How to rebuild GMAT Data Insights from below the 60th

The GMAT Data Insights section rewards candidates who can read a chart quickly, choose a clean arithmetic move, and protect a 2-minute average across roughly 20 items in 45 minutes. For candidates arriving with weak quantitative foundations, the section often feels like the hardest hill on the exam: the items are short, the visuals are dense, and a single missed sub-question can cost a full point on the 60–90 scale. The road map below is designed for the low-scorer who can read English fluently but who has not touched algebra, ratios, or basic statistics in years, and who needs a 60-day plan that is sequenced, not heroic.

What the Data Insights section actually measures, and why weak-foundations candidates underestimate it

The Data Insights section is not a maths test in the traditional sense. It is a measurement of how a candidate turns a piece of information — a table, a graph, a multi-source layout — into a numerical answer, a justified claim, or a chosen next step. The official item families include Data Sufficiency, Multi-Source Reasoning, Table Analysis, Graphics Interpretation, Two-Part Analysis, and the newer style of Data Interpretation prompts. Each family shares the same skill: locate the relevant data, run a single quantitative operation, and either produce a number, judge two statements, or rank a pair of options.

For candidates with weak foundations, the danger is that the section looks approachable. The stems are short, the language is plain, and the early items feel conversational. Underneath, every prompt hides a calculation. A weak-foundations candidate will often read a chart correctly, identify the right bar, and then lose two minutes converting units or computing a percentage change. That two-minute slip is the silent killer of Data Insights scores: it does not register as a content error, so the candidate keeps doing the same thing for weeks.

Three structural facts shape the road map. First, the section is adaptive at the section level on GMAT (formerly the GMAT Focus Edition), so early items set the floor for the entire 20-item set. Second, every item is worth the same on the 60–90 scale, so a Data Sufficiency item that you abandon costs the same as a Table Analysis item that you solve. Third, the section does not test calculus, trigonometry, or advanced algebra — it tests arithmetic, ratios, percentages, linear equations in two variables, and basic descriptive statistics. That last point is the most important for a low-scorer to absorb: the section is mathematically narrower than it appears, but the time pressure is steeper than Quant or Verbal.

Why a diagnostic week is non-negotiable

Most low-scorers skip the diagnostic and start drilling mixed sets. In my experience this is the single most expensive mistake on a 60-day plan. A diagnostic isolates the four or five item families where the candidate is genuinely slow, not just the families that feel uncomfortable. A clean diagnostic is 30 items, untimed, with full review afterwards. The candidate tags each item by family, by error type, and by minutes spent. Within a week, the candidate can see whether the ceiling is arithmetic, reading speed, or conceptual gaps in ratios and percentages.

Stage 1 — Foundations reset: 14 days to rebuild arithmetic confidence

The first two weeks are deliberately unsexy. The candidate abandons practice tests, abandons mixed sets, and works through a foundations sequence in three blocks. The goal is not to score well in Data Insights during these two weeks. The goal is to remove the moments in the section where the candidate freezes because 17 divided by 8 felt intimidating. Foundations work pays off across the entire section because every item family runs through the same arithmetic core.

The first block covers numerical fluency. Fractions, decimals, percentages, ratios, and proportional reasoning. The candidate drills conversions both directions: 3/8 to a percentage, 22.5% to a fraction, a ratio of 5:7 to a percentage share. The drill is timed at the question level, not the set level, because Data Insights punishes slow arithmetic far more than slow algebra. The target for this block is 8 seconds per single-step conversion. Most low-scorers sit closer to 20 seconds at the start, and that gap is the foundation of the recovery.

The second block covers linear equations and two-variable systems, because Data Sufficiency items in particular demand the ability to set up two equations and judge their independence. The candidate practices the form ax + by = c, learns to recognise parallel lines as dependent, and works through roughly 40 hand-picked items where the only skill is writing and reading the system. A common weak-foundations trap is to attempt to solve the system when the stem only requires recognising whether a given statement is sufficient — a habit that costs 60–90 seconds per item.

The third block covers basic descriptive statistics: mean, median, weighted mean, range, standard deviation as a concept rather than a formula. The block emphasises the difference between an average and a weighted average, because a surprising number of Data Insights items hinge on a weighted-mean move hidden inside a Table Analysis prompt. By the end of stage 1 the candidate should be able to compute a weighted mean across three categories in under 90 seconds without a calculator, even if the numbers are awkward.

Common pitfalls and how to avoid them

• Drilling mixed sets before the foundations blocks are finished. Mixed sets hide which block is leaking time.
• Over-reliance on the on-screen calculator. The calculator adds clicks, and clicks add seconds, and seconds add missed items.
• Skipping the rewrite step. Candidates who try to solve in their head on Data Sufficiency items lose 45 seconds per item on average.
• Confusing sufficiency with correctness. A statement can be sufficient even if the candidate cannot easily compute the final number.

Stage 2 — Item-family triage: 21 days to learn the visual fingerprints of each family

Once the foundations blocks are stable, the candidate moves to item-family triage. The objective here is to build a recognisable first move for each of the roughly six families, because the most expensive mistake in Data Insights is reading a Table Analysis stem as if it were a Multi-Source Reasoning prompt. Each family has a visual fingerprint, a default reading order, and a default answer move. The candidate studies all three.

Data Sufficiency items are the densest family and the most punishing for weak-foundations candidates. The visual fingerprint is the two statements labelled (1) and (2), the question stem, and the five-option choice set. The default reading order is: read the question stem, classify it as a value, a yes/no, or an unusual request, then judge each statement independently. The default answer move is to test the statements alone, then in combination, without ever computing the final answer. Candidates who compute final answers on Data Sufficiency lose two minutes per item and never recover.

Multi-Source Reasoning items present three tabs of information, of which one or two are usually a distraction. The default reading order is: skim all three tab labels, read the question stem, then return to the tab that contains the relevant data. The default answer move is to mark the irrelevant tabs and never open them. The weak-foundations trap is reading every tab in order, which burns three to four minutes per item. That habit alone can move a candidate's Data Insights score by a full point on the 60–90 scale.

Table Analysis items present a sortable table on the left and a list of statements on the right, of which several are true. The default reading order is: identify the column that the question targets, then sort mentally before checking statements. The default answer move is to commit to a sort direction early and let the table do the arithmetic. The weak-foundations trap is treating Table Analysis as a reading task; it is an arithmetic task disguised as a table.

Graphics Interpretation items present a single chart, usually with two related y-axes, and a short set of two or three sub-questions. The default reading order is: read both y-axis labels before touching the chart, then read the question. The default answer move is to anchor on one axis, convert, then check the second. The weak-foundations trap is assuming the chart is symmetrical, which it almost never is on adaptive items.

Two-Part Analysis items present a prompt with two simultaneous questions, often with five options in a column for each part, and the candidate selects one option per column. The default reading order is: read the prompt, classify the two parts, identify the shared calculation, then work both parts in parallel. The default answer move is to test the shared calculation once and read off both answers. The weak-foundations trap is solving the two parts sequentially, which doubles the arithmetic cost.

Stage 3 — Pacing architecture: 14 days to lock a 2-minute average

By week five, the candidate knows the families and has rebuilt arithmetic. The next task is pacing, and this is where most low-scorers quietly give back the gains they have made. The pacing architecture is built around a single number: 45 minutes for roughly 20 items, which is a 2-minute average. That average is non-negotiable on a target score of 78 or above. Candidates aiming for 85 or above should target closer to 1 minute 50 seconds per item, because two or three banked minutes are required to handle the two hardest items in the section.

The candidate learns a 3-tier triage. Tier 1 items are answered in under 90 seconds: small Data Sufficiency items, two-part items with clean arithmetic, table analysis items with a single sort. Tier 2 items are answered in 2 to 2.5 minutes: graphics interpretation items with two y-axes, multi-source items with one irrelevant tab. Tier 3 items are the two or three items in the section that will always take 3 to 4 minutes: a Data Sufficiency item with a non-linear setup, a multi-source item that requires cross-tab reasoning. The candidate flags Tier 3 items on the first pass, returns to them after Tier 1 and Tier 2 are cleared, and never lets a Tier 3 item block the section.

Banked minutes are not free time; they are insurance against a single expensive stem. In my experience, candidates who bank 4 to 5 minutes across the section consistently outperform candidates with the same raw accuracy who treat the section as a flat 2-minute run. The reason is that the adaptive algorithm places the harder items in the second half, and those items are exactly the ones that need a small buffer.

Worked pacing example

Consider a 20-item section. The candidate opens the section, commits to a 90-second budget on the first 8 items, accepts a 2-minute budget on items 9 to 16, and reserves items 17 to 20 for a 3-minute budget with a hard cap at 4 minutes. If the first 8 items close in 11 minutes, the candidate has banked 1 minute. If items 9 to 16 close in 14 minutes, the candidate has banked another 2 minutes. The final 9 minutes are then available for the four hardest items, which is more than enough headroom for a 3-minute Tier 3 item followed by three 2-minute items.

Stage 4 — Mixed sets and review loops: 14 days to harden the architecture

The last two weeks of the plan replace family-specific drills with mixed sets of 10 items, drawn from a source that respects the section's item-family mix. The candidate works each set under timed conditions, then performs a structured review. The review is where the score moves from 70 to 78, not the practice set itself. Most low-scorers spend 60% of their time on the set and 10% on the review, then wonder why the score plateaus. The correct ratio is closer to 40/40, with 20% left for the next set.

The review loop has four steps. Step 1, the candidate tags every missed item by error type: arithmetic slip, reading error, concept gap, or pacing overrun. Step 2, the candidate picks the two most common error tags and writes a one-paragraph note on the pattern. Step 3, the candidate redoes the missed items cold, the next day, without looking at the original work. Step 4, the candidate logs the minutes spent and notes whether the re-do cleared the item. If the re-do does not clear the item, the item returns to the foundations queue, not the practice queue.

The weak-foundations candidate should expect to see arithmetic slips dominate the error log for the first three weeks of mixed sets. This is normal and is the point of stage 1. By week seven, the error log should shift towards reading errors and pacing overruns, which is the signal that the foundations blocks have done their job. If arithmetic slips persist into week eight, the candidate should pause mixed sets and return to stage 1 for another week. The plan is staged for a reason; stages are not skipped because a calendar date arrives.

Reading the score report and the next diagnostic

By day 60 the candidate takes a second diagnostic, this time timed and section-realistic. The result is read against three thresholds. Below 60 on the 60–90 Data Insights scale, the candidate repeats stage 1. Between 60 and 70, the candidate extends stage 2 by two weeks and tightens the arithmetic drill. Between 70 and 78, the candidate moves to a maintenance plan of one mixed set per week, with full review, until test day. Above 78, the candidate shifts attention to Quant and Verbal, because the marginal return on Data Insights above 80 is small relative to the time investment.

For most candidates reading this, the realistic target is 76 to 80, which corresponds to the 70th to 85th percentile range on the GMAT percentile band. That target is achievable for a weak-foundations candidate on a disciplined 60-day plan, but only if the plan is staged and reviewed properly. A heroic 30-day sprint almost always fails on this section, because the section punishes unstable arithmetic far more than it punishes slow learning.

Comparative scoring snapshot

Maintenance and the final 14 days before test day

The final two weeks are about protecting the architecture, not extending it. The candidate runs one timed mixed set every five days, performs the full review loop, and avoids the temptation to add new item families. New families discovered this close to test day are a net negative, because they replace confident reading with uncertain reading. The weak-foundations candidate is far better served by sharpening the six families already learned than by chasing a seventh that might appear once on test day.

Sleep, hydration, and a familiar testing interface matter disproportionately on Data Insights. The section's 45-minute window is unforgiving of a slow start, and a candidate who has practised in the same browser, with the same calculator behaviour, and with the same note-board habits is roughly one Tier 1 item ahead of a candidate who has not. In my experience, the interface familiarity is worth about 30 seconds per item, which is two items across the section — a meaningful buffer on a tight 60–90 scale.

The candidate should also rehearse the moment of abandonment. Roughly two items per section will be too expensive to solve in the time available, and the correct move is to flag and move on, not to grind. Rehearsal means practising a deliberate 5-second pause, a flag, and a return to the item only after the Tier 1 and Tier 2 queue is cleared. A weak-foundations candidate who learns to abandon cleanly is almost always a higher scorer than a candidate with the same raw accuracy who grinds.

Frequently observed weak-foundations traps and the small fixes that resolve them

Trap one: reading every tab on a Multi-Source Reasoning item. Fix: skim the tab labels first, then read the stem, then open only the relevant tab. This saves 90 seconds per item.

Trap two: solving Data Sufficiency items to a final answer. Fix: judge each statement for sufficiency, then test the combination, then stop. The five-option choice set is a recognition task, not a computation task.

Trap three: ignoring the second y-axis on Graphics Interpretation items. Fix: read both axis labels before the chart, then anchor on the axis the question references. Most wrong answers come from the wrong axis.

Trap four: treating Two-Part Analysis as two separate items. Fix: identify the shared calculation first, then read off both answers. The shared calculation is the whole point of the family.

Trap five: practising only in untimed mode because timed mode is uncomfortable. Fix: timed mode is the actual exam. Practise in timed mode from week three onwards, with the timer visible, and accept the discomfort as part of the training.

A staged 60-day plan, a clean diagnostic, a serious review loop, and a willingness to abandon items that exceed the budget — these are the four levers that move a weak-foundations candidate from below the 60th percentile to a defensible 76 to 80 on the GMAT Data Insights section. Candidates who commit to the stages, rather than to a single heroic month, are the ones who clear the hill. TestPrep Europe's diagnostic assessment for the Data Insights item families is a natural starting point for candidates building this sharper preparation plan.

Frequently asked questions