Five item families, one section score

The GMAT Focus Data Insights section is the third scored module of the exam and the one most candidates underestimate before they sit down to take a practice test. It blends the table-reading speed of Verbal with the arithmetic discipline of Quant, and it does so inside visual stimuli that are not found anywhere else on the test. An 80+ section score on Data Insights signals to admissions committees that a candidate can read an unfamiliar chart, decide what the question actually wants, and commit to a defensible numeric answer in under two and a half minutes on average. The work to reach that line is mechanical, repeatable, and surprisingly resistant to last-minute cramming, which is why a study architecture built around the section is worth the planning effort.

What the Data Insights section actually tests: five item families, one shared skill set

Before any candidate can plan a serious study schedule, they need a clean mental picture of what the section is built from. Data Insights on the GMAT Focus contains roughly twenty scored questions, distributed across five item families that the test-makers have used for several administrations. The families are: Data Sufficiency, Multi-Source Reasoning, Table Analysis, Graphics Interpretation, and Two-Part Analysis. Each family carries its own stimulus type and its own answer-entry format, but they all sit on the same underlying skill: a candidate has to extract structure from an unfamiliar data display, decide what the question is asking, and either commit to a number, choose among statements, or split a problem into two coordinated answers.

Most candidates reading this will recognise Data Sufficiency from the Quant section of older GMAT formats, and that recognition is useful, because the underlying logic is identical. A Data Sufficiency stem on the Focus does not, however, ask for a value; it asks whether a value can be determined, and the answer choices are always five fixed statements about sufficiency. Multi-Source Reasoning presents two or three tabs of data — a short passage, a chart, and occasionally a small table — and asks three to four questions that move between them. Table Analysis gives candidates a sortable table and asks them to sort, filter, or interpret columns, with the answer entered as discrete options in a drop-down. Graphics Interpretation uses bar charts, line charts, scatter plots, and the occasional infographic, and most of its questions ask candidates to choose between two or three drop-down statements that complete a sentence. Two-Part Analysis presents a single shared stimulus and asks a question whose answer is a pair of coordinated values drawn from a list of six to eight options.

The common thread is that none of these families rewards raw arithmetic. A candidate who can multiply three-digit numbers quickly will still lose marks if they cannot decide what to multiply in the first place. The section score is driven by reading precision, structural thinking, and the discipline to ignore numbers that the stem does not need. A study plan that treats Data Insights as a hybrid of Quant and Verbal drills will outperform one that tries to sharpen arithmetic alone.

The 80+ section score target: what it implies about your accuracy budget

Translating a section score into a raw accuracy target is not an exact science, because the GMAT Focus uses adaptive scoring and the official scale goes from 60 to 90 in five-point increments. The practical rule that I share with candidates is that an 80 reflects roughly 75 to 80 percent of available points in the section, once easy-versus-hard module difficulty is taken into account. That range is worth internalising, because it changes how you triage your weakest family. If a candidate is missing 8 out of 20 questions at random, the section score lands in the high sixties. Cutting the miss count to 4 or 5 will push the score into the mid-to-high seventies. Cutting it to 2 or 3 reliably clears the 80 line on a standard administration.

The accuracy budget matters because it tells you where to spend preparation time. A candidate who is losing points in a single family can climb two or three scaled points by addressing that family alone, whereas a candidate who is scattering errors across all five families needs a broader intervention. In my experience tutoring, the second profile is more common than the first. Candidates tend to over-prepare the family that feels most familiar, which leaves the other four under-drilled even when the cumulative loss is the larger problem.

Setting the budget also disciplines the late-stage review. A candidate who needs two or three additional correct answers to clear 80 can target specific stem types in the final two weeks, rather than trying to relearn the entire section. The 80+ target is also the threshold at which Data Insights stops dragging the total score down. Above that line, the section becomes a quiet contributor to a competitive total, and below it, the section tends to cap the total regardless of how strong the other two scored modules are. Treat the 80 as a hinge, not a vanity figure.

A 10-week study architecture calibrated for the section score climb

The reason to plan across ten weeks rather than six or twelve is that Data Insights rewards a layered build. The first layer is fluency with each item family, the second is the discipline to triage the easiest points first under time pressure, and the third is the stamina to hold accuracy across a full-length practice test. Most candidates who try to compress the build into four weeks end up with a fragile skill set that holds on easy modules and collapses on hard ones. A ten-week window allows each layer to settle.

Across the first two weeks, a candidate should complete one untimed diagnostic from each of the five item families, scoring themselves by family and by stimulus type. The goal here is calibration, not performance. The diagnostic answers three questions: which family is strongest, which family is weakest, and which family has the largest gap between recognition speed and execution accuracy. A candidate who is fast but inaccurate on Multi-Source Reasoning is a different study case from one who is slow but accurate, and the two cases deserve different drills. Write the diagnostic results down. They become the baseline for the rest of the plan.

Weeks three through six are the family-by-family build. Spend roughly one week on each family, starting with the weakest, working through the middle three, and ending with the strongest. For each family, the weekly structure should be: 25 untimed items spread across stem types on day one and two, 20 timed items on day three, a review session on day four in which every missed item is rewritten from the stem, and a 15-item mixed-family set on day five. The review session is the highest-leverage hour of the week. Without it, candidates repeat the same error two and three weeks later, because untimed practice hides the structural mistake that timed practice exposes.

Weeks seven and eight are the integration phase. Move to full-length Data Insights sections, scored, with strict timing. Two sections per week is plenty. Between sections, keep a single-page error log that names the family, the stem type, the error mode, and the fix. Common error modes include: misreading the axis on a Graphics Interpretation question, treating a Two-Part Analysis answer as two independent choices rather than a coordinated pair, and answering a Multi-Source Reasoning question from the wrong tab. Naming the mode is what allows the fix to stick.

Week nine is the pacing reset. Take one full-length Focus mock, score it, and use the result to set a minute-per-question budget for test day. Most candidates I work with end up at around two minutes and fifteen seconds per question, with five to seven extra minutes held in reserve for the two or three hardest items. Week ten is taper. Light review only, no new item types, and a final short section on day three or four of the final week. The taper is what keeps the gains from week nine intact on test day.

Triage logic for the five item families under time pressure

The mistake most candidates make on the section is to treat the questions in presentation order. The five item families are not, however, distributed evenly across the section, and they are not equally generous with their points. A triage logic that I teach for the 80+ climb is to start the section by scanning the first ten questions, identifying the families, and committing to the easy and medium items before going back to the hard ones. The scan should take no more than ninety seconds.

Data Sufficiency items, when they appear, tend to be the highest-yield family for candidates who are comfortable with the sufficiency logic. They reward a specific decision tree: ask whether each statement alone is sufficient, then ask whether the two together are sufficient, and stop. A candidate who follows the tree correctly can clear Data Sufficiency items in roughly one minute and forty-five seconds each, which leaves time for slower items later in the section. Two-Part Analysis is the family that most candidates under-practice. The shared stimulus is a hint: the question is testing whether you can identify two values that jointly satisfy a condition, and the six-to-eight-option answer list is a constraint, not a free choice. Read the stem twice before touching the list.

Multi-Source Reasoning rewards candidates who label the tabs before reading the question. A fifteen-second investment in writing a one-word summary above each tab pays back across the three or four questions that follow. Table Analysis is the most mechanical family, and the discipline is to read the column headers, not the rows, before answering. Graphics Interpretation is the family where candidates lose the most time to second-guessing. Commit to the drop-down answer, mark the question, and move on. If the section score is built from these five habits, the 80+ line is genuinely reachable in one sitting.

Common pitfalls and how to avoid them

Arithmetic-first reading. Candidates read the chart and start computing before they have parsed the stem. The fix is a two-pass read: pass one extracts the chart structure, pass two reads the stem. The two-pass read costs fifteen seconds and saves two or three minutes across the section.
Drop-down paralysis on Graphics Interpretation. Candidates treat the drop-downs as confirmation tasks and reread every option. The fix is to predict the answer before opening the drop-down, then select the matching option. If none matches, that is a signal to reparse the stem, not to reread the options.
Two-Part Analysis treated as two separate questions. Candidates answer the first half, lock it in, and answer the second half independently. The fix is to read the stem for the joint condition and treat the answer as a single pair. The two values almost always interact.
Data Sufficiency statement creep. Candidates use the second statement to revise their reading of the first. The fix is to evaluate each statement in isolation, mark the answer when the tree is clear, and move on. Statement two only matters if statement one is insufficient on its own.
Multitasking across tabs in Multi-Source Reasoning. Candidates switch tabs while reading the question, which costs them the structure of each tab. The fix is to read the question fully, then jump to the named tab. Tab labels in the question are deliberate, not decorative.

Reading the stem twice: a two-pass method that protects accuracy

Most of the section's hardest questions are not hard because the arithmetic is difficult. They are hard because the stem is doing two jobs at once and the candidate notices only one. A two-pass read is the cheapest, most reliable way to catch the second job. Pass one reads for the question type. Is the stem asking for a single value, a comparison, a condition, a sufficiency decision, or a coordinated pair? Pass two reads for the constraints. Which chart, which table, which column, which axis, which units, and which time window.

The two passes do not need to be slow. A candidate who has internalised the pattern can complete both passes in thirty to forty seconds, which is well within the time budget. The discipline is to refuse to read only once. Candidates who read once are reading a different question than the one that is on the screen, and they do not realise it until the answer choices fail to make sense. By the time they go back to the stem, two minutes have evaporated.

In my experience tutoring, the two-pass method is the single biggest lever for candidates who are sitting in the mid-70s. It is a small change in behaviour, it costs no extra knowledge, and it lifts the accuracy rate on the three or four questions that separate a 78 from an 82. The candidates who already read once and got it right do not need the method. The candidates who read once and missed the second job are exactly the ones who will benefit.

How the section score climbs above 80: the difference between 78 and 84

Once a candidate clears 80, the climb to 84 or 85 is a different problem. The errors that were tolerable at 78 are no longer tolerable at 84, which means the preparation has to shift from family-level fluency to stem-level discrimination. The candidates who break into the mid-80s are the ones who can identify, in real time, the two or three stem shapes that signal a question they are likely to miss, and who have a pre-planned response for each. There is no shortcut here. The lift comes from drilling the small set of stems that drive the section score, and from refusing to guess on the stems they have not drilled.

The diagnostic that separates a 78 candidate from an 84 candidate is the following: take twenty timed items from the candidate's weakest family, score them, and then sort the misses by stem shape. A candidate who misses one of each stem shape at random is a 78 candidate. A candidate who misses three or four items from the same stem shape is an 84 candidate in waiting. The fix is to drill that single stem shape until the miss rate drops below one in five, and then re-test. Targeted drilling of a single stem shape can move the section score by two to four points in a single week.

The other lever above 80 is the pacing reserve. A candidate who has been spending the full two and a half minutes on every item will not climb past 82, because the section contains a few items that legitimately require more than the average. The candidates who break 84 hold five to seven minutes in reserve, which they spend on the two or three hardest items of the section. Building the reserve requires cutting time on the easy items first, which is its own discipline.

Building a personal error log that actually changes behaviour

An error log only works if the entries are specific enough to drive a change in behaviour. The format I recommend has four columns: the family, the stem type, the error mode, and the fix. The family is one of the five. The stem type is a more granular description, such as "Graphics Interpretation with a stacked bar" or "Two-Part Analysis with a joint condition." The error mode is the cognitive mistake, such as "read the wrong axis" or "treated the pair as independent." The fix is the single behavioural change that will prevent the same mode next time.

The log should be reviewed weekly, not daily. A daily review creates anxiety without creating change. A weekly review allows patterns to emerge, and the patterns are what the fix column is designed to address. A candidate whose log shows the same error mode three weeks in a row knows exactly which drill to schedule, and a candidate whose log shows a new mode each week knows the build is going well and the section score is climbing.

The error log is also the single best input for the week nine pacing reset. The mock score tells you how the section is performing on aggregate. The error log tells you which two or three stem shapes to bank time against in the final week. Treat the log as the second-most-important document in the preparation, after the diagnostic baseline.

Pacing, the 90-second chart scan, and the discipline of a section-level plan

The chart scan at the start of the section is a tactical habit, not a study habit. It belongs in the test-day plan, but it is rehearsed during the integration phase. The scan has three jobs: identify the families in the first ten questions, flag any stimulus that looks like a Two-Part Analysis or a Multi-Source Reasoning set, and decide the order in which to attempt them. The scan should not try to answer any question. It is a reconnaissance pass, and it ends with a written or mental note that says "start with items three, five, and seven; return to one, two, four at the end."

The minute-per-question budget should be set during week nine, after the integration phase, and it should be honest. A candidate whose integration phase results show a comfortable 75 percent accuracy at two minutes and fifteen seconds should plan for that budget on test day. A candidate whose accuracy drops below 70 percent at the same budget should plan for two minutes and thirty seconds, and should hold a larger reserve. The budget is a forecast, not a target, and it should be updated after every mock.

Finally, the section-level plan should be written down before test day. The plan names the order of attack, the minute budget, the error log entries to watch for, and the two or three stem shapes that warrant a flag. A candidate who walks into the test centre with a written plan is a candidate who has rehearsed the section dozens of times. A candidate who walks in cold is a candidate who is hoping the section will go well. Hope is not a study strategy.

Conclusion and next steps

An 80+ section score on the GMAT Focus Data Insights module is the product of layered preparation, a clean triage logic, and a pacing reserve held in reserve until the section's hardest items appear. The ten-week architecture above is one way to organise that preparation, but the principles are portable: build family fluency first, integrate under timed conditions second, taper into a written plan third. Candidates who follow the principles tend to clear 80; candidates who skip the taper tend to sit at 78 and wonder why the section score refuses to move. The next step for any candidate reading this is to run the week-one diagnostic across all five item families and to write down the baseline. The baseline is what the rest of the plan is calibrated against. TestPrep Europe's diagnostic assessment is a natural starting point for candidates building a sharper GMAT Focus Data Insights preparation plan, especially for those who want a written baseline before they commit to the ten-week build.

Comparative table: the five item families at a glance

Item family	Stimulus type	Answer format	Average time target	Most common error mode
Data Sufficiency	Short stem, no chart	Five fixed statement options	1 minute 45 seconds	Using statement two to revise statement one
Multi-Source Reasoning	Two or three tabs of mixed data	Standard multiple choice	2 minutes 15 seconds per question	Reading the question from the wrong tab
Table Analysis	Sortable table with headers	Drop-down per column	2 minutes per question	Reading rows before headers
Graphics Interpretation	Bar, line, or scatter chart	Two or three drop-downs	2 minutes per question	Misreading the axis or units
Two-Part Analysis	Single shared stimulus	One pair from six to eight options	2 minutes 30 seconds	Treating the pair as independent

What separates the Data Insights section from a Quant section in study design

The two scored Quant-adjacent sections of the GMAT Focus are easy to conflate, and the conflation is one of the most common reasons a candidate's Data Insights score stalls. Quant is built around word problems whose answer is a single value, and the discipline is to translate the prose into algebra, compute, and commit. Data Insights is built around visual stimuli whose answer is often a structural judgement, and the discipline is to read the stimulus, parse the stem, and then decide whether the work is arithmetic, logical, or simply interpretive. The two sections share a numeric vocabulary, but they reward different habits. A study plan that imports Quant drills into Data Insights will sharpen arithmetic without sharpening the section score, because the section score is gated on stimulus reading, not on raw computation.

The other difference is the role of partial credit. Quant questions are all-or-nothing: a single value, a single correct answer. Data Insights items are also all-or-nothing, but the cognitive path to the answer is layered in a way that Quant questions rarely are. A candidate who gets the chart structure right and the stem parsing right but the arithmetic wrong on a Data Sufficiency item still scores zero, which means the cost of a small mistake is the same as the cost of a large one. The implication for study design is that Data Insights rewards drilling the small mistakes harder than Quant does. A two-minute fix to a stem-parsing habit is worth more than a thirty-minute drill on a numeric shortcut.

How to know when you are ready: the signals that precede an 80+ on test day

Readiness for an 80+ is not a feeling. It is a pattern of measurable signals that appear in the last two weeks of preparation. The first signal is consistency: across the two full-length practice tests of the integration phase, the candidate's section score should land within three points of itself, even when the second test is harder. A candidate whose scores swing by eight points between two practice tests is not yet ready, regardless of the higher score. The second signal is family coverage: by week eight, the candidate should be hitting at least 70 percent accuracy on every family, and at least 80 percent on three of the five. A single family below 60 percent will cap the section score no matter how strong the other four are.

The third signal is the error log. By week nine, the log should show fewer new error modes per week than it did in week four. A log that is still generating new modes at week nine is a signal that the diagnostic was incomplete, and the candidate should consider a second diagnostic on the family that is generating the new modes. The fourth signal is the pacing reserve. A candidate who has been forced to spend the full time budget on every item in week eight will not find a reserve in week ten. The reserve is built by deliberately cutting time on the easy items, and that habit is rehearsed, not summoned.

When all four signals are present, the candidate is ready. When two of the four are present, the candidate has another two weeks of work to do. When only one is present, the test date should be pushed, because the section score will not climb under pressure. The signals are not guarantees. They are probabilities, and they are the best probability estimate a candidate can build without sitting the actual exam.

Final tactical notes for the day before and the morning of the test

The day before the test should be a review day, not a study day. Re-read the error log from the last three weeks, re-read the section-level plan, and re-run a fifteen-item mixed-family set at a relaxed pace. The goal is to confirm that the habits are still in place, not to learn anything new. Candidates who try to learn a new stem shape the day before the test are the candidates who walk into the test centre with one more shape in their head than they have rehearsed, and the un-rehearsed shape is exactly the one that will appear on the section's hardest item.

The morning of the test should include a light review of the triage logic: the order of attack, the minute budget, the chart scan, and the two-pass read. Eat a meal that does not include unfamiliar food, hydrate normally, and arrive early enough to sit for ten minutes in a quiet space before the test begins. The first ten questions of the section will set the tone for the rest of the sitting, and a calm entry into those questions is worth more than any single drill from the previous ten weeks.

Frequently asked questions

How many weeks does it take to reach an 80+ on GMAT Focus Data Insights?

Most candidates who already score in the mid-70s need roughly ten weeks of layered preparation: two weeks of diagnostics, four weeks of family-by-family build, two weeks of timed integration, one week of pacing reset, and one week of taper. Candidates starting from the mid-60s usually need a longer build, because the family-level gaps are larger.

Which Data Insights item family is the easiest to improve on?

In my experience tutoring, Table Analysis and Data Sufficiency are the two families where targeted drilling moves the section score fastest. Table Analysis is mechanical and rewards column-header discipline, while Data Sufficiency rewards a clean decision tree that candidates can rehearse in isolation from the other families.

Should I skip Multi-Source Reasoning items to save time?

No. Skipping a family on test day caps the section score, because the scoring algorithm weights every family. A better tactic is to budget two minutes and fifteen seconds per Multi-Source Reasoning question, label the tabs before reading the stem, and accept that the family is slower than the others.

What is the difference between a 78 and an 84 on the section?

The difference is usually two or three additional correct answers, concentrated in a single stem shape that the candidate has under-drilled. A 78 candidate tends to scatter errors across all five families, while an 84 candidate has reduced one family's error rate through targeted drilling of a specific stem shape.

How does Data Insights prep differ from Quant prep on the GMAT Focus?

Quant prep rewards arithmetic fluency and algebraic translation, while Data Insights prep rewards stimulus reading and stem parsing. A study plan that imports Quant drills into Data Insights will sharpen computation without lifting the section score, because the section score is gated on reading precision, not on raw numeric speed.

Five item families, one section score: how to triage GMAT Focus Data Insights for an 80+