GMAT Data Insights is the section where strong quant students lose points they shouldn’t. The arithmetic is rarely the problem; the traps sit in the reading, the logic, and the process.
Data Insights (DI) counts for one-third of your total score, the same weight as Quant and Verbal, and it’s the newest section, part of what changed with the GMAT Focus Edition. Here are six mistakes, an example of each, and the fix.
DI gives you 20 questions in 45 minutes, about 2 minutes 15 seconds each, and it’s the only section with an on-screen calculator; Quant has none. Your DI score runs 60 to 90, and the three sections count equally in the 205–805 total (how section scores add up).
Three rules shape your strategy:
GMAC’s exam content page defines five question types, and the current GMAT exam structure covers the timing.
| Question type | What you see | What you do |
|---|---|---|
| Data Sufficiency (DS) | A question, two statements, five fixed choices (A–E) | Judge whether the statements give enough information |
| Multi-Source Reasoning (MSR) | Multiple tabs, several questions per set | Combine the relevant pieces across tabs |
| Table Analysis (TA) | A sortable, spreadsheet-style table | Mark several statements yes/no or true/false |
| Graphics Interpretation (GI) | A chart or diagram plus short text | Complete sentences using drop-down menus |
| Two-Part Analysis (TPA) | One answer list, two response columns | Pick one answer per column; they may be linked |
What it looks like. You grind to a number when you only need to know whether one answer exists, or call a statement insufficient because it gives a definite no.
A hypothetical example. “Is x greater than 5? (1) x < 3.” Many test-takers see that (1) doesn’t say what x is and call it insufficient, but a firm “no” is an answer, so (1) is sufficient: if (2) alone doesn’t answer it, pick (A).
The fix.
What it looks like. You spot one trick and stop checking, assume what the question never granted (integers, positive numbers, figures drawn to scale), or miss a constraint the context does impose. Hard DS questions get their difficulty from the setup, not the math: good test-takers find the first trap and stop, the best keep looking.
A hypothetical example. A warehouse packs identical boxes, each holding the same whole number of units. How many units per box? (1) Three boxes hold more than 20 units. (2) Four boxes hold fewer than 32 units. Alone, each gives only a range: more than 6.67, or fewer than 8. Together, and because units are whole, the count is exactly 7, answer (C). Treat the ranges as continuous and you pick (E).
The fix. List every constraint in the stem first: integer, positive, “at least,” “distinct.” Then retest your answer with a negative, a fraction, zero, and a large number to see whether it changes. GMAC also notes that DS figures are not necessarily drawn to scale, so don’t trust the picture. Studying official Data Sufficiency questions explained trains you to spot the second trap.
What it looks like. You spend 3–4 minutes absorbing tabs of emails, memos, and tables before reading a question, then reread when it asks about one number.
A hypothetical example. Three tabs: a manager’s email setting a goal of cutting shipping costs 10%, a carrier rate table by region and weight, and a policy update adding a surcharge above a weight threshold. A yes/no question asks whether switching one region to a new carrier meets the goal, which takes only the target, two rate rows, and the surcharge rule buried in the third tab. Reading every word wastes time; skimming too fast misses the surcharge.
The fix. Spend 30–45 seconds mapping the set: each tab’s title and first sentence, and which variables each table or chart holds, not the values. Then read the question and pull only what it needs, watching for rules and exceptions (“applies only when,” “excluding,” footnotes). The map pays off more than once, since the same set returns for several questions. GMAC’s Data Insights prep strategies note that everything you need is on the screen, so outside knowledge shouldn’t sway an answer.
What it looks like. You answer “yes” because a statement is plausible, when the question asks whether the data shows it to be true.
A hypothetical example. A sortable table lists 12 sales offices with columns for revenue rank, headcount, and year-over-year growth (%). Statement: “The office with the highest revenue also has the highest revenue per employee.” The table gives revenue rank, not dollars, so revenue per employee can’t be calculated; the statement can’t be shown true, so the answer is “no,” however likely it seems.
The fix. Sort by the column each statement is about, which is what GMAC recommends the sort feature for. Then ask what data would prove the statement, and check whether the table has it. Watch for ranks vs. values, percentages vs. counts, and items the table doesn’t cover. It’s Critical Reasoning’s “must be true” logic applied to a spreadsheet. Every statement in a TA question must be right, so one careless “yes” zeroes out the question.
What it looks like. You read an axis in thousands when the text says millions, assume an axis starts at zero, read a stacked bar as separate bars, or confuse percent change with percentage-point change.
A hypothetical example. A line chart shows market share rising from 20% in 2023 to 25% in 2025. The sentence to complete: “From 2023 to 2025, market share grew by ___ relative to its 2023 level,” with options of 5%, 20%, 25%, and 125%. Share rose 5 percentage points, but the sentence asks for relative change: 5 ÷ 20 = 25%. “5%” is the trap; the wording decides.
The fix. Read the axis labels, units, scale, legend, and footnotes first, and read all the drop-down options before answering: they tell you what kind of answer the task wants. Estimating first helps, since the options are often spread far enough apart to settle the question once you eliminate the extremes, as these worked Data Insights examples show.
What it looks like. You solve column 1, then start column 2 from scratch, or refuse to pick the same answer twice because it “feels wrong.”
A hypothetical example. A business plans a season of events: revenue must be exactly 12,000,everyeventsells200tickets,thepricemustbeatleast12, and the venue is free only for 5 or more events. Column 1 is the price, column 2 the number of events, and both draw on one list: 3, 4, 5, 6, 10, 12, 15. Price × 200 × events = 12,000,soprice×events=60:only10 and 6, 12and5,and15 and 4 fit, and only $12 with 5 events clears both conditions. GMAC also warns that the same answer choice can be correct for both columns, and TPA isn’t always math; some versions ask you to reason about an argument.
The fix. Decide whether the tasks are independent or linked. If linked, test pairs against the constraint instead of solving each column alone, and don’t rule out an option in column 2 because you used it in column 1.
DI drills fit naturally into the middle weeks of a 10-week GMAT study schedule.
Pair this with a three-step GMAT prep method: refresh the basics, learn and apply techniques, then practice under real conditions.
The section rewards process more than math, so fixing these six habits is usually the fastest DI score gain available.
