GMAT Data Insights: the complete section guide for 2026.

Data Insights (DI) is the section most students under-train. Quant and Verbal have decades of prep material. DI was built for the Focus Edition that launched in late 2023, and most existing prep is either thin, copied from the old Integrated Reasoning playbook, or outright wrong. If you're in the 605 to 705 range, DI is almost always one of the two highest-leverage sections to fix.

I went from a 73 to a 87 on Data Insights over four months — a larger raw jump than I made in either Quant or Verbal. None of it came from new content. It came from understanding the format of each question type well enough to stop wasting time on the wrong approach. This guide is the version of that playbook I wish I'd had at the start.

DI doesn't test five different topics. It tests one skill — extracting the answer from messy information — wearing five different costumes.

What Data Insights actually is

Data Insights is the third section of the GMAT Focus Edition. It is scored on the same 60–90 scale as Quant and Verbal, and it contributes equally to your Total Score. Treating it as a side section because it isn't “real Quant” or “real Verbal” is one of the more expensive mistakes a candidate can make.

The format:

• 20 questions in 45 minutes — 2 minutes 15 seconds per question on average.
• One section, no breaks. The clock runs straight through.
• Five question types, mixed in unpredictable order:
  • Data Sufficiency
  • Multi-Source Reasoning
  • Table Analysis
  • Graphics Interpretation
  • Two-Part Analysis
• An on-screen calculator is provided. Use it for specific computations, not as a substitute for set-up.
• Multi-part questions count as one item. A Multi-Source Reasoning set may include three questions but each one is scored independently.

The single biggest practical change from the old GMAT is that Data Sufficiency now lives in DI, not in Quant. Roughly a quarter of your DI section will look identical to the DS you'd expect from old Quant prep. The other three quarters will not.

The five question types, in detail

1. Data Sufficiency (DS)

DS is the question type most students already know. You get a question stem, two numbered statements, and the same five answer choices every time:

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

Format mistakes that cost points. The classic DS mistake — and it never goes away no matter how much prep you do — is to start solving the question instead of asking whether each statement determines the answer. You are not being asked for the value. You are being asked whether the value is pinned down.

Approach.Read the question stem. Translate it into a single concrete question you could answer with the right information. Look at each statement in isolation and ask: “Does this give me enough to answer that exact question with certainty?” Only combine after you've evaluated each alone. The order matters — combining first is the surest way to mis-grade statement (1) or (2).

Timing target. 1 minute 45 seconds to 2 minutes. DS questions are denser than the type-average suggests; you want to bank time here so you have it for MSR sets later.

2. Multi-Source Reasoning (MSR)

MSR presents two or three tabs at the top of the screen — typically an email, a spreadsheet, a memo — and then asks two to three independent questions about that information. Some questions are standard multiple-choice; others are a row of yes/no judgements.

The trap.The trap on MSR is reading order. If you read all three tabs in detail before you see any of the questions, you will spend two minutes on context you don't need and run out of time on the question you do.

Approach.Skim the tabs in 30 seconds — what is each one and roughly what does it contain. Then go to the first question. Each question will tell you which tab it actually depends on. Read that tab's relevant lines carefully and answer. Treat each question as a fresh search task; do not try to hold the entire information set in working memory.

Timing target. 5 to 6 minutes for an entire MSR set, regardless of whether it has two or three questions. Set a mental budget the moment you see the tabs.

3. Table Analysis (TA)

Table Analysis gives you a single sortable spreadsheet plus a short blurb of context. The questions are usually three statements you must judge as true or false (or “would help” vs. “would not help”) based on the table. To get the question right, all three judgements have to be correct.

The trap. All-or-nothing scoring is what makes TA brutal. Two out of three is wrong. People skim the table, get the obvious one right, and miss a subtle one — losing the entire item.

Approach. Resort the table for each statement individually. The on-screen sort is the entire point of the interface; you are meant to use it. For each statement, identify the column that matters, sort by it, and read off the answer before judging. Do not try to evaluate three statements off a single sort; you will mis-read at least one.

Timing target.2 minutes 30 seconds. The interface is slow even when you know what you're doing.

4. Graphics Interpretation (GI)

GI shows a chart — scatter, bar, line, sometimes more exotic — with two fill-in-the-blank statements underneath. Each blank is a dropdown of values. To get the question right, both dropdowns must be correct.

The trap. GI rewards students who read the axis labels and units carefully and punishes students who eyeball. Charts on the GMAT are drawn precisely; every value you need can be read off the chart if you actually look at it.

Approach.Spend the first 20 seconds on the chart itself, not the question. What are the axes? What are the units? What does each point or bar represent? Only then read the statements. Most GI dropdowns can be answered by translating the statement into “at value X on this axis, what is Y?” and reading directly off the chart.

Timing target. 2 minutes. GI is faster than it looks once the chart is mapped.

5. Two-Part Analysis (TPA)

TPA presents a written prompt — sometimes quantitative, sometimes verbal — and asks you to choose two values from the same column list, one for each of two related slots. For example: “Choose the value of X that satisfies both equations and the value of Y that satisfies neither.”

The trap. The two slots are usually interdependent in a way that hides which one to attack first. People try to solve them in order, hit a dead end on the second, and start the first over.

Approach. Read both slot prompts before you do any work. Identify which slot is more constrained — the one with the narrower set of valid candidates — and solve that one first. Then use the result to filter candidates for the second slot. Quantitative TPA almost always rewards plugging values from the list rather than solving algebraically.

Timing target. 2 minutes 15 seconds. TPA is the most variable — easy ones are sub-2-minute, brutal ones approach 3.

Timing strategy across the section

The 45-minute clock and 20 questions hide a complication: a three-question MSR set takes about three times as long as a single DS question, but counts as three items. If you treat “average time per question” as a hard rule you will run out of time on the MSR set or skip easy DS at the end.

The rule I used:

• DS:1:45–2:00.
• MSR set (2–3 questions):5:00–6:00 total.
• Table Analysis: 2:30.
• Graphics Interpretation: 2:00.
• Two-Part Analysis: 2:15.

Add those up across a typical mix and you land at roughly 43 minutes — leaving a 2-minute buffer for the one or two questions that go long. If a single DS or TPA question crosses 3 minutes, guess and move on. Coming back to it would cost you points elsewhere.

Every minute you spend over your per-type budget is a minute you're stealing from a question you would have got right.

The five mistakes that cost most students DI points

Mistake 1 — Treating DI like Quant

DI rewards information extraction and process discipline. Quant rewards algebraic fluency. Students who try to solve every TPA algebraically, or compute exact values on Table Analysis when a comparison would do, lose minutes per question and rarely catch up. The on-screen calculator and sort interface exist for a reason.

Mistake 2 — Reading MSR tabs front-to-back

The tabs are reference material, not a passage. Read them like a spreadsheet you're looking up a value in, not like an essay you're studying.

Mistake 3 — Solving Data Sufficiency

The DS prompt is asking whether the answer is determined, not what it is. You can spend ten seconds checking whether a value is pinned down. You can spend ninety seconds computing the value itself. The first is the question. The second is a way to fail at being asked the first.

Mistake 4 — Eyeballing Graphics Interpretation

GMAC draws GI charts to scale and on labelled grids. Every answer is in the chart, in numbers, if you bother to read the axes. The wrong answers in the dropdown are usually values that look right if you eyeball but don't survive a careful read.

Mistake 5 — Skipping the section in study planning

DI is the section most students study last and least, so it's also the section with the highest marginal return on a focused week. If you're torn between another month of Quant drilling and a serious week on DI, the second move almost always pays better — both because the headroom is larger and because DI is scored on the same scale as the other two sections.

How to practice Data Insights

The general study loop I recommend is the same one I used for every section, adapted slightly for DI's format-heaviness:

1. Drill one question type at a time first. The instinct to do a mixed set immediately is a mistake. You need the format reps before you need the variety. Two weeks of one-type-per-day drills, then mix.
2. Time individual questions from day one. Untimed practice on DI hides the real problem, which is almost always pacing rather than content.
3. Log every miss with a trap tag.The same five traps repeat across hundreds of questions. Tag them — you cannot fix a pattern you cannot name.
4. Mock the section as a whole, weekly. Time your full 45-minute DI sets early. The full-section pacing is its own skill, separate from per-question pacing.
5. Review with the chart, table, or tabs in front of you. When you review a missed GI question, re-open the chart and trace the path your eyes should have taken. When you review a missed MSR, identify which tab you over-read.

Volume helps less than honest review here. Twenty questions reviewed properly will move your score more than a hundred questions ground through.

What to look for in a DI study plan

If you're building a study plan for DI — either yours or one a tutor or platform builds for you — the things to insist on:

• Explicit per-type drilling before any mixed practice.
• Original questions, not recycled IR. The Focus DI section has different format weights and a tighter clock; old IR questions teach the wrong intuition.
• Format-realistic practice.A printed PDF of Table Analysis questions doesn't teach you to use the sort. You need an interface that mimics the test, or you are practising the wrong skill.
• An error-tagging systemwith a small fixed vocabulary — conceptual, format, time pressure, misread — not free-text notes you'll never re-read.
• A weekly full-section mock with a debrief that breaks the section apart by question type.

The short version

DI is the highest-leverage section for most students because it is the most under-trained and the most format-dependent. Five question types, each with one or two specific traps, all sharing the same underlying skill: extract the answer from messy information without burning the clock. Drill per-type first, time from day one, log every miss with a tag, and mock the full section weekly. The score moves quickly once the format clicks.

The platform

Zakarian GMAT was built for exactly this loop. The DI chapters cover each question type in detail with format-realistic practice, the error log uses the six-tag taxonomy this guide describes, and mocks are scored against the same 60–90 scale as the real exam. The diagnostic identifies your weakest DI sub-type in 30 questions, and the adaptive plan focuses your week on it. If you're working on DI and want a system that holds the loop for you, the diagnostic is free.