GMAT Data Insights

What Data Insights tests

Data Insights (DI) asks you to reason with information presented in different forms. You may see a table, chart, short passage, set of tabs or paired answer fields. The section contains 20 questions in 45 minutes. A calculator is available on screen, but the main task is often deciding which information matters and what the question means.

The five question types are Data Sufficiency, Multi-Source Reasoning, Table Analysis, Graphics Interpretation and Two-Part Analysis. They reward different moves. Sometimes you calculate a value. Sometimes you test whether information is enough, combine two sources or choose a pair of related answers. Recognizing the task before working is often the quickest way to avoid unnecessary arithmetic.

A repeatable first read

Begin by reading the question stem and identifying the requested result. Write down the unit, time period, population and condition. Then inspect the data source that can answer it. A chart might show values in thousands, a table might include only surveyed customers, and a two-part question might ask for both a quantity and a reason. These details determine what counts as a correct answer.

Use the calculator only after the setup is clear. It is useful for cumbersome division or multiplication, but keying numbers before deciding whether to add, average or compare can make an error feel precise. Estimate the expected size of an answer first. If the question compares two quantities, note which should be larger before calculating so you can catch an inverted subtraction.

Data Sufficiency

Data Sufficiency asks whether statements provide enough information to answer a question. You do not need to calculate the final value when a statement proves that one unique answer exists. Test the question, then each statement alone, then both together if needed. The reasoning is about determinacy, not about finding a convenient example that happens to work.

For instance, if the question asks for the value of x and statement one says x + 4 = 11, that statement is sufficient because it determines x. If statement two says x is positive, it is not sufficient by itself because many positive values remain possible. Together they are sufficient because statement one already determines x. Avoid assuming that an unmentioned number must be an integer or positive.

For a yes-or-no question, sufficiency means every value allowed by the statement leads to the same answer. If one permitted case yields yes and another yields no, the data are insufficient. Try boundary values when a condition uses inequalities. Testing examples can expose insufficiency, but one convenient example cannot prove sufficiency unless the logic shows all allowed values have the same outcome.

Multi-Source Reasoning

Multi-Source Reasoning presents information across multiple tabs, panels or documents. Questions may require locating evidence in one source and comparing it with another. Do not read every sentence with equal intensity before you know what is asked. Start with the question, identify likely sources, and record the fact or condition from each one that matters.

A common trap is to rely on one tab when the prompt requires a comparison. Suppose one note gives a project's planned budget and another records its actual spend. A question asking whether the project stayed within budget requires both numbers. Another trap is using a headline summary that omits an exception stated in a footnote or policy panel. Check definitions and conditions when a value appears inconsistent.

If a question has several linked parts, keep the facts in a compact scratch table: claim, source, value and unit. This reduces repeated searches across tabs. Do not carry a value from one project or period into another. Track labels carefully, especially when multiple rows or sources use similar names.

Table Analysis

Table Analysis asks you to evaluate data in a sortable table, often against a stated condition. First identify the requested row or category and the criterion. Sort by the relevant column if that helps, but confirm that the sort direction and units match the question. A table may have many entries, so a systematic filter is usually faster than reading every row sequentially.

Suppose a table shows contribution in thousands of dollars. A value of 145 represents $145,000. If the prompt asks which customers exceeded $150,000, a value of 145 does not qualify. If the table gives growth as a percentage, compare percentage growth rather than absolute dollar change unless the question asks for dollars. Sorting one column does not change what another column measures.

Graphics Interpretation

Graphics Interpretation uses a chart or graphic and asks you to complete a statement, often by selecting values for blanks. Read the title, axes, scale, legend and footnotes before selecting a point. Check whether the x-axis is linear or uses categories and whether an axis begins at zero. A visual slope can appear large because the scale is narrow even when the numerical change is modest.

If a line rises from 40 to 52 over four months, the increase is 12 units, or 30 percent of the starting value. A distractor may report 12 percent, confusing absolute change and percent change. Another may report 23 percent by dividing by the final value. Percent change uses the original value as its denominator: (52 - 40) / 40 = 0.30.

Two-Part Analysis

Two-Part Analysis gives a shared prompt and asks for two related answers, such as two values, a value and a condition or a pair of choices. Read the full prompt before solving either part. The correct pair must satisfy both requirements at once. A choice can be individually plausible and still fail because it conflicts with the other selected value.

For a rate problem, one column might ask for the number of units produced and another for the hours required. If the production target doubles, hours may double only when the rate stays constant. Verify the relationship in the prompt rather than assuming it. When possible, solve one part and use it to test the other.

Original worked questions

Read a table's unit

A table reports monthly subscription revenue in thousands: North 84, South 96 and West 110. How much greater was West revenue than North revenue in dollars? The difference is 110 - 84 = 26 thousand, or $26,000. The answer is not $26, and it is not $194,000, which adds the two regional values.

The first step is to identify what is being compared. The second is to preserve the table's unit. If the question had asked for the percent by which West exceeded North, the calculation would be 26 / 84, about 31 percent. The absolute difference and percentage difference answer different questions.

Use two sources together

A planning sheet gives a project budget of $240,000. A status sheet says the project has spent $178,000 and has committed $49,000 more. Is the project currently under budget if every commitment is expected to be paid? Total projected cost is $178,000 + $49,000 = $227,000, so it is $13,000 below budget. Looking only at spent funds would overstate the remaining room.

A tempting wrong calculation subtracts $178,000 from the budget and reports $62,000 remaining. That is cash not yet spent, not the amount remaining after committed costs. Another wrong answer adds the budget to the actual spend. Keep planned, spent and committed quantities separate.

Test sufficiency without solving every value

A question asks whether positive integers a and b have the same parity. Statement one says a + b is even. This is sufficient: the sum of two integers is even exactly when both have the same parity. Statement two says a is odd. Alone it is not sufficient because b could be odd or even. Together, the statements are sufficient, although statement one already answers the question.

The point is not to find the actual values of a and b. It is to prove whether every allowed case produces the same answer. Testing a = 1, b = 1 is helpful for statement two only if you also test another allowed case such as a = 1, b = 2 to show the result changes.

Timing and calculator habits

Twenty questions in 45 minutes gives an average of two minutes and 15 seconds per question, but difficulty and reading load vary. Use this average as a pacing signal, not a fixed limit. A short chart question may take less than a minute; a multi-source problem can take longer. If one item absorbs several minutes without progress, make the best selection you can and move on.

Practice with the same calculator interface used by the test. Enter one operation at a time, check decimal placement and keep units outside the calculation. For a ratio such as 18:24, simplify to 3:4 before dividing if the question asks for the relationship. Avoid using a calculator for mental arithmetic that is quick and certain, but use it when lengthy arithmetic distracts from reasoning.

At the end of practice, classify each missed DI question as a data-reading, setup, sufficiency, arithmetic or time-management error. That classification points to a remedy. More calculator drills will not fix a row-selection mistake. Faster chart reading will not help if the question asks for a percentage and you calculate an absolute difference.

A focused practice sequence

Start with one question type at a time so you can learn its decision process. Do a small untimed set and explain each answer. Then complete mixed sets in which the type is not announced. The mixed set tests whether you can identify the method. Finish with full 45-minute Data Insights sections and review questions that took too long, not only those answered incorrectly.

For each practice question, explain why the chosen answer works and why the strongest distractor fails. In Table Analysis, name the wrong unit or criterion. In Multi-Source Reasoning, identify the missing panel. In Data Sufficiency, state which possible cases remain open. A precise explanation makes it easier to detect the same trap when the surface details change.

The section is part of a three-section adaptive exam, and its score contributes equally with Quant and Verbal to the total. Data Insights results should therefore be reviewed as one component of your overall profile, not as a separate admissions credential. Use the report's feedback and timed practice to decide whether it deserves more study time.