on what a lawyer actually does, instead of assuming you can do it because you remembered the rule. It also hands you the governing law fairly often and asks you to apply the text in front of you, even where that text reaches further than the rule you learned. Knowing more doctrine doesn’t, by itself, get you any of that. The exam page has the full breakdown of how NCBE builds the exam, with sources.
So the question isn’t really what to study. It’s what to do with the hours you’ve got. Our answer: answer real questions from memory, in the formats the exam actually uses, with the subjects mixed together. Then read feedback that makes you think, and keep an honest record of how sure you felt each time you got one wrong.
Five things that actually work
Each of the five below is a well established finding about how people learn, followed by what it means for your prep. None of it is ours, and you can use all five with any materials you like.
Retrieval practice, not rereading
Pulling an answer out of your memory is what makes it stick. Reading the answer again doesn’t do that. It just makes the material feel familiar, and familiar feels an awful lot like knowing it. Psychologists call this the testing effect. It’s one of the best confirmed findings they have, and one of the most ignored, because rereading feels smooth and making yourself remember feels like wasted effort.
So what do you do with that? Build your sessions around questions you might get wrong, rather than pages you can nod along to. Keep your outlines as reference and go to them when a question exposes a hole, instead of reading front to back and hoping the holes turn up on their own. If a session never risked being wrong, it probably didn’t teach you much. Which brings us to the bench. That’s our name for the practice app that comes with membership: the question bank, the sets, the tasks, the grader, and your record, in one place. There’s nothing on it to sit and watch.
Practice in the exam’s real formats
Whatever conditions you practice in are the conditions you get good at. Learn in one format, then take the exam in another, and part of your attention on the day goes into translating between the two, at exactly the moment you have none to spare.
So what do you do with that? Treat the format as part of the material, not as packaging around it. If you spend months on four-option questions and then meet six options with two right answers, you have a translation problem sitting on top of a legal one. There’s more on this further down, under the real formats.
Mixing subjects, and why it feels worse than it is
Studying one subject at a time, all Contracts on Monday and all Evidence on Tuesday, quietly answers the exam’s first hidden question for you: what kind of problem is this? When every question on the page is a Contracts question, you never once have to work that out. Mixing subjects puts the question back, and yes, it feels worse while you’re doing it. Mixed practice usually scores lower during study and higher on the exam itself. So when it starts feeling harder, that’s the method working, not you falling behind.
So what do you do with that? Study one subject at a time only on your first pass through new material. After that, mix, and keep mixing. Judge a mixed session by what it showed you, not by the score it gave back. Your score will drop, and that drop will tempt you back to one subject at a time, which makes the number look better and teaches you less.
Feedback that explains, right away
Learning that you were wrong teaches you almost nothing. What teaches you is seeing why we built the option you picked to look right, and what would have to change in the facts for it to actually be right. That’s the part that carries over to questions you’ve never seen before.
So what do you do with that? Spend about as much time reviewing as answering; answering is only the setup. Two habits do most of the work. Read the explanation for the option you actually picked, not just the one for the right answer. And if you got a question right but couldn’t say why, treat it as one you got wrong, because on slightly different facts you would have. Every question on the bench explains every option and ends with one line that tells the close calls apart, short enough that you’ll still have it in the exam room.
Knowing what you actually know
The four above improve what you know. This one improves how well you can tell what you know, and that drives nearly every decision you make on exam day: when to commit, when to flag something, when to move on. It’s the least practiced skill in bar prep and one of the most valuable, so it gets its own section.
How sure you were, and why we ask every time
Almost every study product records whether you got the question right. Very few ask whether you thought you had it. That second question turns out to matter more, for a practical reason rather than a philosophical one.
Why being sure and wrong is the miss that matters most
Wrong answers come in two kinds. There are the ones you half expected, where you’d narrowed it to two options and guessed, or you knew the doctrine was thin. And there are the ones you got wrong while feeling completely sure, where you read the facts, saw the answer, picked it, and moved on without a second thought.
The second kind is far more dangerous, for a very ordinary reason: you don’t go back and review the questions you were sure about. A miss you felt shaky on flags itself. You remember the wobble, so you go back to it. A miss you felt sure about hides from the very review that was meant to catch it, and it shows up on exam day untouched. Feeling sure also speeds you up, so you never spend the seconds that might have caught it.
The two usually mean different things. A miss you saw coming is a gap: nothing where the rule should be. A miss you were sure about normally means you do have a rule, and it’s either wrong, or right but attached to the wrong trigger. You aren’t missing information there, you’re carrying bad information, and it gets you the same wrong answer every time that doctrine comes up. A rule like that doesn’t misfire once, which is why a single sure-and-wrong miss deserves more of your time than a pile of honest guesses.
Why we ask how sure you are on every answer
You can’t go back later and recover how sure you felt. Ask yourself a week afterward and you’re answering from a memory that already knows the right option, and it quietly rewrites itself into almost had it. The only honest moment is the second before the answer appears, so that’s where we ask. Every answer gets the question, rather than a sample of them, or one prompt at the end of a session.
The question is blunt on purpose: Sure or Unsure, one tap. A percentage slider would make you stop and think, slow every question down, and turn a gut feeling into an estimate, which is a different and less useful thing. Two buttons are fast enough that you’ll stay honest across hundreds of answers, and staying honest that many times is what makes the number worth anything.
Once you have enough answers, two numbers appear that a raw score can’t give you. The first is a simple count of the questions you answered Sure and got wrong. The second is the gap between how you score when you say you’re sure and how you score when you say you aren’t. Your accuracy tells you where you stand; that gap tells you whether you can trust your own read on where you stand. You’ll be making exactly that call all day long in a nine-hour exam.
What to do with a sure-and-wrong list
Four ways to use it, roughly in order of how much they help.
Start every review with the ones you got wrong while feeling sure, as a group. Not scattered through a session in the order you happened to answer them, next to questions you got right, where they disappear into everything else. Pull them together and go through them in one go, because what you’re looking for is the pattern across them. For each one, do more than reread the explanation: say out loud what rule you were actually running on when you answered it. That part is uncomfortable, and it’s the whole exercise. On the bench it happens by default, because review opens on those questions first.
Read them by subject, not just as a total. These misses cluster. Four in Civil Procedure and one in Evidence isn’t a general problem with your confidence, it’s a Civil Procedure problem, and it tells you exactly what to go fix. The total only tells you something is wrong; the breakdown by subject tells you what.
Drill the doctrine again, don’t just reread the question. Rereading an explanation will make you feel like you understand it again inside a minute, and that feeling proves nothing. The only way to tell a rule you’ve actually fixed from an answer you’ve simply memorized is to meet the same doctrine again days later, on facts you haven’t seen. That’s what the gap sessions are for. In one step the bench builds you a session out of your own record: the sure-and-wrong questions first, then questions you haven’t seen from your weakest subjects.
Watch the trend, not the day. Don’t read much into any single session. What you’re watching for is the direction over a few weeks: the gap between how sure you felt and how right you were should be shrinking. If your accuracy is climbing but that gap isn’t, you’re learning more law without getting any better at judging yourself. That’s how well-prepared people get blindsided.
There’s a mistake running the other way, and almost nobody looks for it. If you keep marking Unsure on questions you then get right, you’re selling yourself short, and it costs you real time. It’s why you’ll spend ninety seconds double-checking something you actually had at thirty seconds. On a hard clock that’s a minute spent on nothing. It feels like being careful and it works like a pacing problem. None of this is about talking yourself into more confidence; it’s about getting your own sense of certainty to tell you the truth, in both directions.
What the habit buys you on exam day
What you’re after is a sense of certainty that means something. If the questions you mark Sure are almost always right, you can move quickly through those on exam day and spend the minutes you save on the ones you flagged. If your Sure is barely better than your Unsure, you have nothing to sort by, so every question gets the same second look whether it needs one or not. On a nine-hour exam you pay for perfectionism, and it costs you before a single answer has gone wrong.
The real formats, and what a near miss costs you
Most courses treat getting the format exactly right as a packaging question. It’s really a training one: each NextGen format builds a specific habit, and a format that’s almost right builds you a habit that’s almost right. The formats page shows how each one works. This section is about what each one trains.
Select-two, with real partial credit. Six options, exactly two right, scored so that getting one earns you something and leaving it blank earns you nothing. That gives you one habit worth building long before exam day: never leave one unanswered. It also asks you to read differently than a four-option question does. You’re weighing options against each other rather than crossing off until one is left, and the examiners build the wrong ones to look like the right ones.
Integrated sets, where the facts arrive in stages and the record keeps growing. These train something an ordinary question bank can’t: holding one matter in your head while it keeps changing, and answering each question on a record that is always already sufficient. You are never missing anything you need: no component asks for a fact that has not arrived yet. So the two ways to lose points here are stalling on a question you could already answer, and leaving a blank you meant to come back to. Both are habits, and habits are what practice is for.
Both kinds of performance task, on a real clock. The standard task hands you a file, a library, and sixty minutes to write one long piece, and it grades your formatting and whether you followed instructions too. The legal-research task goes at research judgment directly. It asks what binds a court and what only persuades it. It also asks what a case actually held, as opposed to what it said in passing, and whether the library you were handed even answers the client’s question. Telling real authority from the rest is a skill the MBE never touched, and you don’t pick it up by reading about it.
The law they hand you does not replace the law you know. Where the exam gives you a statute or an edited opinion, your job is to apply that text as written, and you need the doctrine to see what that text is doing to the rule. Practicing that takes questions built around the trap, and the trap is that what you already know feels like it’s helping.
Getting the format close but not exact has a specific cost, and it’s worth being clear about. Anything you practiced in an almost-right format, you have to translate on exam day, and translating eats the attention your hard questions need. So we built the bench format for format: 1,194 original questions, 750 select-one and 444 select-two with real partial credit, eighteen integrated sets, and twelve performance tasks across both kinds. Our team writes every fact pattern and every option in a made-up state and checks it against NCBE’s published samples before it reaches you, so nothing echoes them.
What a good study record looks like over time
Early on
Expect the first few weeks to look bad, and don’t read much into it. Early accuracy mostly measures how recently you saw the material, not how well you know it. Three other questions are worth asking instead. Are you answering questions rather than reading about them? Are you doing it often enough that your record shows a pattern? And are you marking how sure you were honestly?
That third one is where early records go wrong. Nobody is sure and wrong zero times in week one. What it usually means is that they’re marking everything Unsure to avoid the sting of being sure and wrong. That’s comfortable, and it destroys the one signal worth collecting. You’re the only person who can skew this number, and the only person it costs.
The middle stretch
Once you have a few hundred answers on the record, four things genuinely mean progress. Your accuracy is rising in the subjects you actually drilled, especially your weakest ones. The gap between how sure you felt and how right you were is narrowing, even in weeks when accuracy is flat, which means you’re learning to read yourself faster than you’re learning law. Your pace is settling onto each question’s time target rather than beating it, because fast and wrong is not progress. And your accuracy is improving on the harder questions, since the easy ones top out early and stop telling you anything.
Three things look like progress and aren’t. Accuracy pushed up by questions you’ve seen before and now remember rather than work out. Improvement that shows up only in single-subject sessions and vanishes once the subjects are mixed. And volume without review, which is the most common way to spend a great many hours and change very little.
Reading your own weaknesses honestly
Sort by sure-and-wrong before you sort by accuracy. Your worst subject by accuracy is often just your newest one, and more exposure will move it. A pile of sure-and-wrong misses is a different and more urgent problem. What you know there is wrong rather than missing, and wrong doesn’t fix itself by doing more questions.
Take the uncomfortable reading. When you can read a number two ways, the pessimistic reading is usually the useful one, because your study habits already assume the optimistic one.
Don’t average away the outlier. One subject sitting well below a respectable overall number isn’t a rounding detail. The exam covers a lot of ground, and you have no way to know which parts your day will lean on. Your weak subject is the one most likely to cost you, and your average is the number most likely to hide it.
A word on bands, since a study record invites the question. The NextGen reports on a 500 to 750 scale, each jurisdiction sets its own passing line, and results have historically arrived months after the exam. Any band a study product gives you, ours included, tells you how you did on that product’s material. It doesn’t predict your score. We call ours provisional, which means we don’t have enough real data yet to stand behind the number, and that label stays on until real data says otherwise.
The three screens, at full size
Everything this method promises has a screen where it actually happens, and three of them are below at reading size. You get a real grade from the grader members use, your record of every answer you’ve given, and the weekly Benchmark. These are samples, and each one matches what members actually see.
A real grade, marked point by point
Roughly a third of your score is written work. Written work is where it’s easiest to fool yourself: you read a model answer, nod along, and it feels as if you could have written it. So here is a real grade, unedited. The grader gave one point and quoted the writer’s own words to show why. It refused the other and wrote out the reason where you can read it. The same grader will grade an answer you write.
The gradeMarked element by element
AI grade · assistive, not authoritative
The task: identify two specific facts supporting that the deposit retention is presumptively an unenforceable penalty.
The answer as written, with the grader’s marks“The retained deposit is $12,000 on a $30,000 contract, 40% of the price, far above the fifteen percent threshold, so retention is presumptively an unenforceable penalty under 14-2-7(c). Second, Silver Fir Hall made no effort to rebook the date, so it cannot claim mitigation offsets under subsection (d).”
CreditedRefusedThe grader’s marks, on the answer’s own words.
First item matches representative answer 1: “40% of the price, far above the fifteen percent threshold.”
The second item, “Silver Fir Hall made no effort to rebook the date, so it cannot claim mitigation offsets,” is not among the representative answers (which credit the 75-day advance notice/rebooking window, the all-cancellations forfeiture clause, or the 40% figure), and asserts a fact about the venue’s conduct rather than the timing of cancellation.
The app puts it this way: “AI grading is assistive, not authoritative. The rubric is.” On integrated sets and performance tasks you can enter your own score here. Yours replaces the model’s in every total after that, and we record the change.
Your record, split by how sure you were
Every answer records how sure you were, and your record then compares how you score when you’re sure with how you score when you’re not. Being sure and wrong is the most dangerous kind of miss in bar prep, because it never feels like a miss.
Your recordSplit by how sure you were
| Subject | Accuracy | Sure & wrong |
|---|---|---|
| Civil Procedure | 61% | 7 |
| Real Property | 58% | 5 |
| Contracts | 74% | 4 |
| Evidence | 82% | 3 |
| Torts | 69% | 2 |
Your record lists every subject you’ve answered, sorted by how many you got wrong while feeling sure; these five carry the most. One tap builds a session out of it. The twenty-seven you answered Sure and got wrong come first, then questions you haven’t seen from your weakest subjects. The numbers don’t just report the gap, they build the session that closes it.
- First BenchmarkFiled December 7
- A clean weekFiled December 8 · seven days on plan
- 100 answersFiled December 10
- 250 answersFiled December 25
- Every subject openedFiled January 4
- Three Benchmarks risingFiled January 4
- 500 answers116 to go
Marks are worked out fresh from your record every time you visit, so they always match what you have actually done. None is given just for logging in.
The weekly Benchmark, and its trend
The Readiness Benchmark is a short timed exam you take once a week: questions from every difficulty level, plus one written answer the AI grades. You get a score band and the numbers behind it. The band says how you did on that test, not how you’ll do on the bar; nobody can honestly predict that from it, so we don’t. What matters is watching it move week over week.
The weekly BenchmarkReadiness Benchmark
Readiness Benchmark · your result
Band III · Holding steady at the exam’s level
A score band, not a pass prediction. Nobody can honestly promise you one of those.
| Run | Score | Score, drawn | Sure & wrong | Median pace |
|---|---|---|---|---|
| Aug 4 | 71% | 2 | 71s | |
| Jul 28 | 65% | 3 | 78s | |
| Jul 21 | 59% | 5 | 84s | |
| Jul 14 | 53% | 6 | 91s | |
| Jul 7 | 47% | 8 | 97s |
How the questions are made
A method is only as good as the questions behind it, and questions are easy to write badly in ways that are hard to spot afterwards. These are the six rules everything we publish has to meet. They’re on the about page too, and they sit here because they’re part of the method rather than a separate policy.
- R1What it was like, never what it asked.People who take the exam are forbidden to reveal what was on it. No rule binds us the same way, and we hold ourselves to that line anyway: we don’t ask for exam content, record it, or repeat it. Ever.
- R2We label every claim with its source.We mark official statements official. Press reports carry the number of independent outlets behind them. Inferences say they’re inferences. If we can’t show you where it came from, we don’t say it.
- R3Original, always.Our team writes every fact pattern, statute, and option in our made-up state of Meridian, and checks it against NCBE’s official samples before it reaches you, so nothing echoes them. Nothing licensed, nothing bought in.
- R4Provisional until the data earns otherwise.Score bands and difficulty estimates stay labeled provisional until we have enough real results to drop the label. No claims about how hard the first exam was, no predicted questions, no invented deadlines.
- R5Splits get flagged.Where jurisdictions genuinely divide, the materials say so, instead of pretending the majority rule is the only rule.
- R6The grader explains itself, and you get the last word.AI grading works from a rubric and quotes your own words back as its reason. It’s there to help, not to decide: you can overrule it, and your score is the one that stands.
Rules are just talk unless something enforces them. Every new question has to pass seven checks before it reaches you. If it fails even one, we fix it or throw it out.
Seven checks before you ever see a question.
- Is it complete? Every question arrives whole: the fact pattern, the answer choices, the right answer, an explanation for every choice including the wrong ones, and one line to remember it by.
- Does it teach something new? We check every new question against everything already in the bank, so it deepens what you cover instead of repeating a point you have already practiced.
- Could you confuse it with another question? Every name we invent for a person, a company or a case we check against every other question. If two are close enough to blur together in your memory, we rename one.
- Can the answers be gamed? We count how often each letter turns out to be right, so the answer is never a pattern you could learn instead of the law.
- Does everything still fit together? Whenever anything changes, we recheck the whole bank from end to end. One error anywhere, and nothing new goes out until it’s fixed.
- Is it actually on your exam? Every rule we test has to match a line in NCBE’s published outline, checked against the source. We reject anything the outline doesn’t list, however good the question is.
- Are the wrong answers honest? Every wrong choice has to be a real competing analysis rather than filler, and no choice may give itself away by being the longest. So we score two pretend students: one reads nothing and always picks the longest choice, the other always picks the shortest. If either beats guessing, we throw the question out.
The fourth check exists for the method’s sake. If right answers drifted toward particular letters, a hard-working student would learn the drift instead of the law, and every number the app showed you after that would measure the habit rather than your knowledge. The second and sixth do the same job for coverage, because a question bank that quietly repeats its favorite topics will show you a strength you don’t have.
What this method can’t do for you
Every method has a list of things it can’t do. Here’s ours.
It isn’t a complete course. No videos, no live classes, no tutors. The bench is the practice half, built to sit beside whatever outlines or lectures you already use. If you’ve never seen the doctrine before, questions alone are a slow way to learn it.
It can’t promise you an outcome. No pass rate, no guarantee, no score prediction. Nobody can honestly promise you a result on a licensing exam, and getting more out of an hour still tells you nothing about how your own exam day will go.
It can’t tell you what the exam will ask. There are no predicted questions here and there never will be. We write about what taking the exam was like, never about the questions themselves, and we won’t collect remembered questions, both because it breaks confidentiality and because it’s wrong.
The main thing it measures depends on your honesty. These numbers are only as good as your own marking. If you mark everything Unsure just to be safe, or mark Sure because it stings to admit you were guessing, you’ll get a clean-looking record that measures nothing. No software catches this, and nobody but you will ever know.
It asks you to accept a worse-looking practice score. Mixing subjects and pushing into harder questions pushes down the numbers you see while you study. The trade is deliberate, it’s genuinely unpleasant, and some people quit over it.
It doesn’t reduce the hours. A better method changes what an hour returns. It doesn’t change how many of them this exam takes, and no set of principles gets you out of putting in the work.
The exam itself is young. NCBE held the first one in July 2026, and each jurisdiction will publish results on its own calendar. Anything anyone tells you today about this exam’s curve, how it scores, or how hard it is next to the old MBE is an inference. Ours say so on their face.
What you can check
- Independent of NCBE
- Professor and attorney authored and overseen
- Seventeen years with bar students
- Every claim labeled by its source
- No pass guarantees, no predicted questions
- Seven checks before a question reaches you
The .ai is for convenience. The vetting is not.
Everything above works with materials you already own. The bench is what that looks like in one place: 1,194 original questions in both NextGen formats, eighteen integrated sets, and twelve performance tasks. Add AI grading that quotes your own words back to you, and a record that turns the questions you were sure and wrong about back into practice. Three questions are free to try, and we publish the standard. If that is the practice you want, full access is $399, one payment.