GMAT Data Insights practice: master the five formats most students under-train

The five DI formats, and the one skill each actually tests

Data Insights rotates through five question types in a single 45-minute, 20-question section. A calculator is provided (it is not provided on Quant), which is a signal in itself: DI questions rarely hinge on arithmetic difficulty. They hinge on reading, structure, and sufficiency. Here is what each format is really testing.

Data Sufficiency: not 'solve the problem,' but 'do these statements pin down a single answer?' DS gives you a question and two statements, and the five answer choices are always identical. The skill is constraint-checking, not computing. A statement is sufficient only if exactly one answer survives every consistent interpretation of it; if you can find one edge case (a negative, a fraction, a zero) that produces a second answer, it is insufficient. The classic tax on DS scores is fully solving when a 30-second sufficiency check would have done.

Two-Part Analysis: fill two cells of a small table from a shared list of options, where the two answers are linked by a joint constraint. The mistake is treating the cells as two separate problems. Think of it as a tiny system of two equations: pick a value for the first cell and it often rules out values for the second. Once you name the linking constraint, the answer frequently falls out from logic with no computation.

Multi-Source Reasoning: two or three tabs of mixed text, tables, and charts, with several questions drawing on them. The skill is locating the one relevant fact across sources and ignoring the rest. The trap is over-reading every tab before answering. You read for structure first (what lives in which tab), then pull only what each sub-question needs.

Table Analysis: a sortable table plus usually three true/false-style statements. The leverage move is sorting the table to answer the actual question instead of scanning rows. Most errors here are misread column headers or units, not analytical failures.

Graphics Interpretation: a chart or graph with fill-in-the-blank dropdown statements. The skill is reading axes, scales, and trends precisely. The trap is the same as Table Analysis: a unit or axis you skimmed.

Why DI is the most under-trained section (and why that is good news)

Three structural reasons explain why DI is where most candidates leave points on the table.

First, it is the newest section, so there is simply less practice material in circulation than for Quant and Verbal, and a lot of older 'Integrated Reasoning' advice does not map cleanly onto the current Focus formats. Many study plans front-load the two sections people feel they understand and run out of runway before DI.

Second, DI feels like five small subjects, so it is easy to never build real fluency in any one of them. A student might see a handful of each format and conclude they 'get it,' without ever drilling a format until the recognition is automatic.

Third, DI has its own pacing math that rewards or destroys you fast: roughly 45 minutes for 20 questions is about 2 minutes 15 seconds each, but Multi-Source sets and dense tables eat far more than the average, which means you must bank time on the formats that should be quick. A Data Sufficiency item you can settle in 45 seconds with a clean sufficiency check funds the three minutes a multi-tab MSR set legitimately needs.

The good news: under-trained means high-leverage. Because most people neglect DI, focused practice moves your DI subscore faster than the same hours spent on an already-strong Quant or Verbal score. And because DI reuses sufficiency logic from Quant and critical-reasoning reading from Verbal, the work compounds across sections rather than competing with them.

How to actually practice DI: a method, not just a question count

Doing 200 random DI questions teaches you very little if you do not change what you do between questions. Use this loop.

Drill one format at a time, to fluency, before mixing. Spend a session only on Two-Part Analysis until the 'find the linking constraint first' move is automatic; then a session only on Data Sufficiency decision-tree discipline; and so on. Mixed practice tests recognition, but you cannot recognize a pattern you have not yet built. Build per-format, then mix.

Log every miss by cause, not just by topic. After each wrong answer, label why you missed it: a conceptual gap, a careless slip, time pressure, a misread, a strategy error, or other. DI errors cluster heavily in 'misread' (wrong column, wrong unit, wrong tab) and 'time pressure,' and those are fixed with completely different habits than a conceptual gap. If you do not separate them, you will drill content when your real problem was reading the axis.

Review with a stopwatch in mind. For every DS miss, ask: did I solve when a sufficiency check would have sufficed? For every Table or Graphics miss, ask: did I sort/read the axis, or scan? Write the recurring failure as a one-line rule you carry into the next session.

Space your review. Re-attempt the questions you missed a few days later, from your error log, not the day-of. The recognition you are building only sticks if you retrieve it after a gap. On the Zakarian GMAT platform this is automated: the six-tag error log (Conceptual, Careless, Time Pressure, Misread, Strategy, Other) feeds a daily spaced-review queue, and per-format and per-difficulty analytics show you exactly which of the five DI types is bleeding points. But the method works in a spreadsheet too. The discipline matters more than the tool.

A one-week DI practice block you can run yourself

If you have a week to make DI less scary, here is a concrete schedule that uses the loop above. Adjust the volume to your timeline; the sequence is the point.

Days 1-2, sufficiency and structure: Data Sufficiency only. Drill the five-choice decision tree until selecting A/B/C/D/E is mechanical. Force yourself to test each statement alone first, then combined only if both alone fail. Test edge cases (negatives, fractions, zero) on every problem that involves variables. Goal: stop solving when sufficiency is already determined.

Day 3, the GMAT-only format: Two-Part Analysis only. For each problem, write the linking constraint in words before touching the options. Eliminate from the pair, not from each cell separately.

Day 4, reading under load: Multi-Source Reasoning only. Practice mapping the tabs first (one sentence per tab: what is here), then answering each sub-question by pulling only what it needs. Resist reading everything.

Day 5, the quick-but-trappy pair: Table Analysis and Graphics Interpretation. Drill the habit of sorting the table to the question and reading every axis label and unit out loud before answering. Most misses here are reading, not math.

Days 6-7, mixed and timed: a full mixed DI set under the real clock (about 2:15 average per question, but deliberately bank time on DS and Graphics to fund MSR). Then review every miss by error cause and re-attempt your earlier-week misses from memory. That spaced re-attempt is where the gains lock in.

Where this fits in a full GMAT Focus plan

DI is one of three equally weighted sections, so the planning question is never 'DI or Quant?' but 'how much DI, and when?' Because DI is under-trained for most people, a disproportionate early push usually pays off: a few focused weeks on the five formats often moves the DI subscore more than the same hours added to a Quant score that is already in good shape.

The cleanest way to know where DI sits for you is to take an official mba.com practice exam and read the section breakdown. That official result is your real baseline, not any third-party estimate. From there, you weight your weeks toward whichever section the official report says is costing you the most.

That baseline-driven approach is exactly how the Zakarian GMAT study plan is built: you enter your official-practice-exam result, and the plan sequences chapters and drills around your actual weak sections rather than a generic order. The platform's Data Insights material covers all five formats as dedicated chapters with practice sets, plus full-length mocks with a mock-to-mock trend so you can watch the DI subscore move. It is free to use during the private beta, with no card required. None of that changes the core idea on this page: DI is winnable because it is neglected, and it gets winnable the moment you drill each format to recognition instead of collecting question counts.