Loading…
There's no content to “cover” — the dMAT has three fixed Core task types and a passage-based Subject module. Preparation is about getting fluent with those formats and training yourself to solve them in your head, quickly.
That makes it an unusual exam to prepare for. There is no syllabus to work through and no bank of past papers to grind, because none has ever been published. What separates a good score from a poor one is almost entirely familiarity with four task formats and the ability to hold them in working memory under a clock. Both are trainable, and eight weeks is enough.
The 2026 test is on 26 September 2026; registration for it ran until 15 September 2026. Around five focused hours a week for six to eight weeks is plenty. If you have less time than that, the plan below still works — you just start further down it.
What follows is the whole thing in order: what the exam actually looks like, why you should sit a mock before you have studied anything, how to use the official materials properly, the technique for each subtest, the eight-week plan itself, how to review what it produces, and what to do in the final week. If you are short of time, the plan and the review section are the two that matter most.
Most preparation goes wrong before it starts, because people plan around a vague idea of the exam rather than its actual shape. The shape is short enough to state in one panel, and everything below follows from it.
Three subtests — Figure Sequences, Mathematical Equations, Latin Squares — each around 20 tasks in about 25 minutes.
For the APS India route, the General Academic Module: around 90 minutes of passage-based questions.
Three and a half hours of examination time, including a break between the modules.
Digital, single-choice, in English, at licensed g.a.s.t. test centres. Ahmedabad is one of them.
A score from 0 to 200 for each module, plus a percentile.
Three things fall straight out of that. Twenty tasks in twenty-five minutes is about seventy-five seconds each, which is not long enough to try two methods — so the first move for every task type has to be automatic rather than chosen. The Subject module is longer than any single Core subtest, which makes it the worst part of the exam to leave to chance, and it is the part most candidates leave to chance. And three and a half hours is a long sitting: the last subtest is answered by a more tired person than the first, so stamina is a real variable and not a metaphor.
One more consequence, and it is the one that catches people. Because each module is reported as a score with a percentile rather than a raw mark, a practice result is feedback for your studying and nothing more. Use it to find what to fix, not to predict anything. Anything decision-critical belongs on the official sources, which are linked at the foot of this page.
The instinct is to study first and test later, and it is the single most expensive instinct in dMAT preparation. A first mock is not an exam you pass. It is a measurement, and a measurement is worth far more at the start of preparation than at the end of it, because it is the thing that decides where your weeks go.
Sitting a paper you are unprepared for tells you things no syllabus can: how long a Latin square actually takes you rather than how long you assume it takes, whether your concentration survives a full sitting, which subtest quietly collapses under a clock. Waiting until you feel ready also hands the decision to a feeling, and readiness is not something anyone senses accurately from the inside. The cost of being wrong early is one mildly demoralising afternoon. The cost of being wrong late is weeks spent on the wrong thing.
Make it as close to the real format as you can: full length if you have the time for it, on a screen rather than on printouts, in one sitting, and without writing anything down. A baseline sat under comfortable conditions measures your comfortable conditions. Expect the score to be low, and expect that to sting more than you predicted — a cold first paper measures unfamiliarity at least as much as reasoning, and unfamiliarity is the cheapest thing on the list to remove.
Then write down four things the same evening, while you still remember. Where the time went, subtest by subtest, and specifically how far you got before the clock ran out — questions you never reached are a pacing problem and nothing else. Which wrong answers you understood the instant you saw the solution, and which you could not have got with another ten minutes; the first group is carelessness or pressure, the second is method. Where your attention dropped, which on a long paper is usually one specific point rather than a gradual fade. And which formats were simply unfamiliar. That page of notes is your plan. The free mock is enough to produce one.
Generic Indian aptitude books (RS Aggarwal, bank-exam or CAT DILR prep) train the wrong workflow. Use the official g.a.s.t. / d-mat.de preparation PDFs and videos to learn the exact task formats and their solution paths, then practise them here. Quality of format beats quantity of questions.
The official material is the only source with real authority about the format, and there is not very much of it — which is an argument for working it slowly rather than reading it once. Reading every solution through is not the same as being able to solve an unfamiliar task under time. Seven steps turn the material into training rather than reading:
Independent practice is for what the official material cannot give you: volume, varied difficulty, timed sets, and the same skill met again in a new form. The official material defines the format; everything else builds fluency in it. Treat any provider claiming more certainty about the exam than the official sources support — including us — as something to check rather than to trust.
This deserves more than a warning, because the books are not bad and the instinct to buy one is not stupid. They simply train three habits the dMAT removes.
The first is the page. Every method in a CAT or bank-exam book assumes you can offload — a half-formed rule in the margin, two struck-through options, a partial sum parked at the top of the sheet. Take the page away and every one of those small deposits has to be held somewhere else. A candidate who has practised for months with a pen has been training the wrong thing very diligently.
The second is back-solving. A great deal of Indian aptitude technique is really option manipulation: plug each answer in and see which survives. It is a good habit on the exams those books were written for, and the Core module does not reliably offer options of the kind you can work backwards from.
The third is the coverage instinct — the sense of progress that comes from chapters closed and pages covered. There is no syllabus here, so that sensation is simply not available, and its absence makes people think they are not improving when they are. Progress on this exam looks like the same task taking less time and going wrong less often. It is a quieter signal and very easy to miss.
Plenty does transfer, and it is worth taking credit for. If you have been through NEET or CAT you can already sit still and think hard for hours, you are fluent with single-choice questions under a clock, and you know how to keep a schedule through a bad week. Those are the expensive capacities and you already have them. What you are putting down is the recall reflex and the belief that more questions is always the fix.
Each Core task rewards a different first move. Knowing which one to make before you start reading is most of the speed.
Track each figure separately and verbalise the rule (movement, rotation, colour, count) before choosing an answer.
Start from the simplest equation and substitute outward; the letters are small whole numbers, so solve mentally.
Scan the most-filled rows and columns first, force the cells you can, and let the constraints cascade.
Figure tasks feel unpredictable because the information is visual rather than verbal, and the fix is to stop looking for the answer and start testing a short list of properties in a fixed order. Compare two adjacent cells and describe what changed in one sentence; then check whether that same sentence explains the next transition. A rule that works once and fails immediately is not a rule. Check rows and columns separately rather than assuming the direction, and separate the layers — an outer shape that rotates, an inner dot that moves and a shading that alternates are three independent things, and trying to explain the whole image in one sentence invents complexity that is not there.
Two habits do most of the work. Predict before you look at the options — say what the missing figure should contain, then use the options to confirm it, because looking early pulls your attention towards decoy similarities. And beware overfitting: if your explanation needs an “except in this cell”, it is the wrong rule. When a sequence looks inconsistent, compare positions one, three and five against two, four and six; alternation is the most commonly missed family. Review the slow correct answers too, not just the wrong ones, and ask which property you should have checked first.
This subtest does not test advanced mathematics. It tests whether you can reason cleanly about simple relationships while the clock runs, and a steady method beats quick arithmetic every time. Read the whole system before you touch any numbers: the fastest solve almost always comes from noticing a relationship between the equations rather than grinding through one of them. Do two share the same variables? Is one nearly a copy of another?
Then choose your move before you start — add when a variable appears with opposite signs so it cancels, subtract when the coefficients match so it drops out, substitute when one equation already gives a variable cleanly. The classic time sink is starting to substitute, realising adding was simpler, and restarting. Verify by dropping your value back into the simplest original equation; that one-line check catches most sign slips, and when it fails you usually only need to re-examine the last step rather than the whole problem.
Estimation belongs here as well. Round before you calculate and you will know the size of the answer before you know the answer, which is often enough to choose between widely spread options and always enough to catch a result that came out the wrong order of magnitude. Log your errors by cause rather than as “careless”: translated the phrase wrongly, used the wrong variable, copied a sign, solved correctly and answered the wrong quantity. Those four have four different fixes.
Latin squares test whether you can apply several simple constraints at once without losing track. The idea is easy; the difficulty is keeping accuracy while the grid gets visually dense. A fixed scan order is faster than repeatedly searching the whole grid, because it stops you re-checking places you have already been:
The single most useful principle is candidate intersection: a blank cell must take a value still missing from its row and still missing from its column, so the answer is whatever survives both lists. The single most common failure is placing a value because it is possible rather than because it is forced — hold unresolved cells as a small set of candidates in your head and commit only when one remains. It also helps to flip the question occasionally: instead of asking what belongs in this blank, ask where a particular symbol can still go in this row. If it is blocked everywhere but one place, that cell is determined even though it had several candidates a moment ago.
Two endings save time. Use the options as constraints — an option that duplicates a value in its row or column is gone without any further deduction — and stop solving the moment the asked-for cell is proven. Finishing the grid for its own sake is the most common way to run a minute over on a question you had already answered. When you get one wrong, do not look at the final option; reconstruct the first point at which the grid became invalid and write that missed constraint down in one sentence.
For the APS India route the Subject module is the General Academic Module: around ninety minutes of questions attached to academic passages. It is the part of the exam that looks least like preparation, which is why people skip it, and it is the fastest part to improve because the gains come from a reading routine rather than from knowledge.
Run the same four stages on every passage regardless of subject. Identify what the question actually wants — a calculation, an inference, a comparison, the best-supported conclusion — before you process a single detail. Locate the evidence, treating the passage as a reference document you search rather than a story you absorb. Apply the method. Then verify: the right unit, the exact question asked, nothing that goes beyond what the evidence supports, and the best option rather than merely a plausible statement.
Read for structure first and detail later. Your first pass is for building a map — what each paragraph is doing, not what it contains — because details are cheap to look up once you know where they live. Watch the turning words: however, in contrast, therefore. Questions are set at those points, because that is where a careless reader misreads. Then let each question send you back to a specific sentence. Even questions asking what a passage implies are anchored; if you are justifying an answer with something the passage never states, you have drifted from comprehension into assumption.
Expect the topics to be outside your field, and treat that as the format working rather than bad luck. The passage cannot assume knowledge you may not have, so terms are defined where they appear and relationships are stated rather than assumed. Demote the terminology to a label: you do not need to know what some unfamiliar process is in the wider field, only that the passage says it increases with temperature and damages the joint. The real trap runs the other way — a passage from your own discipline, answered from memory, when the text has simplified or restricted the case. Every answer has to be supported by something on the screen.
Two more things pay for themselves. Absolute language — always, proves, only, impossible — makes a claim stronger than most evidence in a passage supports, so an option carrying one deserves a second look. And when a table or chart appears, read the frame before the data: title, axes, units, whether the figures are totals, averages or percentages, and what period is being compared. Do not start arithmetic until you know what each number represents.
If the module is costing you time rather than accuracy, the problem is usually re-reading rather than slow thinking. Most slow reading is the same words twice — eyes jumping back out of a vague worry that something was missed. Widen what you take in per glance on the connective tissue of a passage, and slow right down on the sentence that carries the argument or the number a question will hinge on. Fast reading earns its keep by buying you the time to be slow where it matters. Fifteen minutes a day of serious writing from outside your subject, summarised afterwards in two sentences from memory, trains all of this at once.
The Core module is meant to be solved mentally. Working memory is trainable — build it in three stages instead of trying to go paper-free on day one:
Work through the official materials with full notes. Say each solution path out loud so it sticks.
Solve by pointing and speaking only — no writing, untimed. Add General Academic passage practice twice a week.
Solve entirely in your head, working toward ~60 seconds per task. Sit full-length mocks and review errors by rule type.
Working memory is the small mental workspace where you hold and manipulate things for a few seconds — the two partial results you are carrying, the rule you have half-worked-out, the option you provisionally ruled out. All three Core subtests lean on it hard, and it is the reason two candidates who both know how to solve a question type can perform completely differently under exam conditions. The one whose workspace is full loses accuracy not because the method failed but because a piece of it fell out at the wrong moment.
Be sceptical of brain-training games; the evidence for that kind of transfer is modest. What works is direct and unglamorous. Practise under the real constraint rather than saving it for later, because if you always lean on paper you never build the capacity to work without it. Chunk aggressively — working memory holds a handful of items, so the trick is to make each item carry more, seeing a figure transformation as one named move rather than three separate changes. And train little and often: five focused minutes of head-only work on most days beats an hour once a week, and is far less draining.
Treat the moments when your head feels full as data rather than as failure. They point at the exact step where your method has too many loose parts, which is usually a sign to chunk harder or to rehearse that step until it runs by itself. Weeks 3 and 4 above will feel like going backwards. That is the load you are training, and it is the whole point.
Most preparation happens on paper: you print a set, spread it out, circle the promising options, scribble a step in the margin. Then you sit an exam where the questions are on a monitor, the answers are selected with a mouse and the working happens in your head. The gap between the two is small enough to ignore for months and large enough to notice on the day.
The mechanical losses are quiet but real. You cannot annotate a figure to mark what changed between two frames. You cannot put a diagram beside its options by folding the sheet. Your eyes travel between a matrix at the top of the screen and the choices below it instead of taking in a page at a glance. None of that is difficult — it is unfamiliar, and unfamiliarity is expensive when the clock is running.
So stop printing. Use a real computer rather than a phone, because a phone reflows text and shows you one element at a time while the exam adapts to nobody. Read at full size and work one question at a time instead of scrolling between several. Do not annotate the file: highlighting a PDF is the paper habit wearing a different coat. Keep the desk as it will be, with notes and formula sheets put away, since practising with a reference within reach quietly inflates your results. And use official on-screen material wherever it exists, in preference to a better-looking third-party set on paper.
Without a pencil, keeping your place becomes a matter of structure rather than effort. Scan in a set order so you are never re-checking at random. Use the cursor as a pointer — it is the one substitute for a finger on the page, and it holds a row or a cell while your attention is elsewhere. Name things instead of marking them: a rule you can say survives in your head far better than a shape you are trying to picture, and for elimination it is easier to carry the short list of options still standing than the longer list of ones you have killed. Expect the first few screen sessions to feel worse than paper did. That drop is a cost you were going to pay anyway, moved to a week where it does not count.
One thing to do each week, and one way to tell it worked. The checks matter more than the tasks: “practise more” is advice everyone already has, and what people actually lack is a way to know whether a week was any use.
Read the official g.a.s.t. materials for all four task types and work ten of each on paper, notes allowed, no clock. Write down the solution path for each in your own words.
You know it worked when: You can say what each task type asks for without looking it up.
Drill each Core type separately, still on paper. For figure sequences, name the rule out loud before choosing — movement, rotation, colour, count. For Latin squares, find the forced cell first every time.
You know it worked when: You recognise a rule type within a few seconds of seeing the matrix.
Same drills, no writing at all. Point at the grid and talk through it instead. It will feel much slower — that is the working-memory load you are training, and it is the whole point.
You know it worked when: You finish a Latin square in your head, however slowly.
Two General Academic passage sets a week from here on. Read the questions first, then the passage — the passage is long and most of it is not being asked about.
You know it worked when: You stop re-reading passages from the top.
Time individual questions, not whole sections yet. Aim at roughly 75 seconds per Core task, which is the real pace with a little slack. Stop when the timer goes even if you were close.
You know it worked when: You abandon a stuck question instead of sinking three minutes into it.
One Core module end to end, then review it by rule type rather than question by question — group every error by the rule you misread and you will usually find two or three causes, not twelve.
You know it worked when: You can name your weakest rule type, not just your weakest subtest.
A full mock in one sitting, at the time of day your real test is booked for. Review it the same way. Then drill only the rule types that failed.
You know it worked when: Your accuracy on last week's weakest rule type has moved.
Short daily drills to stay warm, one light mock early in the week, nothing new. Sort out documents and travel instead — that is the part that actually goes wrong on the day.
You know it worked when: You are bored rather than panicking. That is the right state to arrive in.
Five to six hours, spread out rather than stacked. Almost nobody preparing for this exam has an empty calendar, so plan around the week you actually have rather than your best imaginable one — four protected thirty-minute slots beat a heroic three-hour block that never survives contact with a late shift. Assign each slot its task in advance, too, because the enemy of a short window is spending half of it deciding how to spend it.
Rest is part of the plan and not the thing you sacrifice to make room for it. Tired preparation teaches you very little: you retain less, you make slips that produce no useful information, and you arrive depleted. Build the slack in ahead of time so that a heavy week costs you a planned light session rather than a guilty missed one.
Almost every candidate could answer most of these questions correctly given unlimited time. The exam does not offer unlimited time, so a substantial part of what it measures is how you behave when the clock is genuinely tight. That is trainable, and it is trained separately from technique.
Start with a budget. Around twenty tasks in about twenty-five minutes is roughly seventy-five seconds each. You do not need to hit that on every item — you need an internal sense of when one is running long, and that sense only comes from timed practice. Then decide your skip rule before test day rather than in the moment: if a question clearly exceeds the budget, mark it, take your best current option and move on. Nothing punishes a score more than three minutes on one hard item and five easy ones never reached.
Work in two passes. The first pass takes everything whose route is reasonably clear; the second returns to the items that need longer calculation or more pattern testing. The point is not to avoid hard questions forever but to stop one of them taking time away from several solvable ones. Bank time where you can go quickly, because that banked time is what lets you think calmly where you cannot.
When a question resists, stop solving and start eliminating. The answer is already on the screen, so the useful question changes from what is the answer to which of these can I prove is wrong. Start with the extremes in quantitative items, where a rough sense of scale often kills the largest and smallest options outright. Test each option against whatever part of the rule you are sure about — an option that fails on a step you trust is gone regardless of what remains uncertain. Look for the option that is almost right but carries one wrong detail. Then compare the survivors: once you are down to two, the difference between them usually points straight at what the question is testing. Eliminate on evidence, never on feeling; this option looks unlikely is not a reason.
Two more rules. If you have narrowed four to two and cannot separate them, take the stronger candidate and move — a reasoned two-way choice is a good outcome for a hard question, and lingering to force certainty costs more than it returns. And do not let one question rattle you: rushing spreads errors across everything that follows, so one slow breath and a reset is a pacing strategy rather than a luxury. Check the current marking rules for your sitting on the official sources before you decide how to handle the items you could not answer.
Most preparation happens one subtest at a time, and then a full mock comes back worse than the practice suggested, with the damage concentrated in the opening minutes of each new subtest. Usually nothing has gone wrong with the methods. What has gone wrong is the changeover, and it is the least practised part of the Core module and one of the cheapest to fix.
Two things go wrong at a seam. Carry-over: the mode you were in persists for a minute or two, so coming out of equations you attack the first figure matrix analytically, decomposing every element in order instead of letting the pattern surface, and spend ninety seconds on something you would normally read in twenty. And residue: you are still arguing with the block you just finished, which from the inside feels like concentration.
The fix is a reset short enough to survive pressure. Sit back for one breath and change your posture — a physical change is the clearest signal that one thing has ended. Name the mode you are entering in three or four words, the same phrase every time: look, do not calculate; substitute, then check; what is forced first. You are not motivating yourself, you are loading a method. Then decide your first move for item one specifically, not a plan for the whole block.
Practise the joins rather than only the blocks. Take the last three items of one type and the first three of the next, run the reset between them, and repeat that pairing until it happens without thinking — six items is small enough to do daily. Then watch your accuracy on the first three items of each block specifically, because that is where carry-over shows up and it responds quickly once you start looking at it. A finished block is finished: give it one sentence, and run the same reset you would after a good one.
This is the step most people skip, and it is worth more than the mock itself. Going through a paper question by question tells you which twelve questions you got wrong. Grouping them tells you why, and there are usually only two or three reasons.
Before any of that, sort the paper into four groups rather than two: correct and efficient, correct but slow, incorrect, and guessed or low-confidence. The second and fourth groups are where your next drop in score is hiding, and they are invisible if you only look at what you got wrong. Then give every problem a cause. Avoid the word careless — it names a feeling rather than a behaviour, and there is nothing to practise on the other side of it.
You did not have a reliable approach and improvised. Fix with slow, untimed practice of that one task type until the first move is automatic.
You misread a phrase, a diagram, a unit or a qualifier. Fix by pausing deliberately on the small words — not, except, least, next, only.
The method was right and the arithmetic or the placement was wrong. Fix with a one-line size check before you commit.
You overspent, or never reached the question. Fix with short timed sets and a skip rule decided in advance, not with more content.
You selected, flagged or navigated wrongly. Fix by practising on screen and rewatching the official interface videos.
You changed a right answer without evidence, or submitted a guess as if it were sure. Fix by recording your confidence before you see the result.
Keep the causes in one running error log rather than in the margins of individual papers. A single review tells you about one afternoon; ten entries on one page show you a pattern — that half your losses come from misreading the question rather than from the reasoning, or that you rush the opening items of every timed set. Keep each entry to a line: the task type, what actually went wrong, the one move that would have flipped it, and whether it was a knowledge gap or a process slip. If logging a mistake takes three minutes you will stop doing it.
The log also sets your mock frequency, which should be counted in review cycles rather than papers. Sitting one on Monday and another on Tuesday is two tests and no cycles. One full mock every seven to ten days through the middle of your preparation gives each paper a week to change something; two or three in the final fortnight is enough, by which point the goal has shifted from learning to rehearsing pacing, stamina and calm. The honest test is simple: after each mock, can you name one specific thing you will do differently next week? If not, either the reviews are too shallow or you are testing too often to think in between.
Our free mock gives you the per-subtest breakdown for this, and the free practice questions let you drill a single rule type once you have found it.
On a first timed attempt, most lost marks are not lost to weak reasoning. They go to the format. The clock runs while you are still reading the instructions. A grid you would have solved on paper in twenty seconds takes ninety in your head. You reach for the trick of working backwards from the options and find the options are not the kind you can work backwards from.
This is good news, because format problems are the cheapest kind to fix. It is also why the order of the plan matters: going straight to timed mocks in week one measures your unfamiliarity rather than reducing it.
Most people improve quickly at first, then settle into a narrow band that refuses to climb. A plateau is almost always a signal rather than a ceiling: it means the easy gains are collected and the way you are practising has stopped matching what you now need. The usual cause is practising on autopilot — set after set, mostly reinforcing what you already do well, with review thin enough that the questions you got wrong teach you nothing.
Diagnose before you change anything, and do it with one specific test. Take a set of your weakest subtest and do it untimed. If accuracy is still poor, it is a technique gap: you have no reliable first move and each item is being improvised. That is the case where concentrated work on one task type is exactly right. If accuracy is fine untimed and collapses under the clock, it is a speed gap, and more question sets will not touch it — what is broken is fluency and pacing, not understanding. The two look identical from the outside and their fixes barely overlap.
There is a third possibility worth naming, because it is specific to this exam. A subtest can look like a reasoning weakness when it is really a holding-things-in-mind weakness — you lose the partial result before you can use it. If that description fits, the working-memory work above will do more for you than another block of practice questions.
Then allocate by the principle that early points are cheap and late points are expensive. Moving a subtest from poor to competent means learning two or three methods and applying them consistently; moving one from strong to excellent means chasing the hardest items in the set. So the weak subtest gets the larger share of new learning — but a majority, not a monopoly, because a strength is only banked if it survives the clock, and a subtest you have not touched for a month will cost you more than the weak one gains. Maintenance is cheap. Reconstruction is not, and late in the plan you should be maintaining rather than rebuilding.
Do not compress all eight weeks into the time you have; that produces a shallow version of every stage. Drop stages instead, keep the order, and accept a narrower plan you can finish.
Run weeks 1 and 2 as a single week — learn the formats fast, on paper, then put the pen down immediately. Give weeks 3 to 5 two weeks, and keep the last fortnight for timed modules and review.
Sit a timed mock on day one, no preparation at all. Spend eight days on the single weakest subtest plus mental solving, two days on the Subject module, and the last two on a light mock and logistics.
One mock, then pacing only. Fix a per-question budget, practise abandoning questions, drill the one rule type that failed most, and protect your sleep. Nothing new goes in this week.
The same logic applies if a plan has slipped rather than started late, which happens to almost everyone. Resist scrapping it — the sequence of method, then accuracy, then rehearsal is as sound as it was on day one, and only the calendar has changed. Mark what you genuinely completed rather than what you intended, count the usable hours that actually remain, and rebuild around the two or three task types your error log says are costing the most. Focused imperfection beats even, shallow coverage. And set a floor rather than a ceiling: three or four protected hours you will really keep are worth more than an eight-hour plan that collapses again next week.
The reflex in India is to ask where to enrol rather than what to study, and the dMAT meets that reflex with a market that barely exists yet: few established courses, no long history of past papers, no institute with a decade of results behind it. The useful question is narrower. A course sells four separable things, and self-study is not a lesser version of it — it is the same work with those four things sourced differently.
Structure is a schedule someone else has already thought about. Supply it by working backwards from your test date: count the weeks you actually have, subtract the ones you know will go to work, travel or your own university exams, and commit what remains to named slots in the week. Named times survive; vague intentions do not. Revisit the plan once a week rather than every day, because daily renegotiation is how a schedule quietly dissolves.
Material is questions close enough to the real thing to be worth your time. The instinct without a course is to compensate with quantity — two or three question banks and assorted PDFs from messaging groups. That is usually a mistake: material that does not resemble the exam trains you for something else, and material of unknown provenance can be confidently wrong. Start official, then prefer a smaller amount worked carefully to a large amount skimmed.
Feedback is the function self-study genuinely struggles to replace, so it deserves the most deliberate effort. Without a teacher you have to become the person who explains your own mistakes, and that is a habit rather than a talent: the written error log above is the whole mechanism. One technique costs nothing and works surprisingly well — explain a solution aloud, as if to someone else, without looking at your notes. The points where you turn vague are the points you have not understood yet.
Accountability is the easiest to replace and the one most people skip. You do not need a classroom, but you probably do need someone who notices. A single person preparing on the same timeline is enough; failing that, tell someone at home what your weekly target is and report against it. Being mildly answerable to one person does most of the work a paid seat does. What a course cannot give you is understanding — that is produced by you, at a desk, over weeks — which is why what separates candidates has far less to do with where they studied than with whether they reviewed.
The last seven days are not the time to learn anything. By then your methods are formed, and the real risk is a week of anxious cramming that leaves you tired on the morning that matters. Taper: short focused sets in your two weakest task types early in the week, one final timed simulation around day four, then light revision of your own notes only. Resist a fresh difficult mock two days out — a poor score then shakes your confidence far more than it improves your reasoning.
Review what you already know rather than what you fear. A topic you have avoided for eight weeks will not become a strength in seven days, and the hours spent forcing it come out of the methods that already earn you marks. Read your error log slowly instead, find the two or three mistakes you keep repeating, and rehearse the correction rather than the error. Sleep is part of preparation at this point and not a luxury traded for another hour: reasoning speed, working memory and staying calm under time pressure all run on being rested.
Settle the logistics two or three days ahead so the morning contains no decisions — the route and how long it realistically takes, the documents and identification, the reporting time, and what you are permitted to bring. Check that against the official sources and your own participant instructions rather than against anything you read online, this page included. Then give yourself a short warm-up on the day: three or four easy questions from a topic you are comfortable with, fifteen minutes at most, deliberately avoiding your weakest area. A warm-up that ends in a mistake plants doubt at the exact moment you need steadiness.
Plan the break between modules too, because it is the one stretch of the day with no structure and most people arrive at it with no plan. Keep whatever you intend to eat light and familiar, take the washroom stop early rather than late, stand up and look at something further away than a screen, and decline the two conversations on offer — the one with yourself about the module just finished, and the one in the corridor comparing answers. Neither has ever improved anyone's score. Spend the last minute bringing to mind how you intend to open the module ahead. Rehearse all of it in at least one full-length practice run, including the gap, so that on the day it is a routine rather than an improvisation. Certificates for this cycle become available on 12 October 2026.
Around five focused hours a week for six to eight weeks is enough for most people. The limit is not how much content there is — there is almost none — but how long it takes to get comfortable solving these formats in your head without rough paper. If you have done GRE or GMAT preparation recently you can compress the first two weeks.
They will not help much and can actively slow you down. Those books train a workflow built on rough paper and on eliminating multiple-choice options, and the dMAT Core module gives you neither. The reasoning is genuinely different, not just harder or easier.
No. No official paper has been released, so there is nothing to memorise and anything advertised as a past paper is not what it claims to be. What you can rehearse is the format, which is why practice written to the official task structures is the only useful kind.
Yes, but change the order. Skip the two weeks of paper-based learning, sit a timed mock immediately to find your weakest subtest, and spend what time you have on that one thing plus pacing. A narrow, honest plan beats an ambitious one you cannot finish.
Enough that leaving it to guesswork is a mistake. It is the one part that rewards a reading strategy rather than a solving technique, which also makes it the fastest to improve — reading the questions before the passage is often worth several marks on its own.
No. A course sells four things — structure, material, feedback and accountability — and you can source all four yourself: a schedule built backwards from your test date, the official materials plus format-accurate practice, a written error log in place of a teacher, and one person who asks how the week went. Coaching is a reasonable purchase if you have tried to start three times and not managed it, but it is not a requirement.
Prepare as though you cannot. What you may take to the desk, and what the centre provides, is set out in the rules for your sitting rather than on the public pages — confirm it there. Either way the Core subtests are built to be answered without written working, which is why the plan spends two whole weeks weaning you off it rather than leaving it to test day.
Count review cycles, not mocks. One full mock every seven to ten days through the middle of your preparation is a sustainable rhythm, because the week in between is where the improvement actually happens. Two or three in the final fortnight is plenty, and each one still needs a review that changes something.
Around twenty tasks in about twenty-five minutes works out to roughly 75 seconds each, so aim your drills at that and try to get comfortable a little under it. The pace matters less than the habit of abandoning a question when it passes your budget.
Do one set of your weakest subtest untimed. If your accuracy is still poor, it is a technique gap and needs concentrated work on that one task type. If accuracy is fine untimed and collapses under the clock, it is a pacing problem and more question sets will not fix it. The two look identical from the outside and have almost nothing in common.
The plan above is the shape of eight weeks. These go into the parts people most often get wrong.
The same order of work, compressed, with the parts that survive the cut marked.
What to do with the extra time so it becomes fluency rather than more of the same.
What is realistic from a standing start, and what is not worth beginning.
They are the only genuine source of format. How to use them without burning them early.
The test is digital, and on-screen work changes pacing more than most people expect.
The single habit that turns a wrong answer into something you do not repeat.
The review is where the score moves. Sitting another one is not the same thing.
What coaching actually provides, and which parts you can supply yourself.
We link out so you can verify everything yourself. We are not affiliated with any of these bodies.