Which is more important on the score report: scaled score or percentile? How should I allocate my time for each section?

Percentile is more important than scaled score. Schools look at percentiles (relative ranking). Therefore, you should decide how to invest your time based on the slope of your scaled score (Scale) vs. percentile from your last test or mock test:

Analyze the impact of scaled score improvement in each section (Verbal, Q, DI) on your percentile.

Subjects with a high slope mean that the same unit of ability improvement has a greater impact on the percentile, which schools value. Therefore, you should invest more time in them.

Is the GMAT section order really that important?

GMAC confirmed in an official Q&A in February 2026 that the test does include an inter-section adaptivity element: performance in one section slightly adjusts the starting difficulty of the next. The same answer notes that the impact on the start of the next section is “usually limited” — the first few questions of each section are still drawn from a narrow band of medium difficulty, not from either extreme.

And remember, scoring a basket from the three-point line is different from scoring from the two-point line. So, it’s most important to start from the three-point line (high-difficulty question bank) as much as possible!

Therefore, the conclusion is clear: it’s recommended to take your strongest section first to aim for the best starting difficulty!

What if I rush through questions, will I be penalized?

 GMAC stated at a 2019 test prep seminar that it detects abnormal response speeds (e.g., extremely fast but correct answers). If there are too many such signs, it may indicate cheating or improper test-taking. Although the official statement says that only too many abnormal indicators will lead to a score being frozen for review, we don’t know how the official algorithm judges “minor abnormalities.” Will it be unscored or something else? No one knows.

However, we must protect ourselves. Therefore, if you solve questions too quickly, you should at least spend 30 seconds to 1 minute reviewing or re-checking your answers. This not only prevents careless mistakes but also avoids being flagged for abnormal activity.

If time runs out, should I guess all remaining questions, or focus on a few?

For reference: Quant has 21 questions, Verbal 23, and Data Insights 20, with 45 minutes per section. Total scores run from 205 to 805 in increments of 10 (always ending in 5), and each section is scaled 60–90.

Unanswered questions will be penalized, and the penalty is calculated based on the number of unanswered questions, deducting 30 * (number of unanswered questions / total number of questions) from your scaled score.

The figures below come from Dustin GMAT’s internal model of the scoring engine — they are a reverse-engineered estimate, not a GMAC-published number. Use them for pacing decisions, not as an official penalty table.

So I’ve calculated the scaled score deduction for each unanswered question:

• Q: Approximately 1.4 points deducted per question
• V: Approximately 1.3 points deducted per question
• DI: Approximately 1.5 points deducted per question

If you sit in the middle of the score range, where the percentile curve is steepest, each scaled score point lost can cost you close to 10 percentile points in ranking (an internal estimate, roughly -5% to -15%). Conversely, if you simply guess incorrectly rather than leaving questions unanswered in the later part of the test, the impact on your scaled score and percentile is small (an internal estimate of roughly 1-2 percentile points) — far cheaper than leaving them blank.

Make sure to complete all questions to avoid significant score loss from unanswered items.

For each question type, what is Dustin’s recommended time per question?

1. Small Questions

Includes: Critical Reasoning (CR), Problem Solving (PS), Data Sufficiency (DS)

Question stems are relatively short (usually no more than 100–110 words). They do not require multiple analyses or additional data references. Reading can be completed quickly by scanning the screen, and option structures are simple.

Recommended time: 1.5–2 minutes/question.

Due to the limited information in these question types, the logical reasoning or calculations required for answering are concentrated in the question stem, making them suitable for training quick judgment skills.

2. Medium Questions

Includes: Two-Part Analysis (TPA), Graphics Interpretation (GT)

Each question includes multiple options (usually presented in a two-column design or dropdown menu format). The amount of data is moderate, and comprehensive analysis across conditions or data is required. The process of reading and filtering questions takes more time than for small question types.

Recommended time: 2.5–3 minutes/question.

Reason: These question types involve multiple decisions, especially the two-column choices in TPA and chart analysis in GT, requiring time to read data and make comparative judgments.

3. Large Questions

Includes: Reading Comprehension (RC), Multi-Source Reasoning (MSR)

Involves analysis of multiple paragraphs or multiple pages of data (RC is a long passage; MSR involves 2–3 pages of paginated data). Each question set contains multiple questions, requiring an understanding of the overall text or data before answering. Requires extensive reading and locating information, which is time-consuming.

Recommended time:

RC (per set): 6–8 minutes (including 3–4 questions).

MSR (per set): Approximately 6 minutes per set (usually including 3 questions).

Reason: The nature of large question types requires test-takers to grasp multiple pieces of information at once, thus requiring more time to ensure accuracy in reading and locating information.

The pacing to maintain for the beginning, middle, and end sections of each subject

Ability-estimate movement is largest in the first 1/3 of each section, converges through the middle 1/3, and is small in the last 1/3. Note what this does and does not mean: GMAC confirmed in an official Q&A in February 2026 that there is no question-order weighting in scoring — every question contributes equally to your final ability estimate, and a question answered first is worth exactly as much as one answered last. With that in mind, here is the most comprehensive test-taking strategy:

Early Stage: First Third of Questions

Early questions matter for a structural reason, not a scoring-weight reason. Errors in roughly the first third steer the adaptive engine toward lower-difficulty items, which reduces the amount of information later questions can carry about your true ability. That makes the rest of the section a slower climb — it does not lock in a score ceiling, and it is fully recoverable within the section.

Answer carefully, avoid careless mistakes, and ensure the completeness of the solution logic and steps for each question. If you encounter simple question types (such as CR or DS), even if you intuitively know the answer, you should double-check or re-verify to avoid losing points due to carelessness.

If time permits, you can spend an extra 10–15 seconds confirming each question (e.g., setting the time for small question types to 2 to 2.5 minutes). This stage has a significant impact on your score, so prioritizing accuracy is worthwhile, even if it means slightly sacrificing time later.

Try multiple solution methods. If you are unsure, fall back to plugging in numbers or estimation rather than stalling. Do use the Question Review & Edit tool deliberately: you may change up to 3 answers per section, and any answer you revise is scored as your final answer and counts fully toward your ability estimate. What you should avoid is flagging questions casually — review time comes out of the same section clock, so bookmark only the one or two items you genuinely expect to resolve.

Middle Stage: Middle Third of Questions

By this stage the engine has narrowed in on your ability level, so each additional answer moves the estimate less than an early one did. That is a property of the estimate converging — not a discount on the questions themselves; every item still counts equally.

Answer at a normal pace, strictly adhering to the recommended time per question (1.5 to 2 minutes for small question types, 2.5 to 3 minutes for medium question types, 6 to 8 minutes per set for large question types), avoiding procrastination. For uncertain questions, you can make an educated guess and mark them for review if time permits later.

Maintain a steady mindset. Most questions at this stage are of moderate difficulty, so test-takers should remain calm and avoid letting excessive guessing or delayed answers affect subsequent progress.

Late Stage: Last Third of Questions

This stage has the highest time pressure, and the main goal should be to complete all questions. Unanswered questions will be severely penalized according to the penalty mechanism, affecting the total score by approximately 1.4 to 1.5 points per question, with an even more significant impact on the percentile.

Ensure all questions are completed. If you encounter a question you completely don’t understand, you should quickly guess an answer and move to the next question to avoid wasting too much time and leaving later questions unanswered.

Adjust time flexibly. If time is running out in the later section, you should reduce the time spent on each question and quickly select an answer. Even if you are unsure of the answer, you must ensure the question is “submitted” rather than left blank.

Avoid losing big by focusing on small things. Late questions move the estimate less than early ones, but they are worth the same, and burning the clock on one item to the point of leaving questions blank is the single most expensive mistake available to you.

How does the official test’s question order differ from mock tests?

(Note: There may still be changes in the future; refer to Dustin’s latest announcements.)

Verbal:

Since the launch of the new GMAT system in November 2023, and for the year leading up to October 2024, the question order for the GMAT Verbal section typically dispersed different question types. For example, CR questions would be inserted between RC passages, allowing test-takers to switch between question types and maintain a more relaxed pace.

However, starting from November 2024 (the one-year anniversary of the new system’s launch), there has been a noticeable change in the Verbal question arrangement, especially for RC questions, with a trend of multiple RC passages being arranged consecutively. Almost all test-takers report encountering 3-4 RC passages in a row. The intervals between RC passages are very short, or even non-existent, leading to a need to handle continuous and high-intensity reading during the exam.

To adapt to this new pattern, please make the following adjustments:

1. Strengthen RC endurance training. In your daily mock practice, complete 3-4 RC passages consecutively instead of taking a break or reviewing immediately after each passage. This simulates actual exam pressure and helps avoid fatigue or decreased attention during continuous RC reading.
2. In the current question design, CR questions may also appear consecutively (e.g., 4-6 questions at once). It is recommended to practice in sets of 5-6 questions to improve your ability to handle consecutive CR questions.

Data Insights:

In the past, the question pattern typically distributed different types of DI questions (DST, TPA, GT, MSR) evenly among the first 1-6 questions in the early part of the exam, allowing test-takers to quickly familiarize themselves with each type. MSR, in particular, usually appeared consistently between questions 4-7.

However, there have been significant changes in the DI question order recently:

1. A large number of DS questions are concentrated in the first few questions of the exam, for example, 3-4 out of the first five questions might be DS questions. This design requires test-takers to complete a large number of DS questions in the shortest possible time, demanding higher speed and accuracy.
2. The position of MSR question sets is no longer fixed. MSR question sets may sometimes be delayed until the middle or near the end of the exam, rather than consistently appearing between questions 4-7 as before. This requires test-takers to maintain a steady pace while switching between different question types, especially when dealing with MSR sets in the later part of the exam while facing test fatigue.

I would recommend the following:

1. Strengthen DS question training: DST is the DI question type with the least calculation and most straightforward structure among the four, but the concentration of many questions early on requires test-takers to process them quickly. Therefore, you need to compress the time spent on DS questions in daily practice, aiming to complete each question within 1.5 to 2 minutes. At the same time, you should improve DS accuracy, as these question types usually appear early and have a greater impact on the score.
2. Test-takers need to improve their MSR solving speed. If MSR question sets appear later in the exam, test-takers need sufficient time and energy to process multiple pages of data and multiple questions. During regular training, you can intentionally place MSR question sets at the end of mock exams to simulate the pressure of the late stage of the test.
3. If you encounter a large number of DS questions in the early part of the exam, you can save more time for subsequent TPA, GT, and MSR question types by compressing the time spent on DS questions. If MSR appears in the middle to late stages, ensure you still have enough time for data analysis and answering after completing the earlier questions.

How to self-analyze your score report?

Related posts:

Expert vs. Novice GMAT Thinking GMAT Math: Penalized Heavily for One Mistake? Confusing “Fact Inference” in RC/CR GMAT Critical Reasoning: Common Mistakes & Smart Strategies