GMAT Data Insights: the 5 question types and how to approach each

Updated 23 September 2026 · 10 min read

Data Insights is the section most test-takers have practised least. It did not exist in its current form before the GMAT Focus Edition, it counts as much toward your Total Score as Quant and Verbal, and it mixes five question formats that each reward a different habit. The upside is that the section is very learnable: once you know what each format is really asking, most of the difficulty is in reading carefully and managing time, not in hard mathematics.

This guide covers what the section contains, a realistic pace for it, and a method and the common traps for each of the five question types.

What the Data Insights section is

According to GMAC, the Data Insights section has 20 questions in 45 minutes, and an on-screen calculator is available (it is not available in the Quantitative Reasoning section). It contains five question types: Data Sufficiency, Multi-Source Reasoning, Table Analysis, Graphics Interpretation and Two-Part Analysis. Your Data Insights score is reported on the same 60–90 scale as the other two sections, and GMAC states that the three section scores contribute equally to the Total Score.

Three features shape how you should approach it:

• Many questions have several parts. A Table Analysis question may ask you to judge three statements; a Two-Part Analysis question needs two selections. Prep companies consistently report that there is no partial credit: you need every part right to get the question right.
• The section is adaptive, like Quant and Verbal, so the difficulty you see depends on how you are doing.
• Unanswered questions hurt. GMAC lists the number of questions answered as one of the scoring factors, alongside correctness and difficulty. Finishing the section matters.

Here is what each format looks like at a glance.

A realistic pace for 20 questions

Forty-five minutes for 20 questions is an average of 2 minutes 15 seconds per question. That average hides a lot of variation. A Data Sufficiency question you recognise can take 90 seconds. A Multi-Source Reasoning set asks you to read two or three tabs before you answer anything, so the first question of the set takes much longer than the ones that follow.

A workable plan:

• Check your pace at the 5, 10 and 15 question marks. You want roughly 11, 22 and 34 minutes used. If you are behind, the next hard question is the one to guess on.
• Put a ceiling on any single question. Around three minutes is a sensible limit. Past that, you are usually rereading rather than solving.
• Guess strategically, not randomly. If a multi-part question is going badly, answer every part with your best reasoning and move on. A blank part guarantees a miss; a reasoned guess does not.
• Use Question Review & Edit sparingly. On the GMAT Focus Edition you can bookmark questions and, once you have answered every question in the section, return to them and change up to three answers, within the section's time. Leave time for it only if you have specific bookmarked questions worth a second look.

Data Sufficiency

Data Sufficiency moved from the old Quant section into Data Insights. You get a question and two statements, and you decide whether the statements give enough information to answer it. You never need the actual answer.

The five answer choices are always the same:

• A: Statement (1) alone is sufficient, but statement (2) alone is not.
• B: Statement (2) alone is sufficient, but statement (1) alone is not.
• C: Both statements together are sufficient, but neither alone is.
• D: Each statement alone is sufficient.
• E: The statements together are still not sufficient.

Method. First decide what kind of question it is. A value question ("What is x?") needs exactly one value. A yes/no question ("Is x > 0?") needs a definite answer, and "definitely no" counts as sufficient. Then judge statement (1) alone, then statement (2) alone, deliberately forgetting statement (1). Combine them only if neither works alone.

Example. If n is an integer, is n even? (1) n² + n is even. (2) 3n is even. Statement (1) looks useful, but n² + n = n(n + 1) is the product of two consecutive integers, which is always even. It tells you nothing about n, so it is insufficient. Statement (2): if 3n is even, n must be even, because 3 is odd. Sufficient. The answer is B.

Traps. Carrying information from one statement into the other; assuming a variable is a positive integer when the question never says so; and stopping once you find an answer instead of checking that it is the only answer. Test cases on purpose: a negative number, a fraction, zero, and the smallest allowed values. One counter-example makes a statement insufficient.

For worked examples, see our free Data Sufficiency practice questions.

Two-Part Analysis

Two-Part Analysis gives you a short scenario and a table with two answer columns and one shared list of options. You pick one option for each column. The two selections are often linked: a price and a quantity that must satisfy one equation, or a strengthener and a weakener for the same argument.

Method. Read the column headings before the scenario, because they tell you exactly what to find. Then write down the constraint that links the two answers. For quantitative versions, it is usually an equation, and the answer choices let you test values rather than solve in general.

Example. A shop sells notebooks for $4 and pens for $1.50. Maya spends exactly $20 and buys at least one of each. Select the number of notebooks and the number of pens she could have bought. Double everything to clear the decimal: 8n + 3p = 40. Since 40 − 8n must be divisible by 3, n can only be 2 or 5, and n = 5 leaves no money for pens. So Maya bought 2 notebooks and 8 pens. The constraint "at least one of each" is what makes the answer unique, and it is exactly the kind of detail that is easy to skip.

Traps. Answering each column in isolation when the question makes them depend on each other; and forgetting that both columns can take the same option unless the question says otherwise. Verbal versions often ask for two things that sound alike ("an assumption of the argument" versus "a claim the argument rejects"), so reread the headings before you commit. Worked examples and a four-step method: Two-Part Analysis guide.

Table Analysis

Table Analysis shows a table of data that you can sort by its columns. Several statements follow, and you judge each one with a two-way choice such as true/false or yes/no.

Method. Before reading the data, skim the column headings and any note above the table (units, what a row represents). Then take each statement one at a time and decide which column answers it. Sorting is the main tool: a statement about the "top five" or "the median" becomes easy once you sort by the right column.

Traps.

• Counting errors with the median. With an even number of rows, the median is the average of the two middle values.
• Percent versus percentage points. A share rising from 10% to 15% is a 5-point increase but a 50% increase.
• "Must be" versus "could be". Some statements ask what the data prove. If the table only shows a total, it cannot prove anything about an individual row.
• Sorting and forgetting. A sort changes the whole view. After sorting for one statement, re-sort for the next rather than assume the order still holds.

For a full worked table with three statements, see the Table Analysis guide.

Graphics Interpretation

Graphics Interpretation gives you one visual (a bar chart, scatter plot, line graph, pie chart, flow diagram or similar) and two statements with drop-down menus to complete.

Method. Spend your first 20 seconds on the graphic itself, not the statements: the title, both axis labels, the units, and the scale. Check whether an axis starts at zero, whether it is logarithmic, and whether a legend changes the meaning of colours or shapes. Then read each statement and look at its drop-down options before calculating. The options are often far enough apart that a rough reading is enough.

Traps. Reading the wrong series in a multi-line chart; misreading the scale by one gridline; confusing a change with a rate of change; and treating a correlation in a scatter plot as proof of cause. On scatter plots, questions often ask about the line of best fit or about the number of points above or below a line. Count carefully and keep track of points that sit exactly on the line. The Graphics Interpretation guide works through a chart with drop-down answers.

Multi-Source Reasoning

Multi-Source Reasoning presents information on two or three tabs, such as an email, a table and a policy memo, followed by a set of questions about the same sources. Some questions are ordinary multiple choice; others give several statements to judge as yes/no, true/false or similar.

Method. Do not try to memorise the tabs. Skim each tab once to learn what kind of information it holds and how the tabs relate: one may set rules, another may give data those rules apply to. Then go to the question and return to the specific tab it needs. Keep brief scratch notes such as "Tab 1: pricing rules; Tab 2: Q3 orders; Tab 3: exceptions" so you know where to look.

Traps. Missing a condition or exception on one tab that changes the answer on another; bringing in outside knowledge the sources do not support; and treating "can be inferred" as "is plausible". Because the sources stay the same for the whole set, the investment in reading pays off over several questions, so do not rush the first read and then pay for it on every question. The Multi-Source Reasoning guide has a full three-tab set with solutions.

Using the calculator well

The on-screen calculator helps with ugly arithmetic, such as percentages of large numbers or averages of many values. It does not help with deciding what to calculate, and it is slow to use for simple steps. Estimate first: if the answer choices differ by a lot, a rough figure is enough. Use the calculator for the final step only when the options are close together. On Data Sufficiency, you rarely need it at all, because you are not solving to the end.

How to practise Data Insights

• Learn the formats before you time yourself. Untimed work on each type teaches you what the questions look for. Then move to timed mixed sets.
• Use official material for the formats that depend on the interface. Table sorting, tabbed sources and drop-downs feel different on screen. GMAC's free Official Starter Kit includes two full-length practice exams.
• Keep an error log by question type. Record why you missed each question: misread the graph, a statement-carrying error in DS, ran out of time. Patterns appear quickly. Our guide to keeping a GMAT error log explains how.
• Practise Quant alongside Data Insights. Data Sufficiency and Two-Part Analysis use the same arithmetic, algebra and number properties as Problem Solving. Topic sets such as number properties and statistics help both sections.

If you are not sure how much Data Insights matters for your target, the GMAT Focus score calculator shows how a change in one section score moves the estimated total and the percentile.