GMAT Data Insights: The 5 Question Types, Timing and Scoring
What each GMAT Data Insights question type looks like, how multi-part answers are scored, the section's timing and score range, and a short strategy for each.
By the Relvora team · Updated · 6 min read
Data Insights is the GMAT section that tests whether you can use data, not just do math. It mixes five question types, and most of them look nothing like a standard multiple-choice question. Knowing each format before test day saves you real time.
In short: GMAT Data Insights has 20 questions in 45 minutes, with an on-screen calculator. Its five question types are Data Sufficiency, Multi-Source Reasoning, Table Analysis, Graphics Interpretation and Two-Part Analysis. Many questions have several parts, and you only get credit if every part is right. The section score ranges from 60 to 90.
What are the official facts about Data Insights?
These come from GMAC, which runs the GMAT.
| GMAT Data Insights | |
|---|---|
| Questions | 20 |
| Time | 45 minutes (2 minutes 15 seconds per question on average) |
| Question types | Data Sufficiency, Multi-Source Reasoning, Table Analysis, Graphics Interpretation, Two-Part Analysis |
| Calculator | On-screen calculator available |
| Multi-part questions | All parts must be right to get credit |
| Section score | 60–90 (total GMAT score 205–805) |
| Section order | You choose the order of the three sections |
| Changing answers | You can bookmark and review questions, and change up to 3 answers per section |
GMAC says your Data Insights score depends on how many questions you answer, whether each answer is right, and the difficulty of the questions you answer. GMAC doesn't publish how many questions of each type you'll see, so treat any exact mix you read online with care.
What does a Data Sufficiency question look like?
You get a question and two statements. You don't solve the question. You decide whether the statements give you enough information to answer it. Every Data Sufficiency question has the same five answer choices, so you can learn them once.
Is x > 0? (1) x³ > 0 (2) x² > 0
- Statement (1): a cube is positive only when the number is positive. So x > 0. That's a definite answer, so (1) is sufficient.
- Statement (2): x could be 1 (yes) or −1 (no). Not sufficient.
- Answer: (A), statement (1) alone is sufficient, but (2) alone is not.
Strategy: judge statement (1) alone, then (2) alone, then both together only if neither works alone. Try to break each statement with two cases that fit it but give different answers. There's a full walkthrough in the AD/BCE method.
What does a Table Analysis question look like?
You get a table you can sort by any column, plus a few statements. Each statement has two choices, such as true/false or yes/no.
Region Sales ($ thousands) Sales reps North 420 6 South 315 5 East 512 8 West 288 4 Central 365 5 True or false?
- The region with the most sales also has the most sales reps.
- The median sales figure is above $350 thousand.
- More than half of the regions have sales per rep above $70 thousand.
- True. East has the most sales (512) and the most reps (8).
- True. Sorted: 288, 315, 365, 420, 512. The median is 365.
- False. Sales per rep: North 70, South 63, East 64, West 72, Central 73. Only West and Central are above 70. That's 2 of 5, not more than half. North sits exactly at 70, which is the trap.
Strategy: sort the column a statement asks about before you read numbers off. Sorting finds a median or a maximum in one click. Watch the boundary words: "above," "at least," "more than half."
What does a Graphics Interpretation question look like?
You get a graph or chart and one or two statements with drop-down menus. You pick the option that makes each statement true.
A line chart shows monthly units sold: January 120, February 150, March 135, April 180.
From February to April, units sold rose by [16.7% / 20% / 30%]. The largest month-to-month change was from [Jan to Feb / Feb to Mar / Mar to Apr].
- The rise is 180 − 150 = 30 on a base of 150, so 30 ÷ 150 = 20%. The 16.7% option divides by the new value (30 ÷ 180). The 30% option reads the raw change as a percent.
- The changes are +30, −15 and +45, so the largest is March to April.
Strategy: read the axis labels and units first. Check whether the scale starts at zero, and whether values are in thousands or percents. Then use the calculator for any division you can't do at a glance.
What does a Two-Part Analysis question look like?
You get a short problem and a table of options with two answer columns. You choose one option for each column. The same option can be correct for both columns.
Two numbers have a sum of 20 and a difference of 6. Select the larger number and the smaller number.
Options: 6, 7, 10, 13, 14
- Larger + smaller = 20 and larger − smaller = 6. Adding the two equations gives 2 × larger = 26, so larger = 13 and smaller = 7.
- The traps each fit one condition only: 10 and 10 add to 20 but differ by 0; 14 and 6 add to 20 but differ by 8.
Strategy: treat each column as its own question, and check your pair against every condition at the end. See Two-Part and Multi-Source strategy for worked examples.
What does a Multi-Source Reasoning question look like?
You get information split across tabs. A tab can hold text, a table or a chart. Several questions use the same set of tabs. Some are standard multiple choice. Others give you a few statements to judge, each with two choices such as yes/no.
A small illustration: one tab is an email saying orders over $500 ship free and all other orders pay $25 shipping. Another tab is a table of orders for $480, $520 and $500. Only the $520 order ships free. The $500 order isn't over $500, so it pays $25. The answer depends on a rule in one tab and a number in another, which is what this type tests.
Strategy: read every tab once before the first question, and write down each rule in a few words. Then apply the rules to the exact rows each question names.
How should you pace the section?
Twenty questions in 45 minutes is 2 minutes 15 seconds each on average, but the types don't take equal time. A Multi-Source set asks you to read several tabs once, then answer several questions from them. A Data Sufficiency question has no table or chart to read at all.
- Use checkpoints, not a stopwatch per question. At 2:15 each, you should be about 10 questions in by 22:30.
- Don't sink time into one multi-part question. You need every part right to get credit, so a question you can't finish is worth moving on from.
- Use the calculator for division and percents, not for reading. The calculator can't fix a misread label or a wrong row.
- Save your answer changes. You can change up to 3 answers in the section, so bookmark questions you were unsure about as you go.
How does all-or-nothing scoring change your approach?
GMAC says that when a Data Insights question has several parts, all of them must be correct to get credit. That changes two habits:
- Check each part on its own. In Table Analysis and Multi-Source statements, one careless "true" sinks the whole question.
- Be careful at boundaries. A lot of part-level mistakes come from words like "more than," "at least" and "exactly." The North row above, at exactly 70, is a typical example.
Relvora's GMAT Data Insights practice and lessons cover all five question types. You can find both at GMAT Data Insights.
GMAT™ is a trademark of GMAC. Relvora is not affiliated with or endorsed by GMAC.
Sources
Frequently asked questions
How many questions are in GMAT Data Insights?
Data Insights has 20 questions in 45 minutes, which is 2 minutes 15 seconds per question on average. An on-screen calculator is available in this section.
Do you get partial credit on GMAT Data Insights?
No. GMAC says some Data Insights questions have multiple parts, and all parts must be answered correctly to get credit for the question.
What are the five Data Insights question types?
Data Sufficiency, Multi-Source Reasoning, Table Analysis, Graphics Interpretation and Two-Part Analysis.