GMAT Data Insights: The Complete Guide

Data Insights (DI) tests how you use data to make decisions: judging whether information is enough, pulling facts out of tables and charts, combining sources, and weighing trade-offs. It's 20 questions in 45 minutes, and it's the only GMAT section with a calculator.

DI isn't a pure math section. Some questions are arithmetic in disguise; others are logic or careful reading. It took Data Sufficiency from the old Quant section and the four formats from the old Integrated Reasoning section, and unlike Integrated Reasoning, it counts toward your 205–805 total.

GMAC doesn't publish a fixed number of questions per type, so prepare for all five.

• Multi-part questions are all-or-nothing. A Table Analysis question with three statements earns credit only if all three are right. One careless row costs the whole question.
• The calculator is basic and on-screen. Use it for ugly division and percent checks, not for 12 × 5. Clicking takes longer than you think.
• It's adaptive, like the other sections, and Question Review & Edit applies: you can change up to three answers in the section.
• The average pace is 2:15 per question, but it's uneven. A Data Sufficiency question can take a minute; a Multi-Source Reasoning set can eat six or seven minutes for its three questions.

You get a question and two statements. Your job is to decide whether the statements give enough information to answer it, not to find the answer. The five answer choices never change:

1. Statement (1) alone is sufficient, but (2) alone is not.
2. Statement (2) alone is sufficient, but (1) alone is not.
3. Both together are sufficient, but neither alone is.
4. Each statement alone is sufficient.
5. The statements together are not sufficient.

Worked example

A bakery sold 50 pies yesterday, each either apple or cherry. How many cherry pies did it sell?
(1) Revenue from apple pies was $240.
(2) Apple pies sold for $12 each and cherry pies for $15 each.

(1) alone: no prices, so $240 could be 10 pies or 40. Not sufficient. (2) alone: prices with no revenue figure. Any split of 50 pies is possible. Not sufficient. Together: 240 ÷ 12 = 20 apple pies, so 50 − 20 = 30 cherry. Sufficient. The answer is "both together."

You get two or three tabs (an email, a policy, a table) and usually three questions about them. The answer often depends on a detail sitting in a different tab than the one the question seems to be about.

Worked example

Tab 1, Travel policy: Meals are reimbursed up to $60 per day. Any single meal over $25 needs a receipt.
Tab 2, Expense report: Day 1: $12, $18, $34 (receipt for the $34). Day 2: $22, $31 (receipt for the $31). Day 3: $9, $24, $26 (no receipts).

For each day, would the full amount claimed be reimbursed? Yes or No.

Day 1: No. The receipt is fine, but $64 is over the $60 cap. Day 2: Yes. $53 is under the cap, and the only meal over $25 has a receipt. Day 3: No. $59 is under the cap, but the $26 dinner has no receipt. Each "No" fails for a different reason. That's typical: the question tests whether you checked every rule against every row.

You get a data table (sortable by column on the real test) and three statements to mark true/false or yes/no. Most statements need one sort or one quick calculation.

Worked example

Return rates: A 5%, B 7%, C 3%, D 8%.

• The product with the most units sold has the lowest return rate. True: C sold the most and returns 3%.
• The median return rate is above 6%. False: sorted rates are 3, 5, 7, 8, so the median is (5 + 7) ÷ 2 = 6%, which isn't above 6%.
• Product B had more returns than Product D. True: 56 vs. 48, even though D has the higher rate.

You get a chart (bar, line, scatter, pie, or something custom) and one or two statements with drop-down blanks. Read the axes, units, and scale before you read the data.

Worked example

From January to October, active users grew by about [31% / 38% / 62% / 81%], and the largest increase between consecutive data points came from [January to April / April to July / July to October].

Blank 1: 62%. Growth is 81 − 50 = 31 thousand, and 31 ÷ 50 = 62%. Picking 31% is the classic slip: that's the change in thousands, not the percent. Blank 2: July to October. That's +21 thousand, compared with +15 thousand from January to April (April to July is a drop).

You get a scenario and a grid with two columns (for example "Part A" and "Part B," or "Strengthens" and "Weakens"). You pick exactly one row in each column. It can be quantitative, verbal, or pure logic.

Worked example

A café orders exactly 4.5 kg of coffee in 9 bags. Each bag holds either 250 g or 1 kg. Select the number of 250 g bags and the number of 1 kg bags. Options: 2, 3, 4, 5, 6, 7.

Let s = small bags and k = large bags: s + k = 9 and 0.25s + k = 4.5. Subtract: 0.75s = 4.5, so s = 6 and k = 3. Check: 1.5 kg + 3 kg = 4.5 kg.

The trap is swapping the columns: 3 small and 6 large gives 6.75 kg. Always plug both picks back into every condition. Also, the two columns can sometimes share the same row, so don't assume they must differ.

• Solving Data Sufficiency instead of judging it. Once you know a statement pins down one value, stop.
• Forgetting the negative root or the fraction. In DS, test numbers like −2, 0, and ½ before calling a statement sufficient.
• Percent vs. absolute change in graphs, and rate vs. count in tables.
• Missing a footnote in Multi-Source Reasoning: minimums, fees, and exceptions are where the questions hide.
• Sinking five minutes into one set. If a question is eating your clock, make your best pick, bookmark it, and move on. Unanswered questions at the end cost more than one wrong guess.

1. Week 1: Data Sufficiency. Learn the five answer choices cold, then do 15–20 DS questions a day. Review every miss for the case you didn't test.
2. Week 2: Tables, graphs, and Two-Part. Alternate Table Analysis, Graphics Interpretation, and Two-Part Analysis sets. Practice with the on-screen calculator so it doesn't slow you down on test day.
3. Week 3: Multi-Source Reasoning and mixed timed sets. Do full 20-question sections in 45 minutes, including the DI section of our free GMAT practice test.

For official questions, use the Data Insights Review book or GMAC's online practice (see our prep book picks). For the full format and scoring, see the GMAT overview.