The GMAT Data Insights section rewards a specific kind of reading: a fast, almost surgical triage of multi-source prompts, where the candidate separates the information the question actually needs from the data the test designer placed there to distract. Learning how to identify relevant information on the GMAT is one of the highest-leverage skills in Data Insights, because every other analytical move — calculating, comparing, choosing — depends on pulling the right thread first. Candidates who treat the section as a quantitative test lose minutes on irrelevant rows. Candidates who treat it as a reading test under a clock win the section before opening the calculator.
What the data-heavy Data Insights items actually ask of you
Data Insights has 20 questions in 45 minutes, drawn from five question types: Data Sufficiency, Multi-Source Reasoning, Table Analysis, Graphics Interpretation and Two-Part Analysis. There is no separate "Data Interpretation" item type. The items where relevant-information triage matters most are the ones built on displays. Multi-Source Reasoning spreads its information across two or three tabs — a short text, a chart, a table — and asks several questions about them. Table Analysis gives one sortable table and a set of statements to judge. Graphics Interpretation gives one graph or image and asks you to complete statements from drop-down menus. In each case the candidate's task is rarely to perform an exotic calculation. It is, more often, to decide which display, tab or column actually contains the answer, then to extract a single value or ratio cleanly.
Three structural facts shape how these items behave. First, the answer options are designed to punish mis-extraction — a value from the wrong row or the wrong year often produces one of the trap options. Second, when several displays are involved, they are redundant in places and complementary in others, so the relevant information often lives in the overlap, not in either display read alone. Third, the prompt names the variable being asked about in plain language, but the variable is often one step removed from the column heading you might first look at.
Why multiple displays change how you should read
Most candidates come into GMAT prep trained to read a single chart from top to bottom: title, axis, legend, then data. That habit is slow and wasteful on a multi-display item, because the question stem is doing some of the work a chart title would normally do. Reading the stem first, then jumping to the display that contains the named variable, is a more efficient sequence. It also reduces the chance of forming a false impression from a visual that the question never asked you to evaluate.
The time this re-ordering saves compounds across the 20-question section, which averages 2 minutes 15 seconds per question. In a section where pacing shapes the score as much as accuracy does, those recovered seconds matter.
The seven information categories that appear in Data Insights prompts
Before a candidate can decide what is relevant, they need a mental inventory of the kinds of information that can appear. The variables that show up in chart, table and multi-source items fall into seven recognisable families. A practiced reader recognises the family from the first few words of the stem and pre-loads the display they expect to consult.
- Absolute values — single-point readings such as revenue, market share, or unit cost for one named category in one named period.
- Period-over-period change — percentage or absolute movement between two adjacent or non-adjacent periods.
- Cross-section comparisons — the value of one category versus another within the same period, often a sub-segment against a total.
- Ratios and proportions — share of total, ratio of two named categories, or contribution to an aggregate.
- Trend slope and direction — the shape of change across a series, often read off a line graph rather than a table.
- Rank and order — which category sits first, second, or last on a given metric in a given period.
- Threshold and conditional readings — values above, below, or equal to a stated cutoff such as a target, a budget, or a baseline.
When the stem names a family, the relevant display is usually the one that holds that family cleanly. Tables suit absolute values and period-over-period change. Charts suit trend slope, rank, and threshold reading. Cross-section comparisons and ratios tend to live at the intersection: a chart for the visual ordering, a table for the precise value once you have narrowed the category.
Mapping the seven families to the displays
A useful exercise during preparation is to take ten chart, table or multi-source items and tag every question stem with one of the seven families. After the tagging pass, look at where the correct answer came from. The distribution in your own practice set tells you where to invest your reading speed: whichever families dominate are the volume work, and they reward a clean first read of the display that holds them.
A two-pass reading protocol for identifying relevant information
The protocol has two passes, each with a different goal. The first pass is structural: where is the named variable, and which display holds it. The second pass is extractive: pull the value, confirm the unit, and check the answer option. Many candidates collapse the two passes into one, which is a common source of careless errors on this section.
Pass 1: structural triage
Read the stem once. Underline or mentally mark the noun phrase that names the variable — for example, "average revenue per store" or "percentage change in operating margin." Glance at the available displays or tabs only to identify which one carries that noun phrase most directly. The chart usually shows the trend; the table usually shows the precise numbers. If the stem asks for a precise value, the table wins. If the stem asks for the shape of change or the rank, the chart wins. This is the moment where you decide which part of the prompt is relevant and which part is decoration.
Pass 2: extractive read
Now open the relevant display and find the row, column, or point that the stem specifies. Note three things: the period, the unit, and the basis (per store, per employee, indexed, inflation-adjusted). These three notes are where careless errors hide. A candidate who reads "operating margin" without checking whether the chart shows margin in points or in percent will pick a trap option that is far too small. A candidate who reads "2023" when the table is in fiscal years will mis-extract by one column.
The final seconds of pass two are spent inverting the question: ask yourself what option would be a reasonable distractor, and check that you have not produced it. For example, if the question asks for the difference between 2022 and 2024 and your calculation gives a positive value, the option that is the negative of your number is a designed trap. If you see it sitting in the list, you have just saved yourself from picking it.
When the second display is the relevant one - and how to spot it
Many multi-display items hide the relevant information in the secondary display rather than the visually dominant one. A line chart of total revenue per year sits on one tab, and a small table of segment breakdowns sits on another. The candidate's eye lands on the chart, but the question is about a specific segment. The chart is a decoy; the table is the answer. Recognising this pattern is the difference between a quick solve and a long struggle.
Three signals that the secondary display is the relevant one
First, the stem names a noun that does not appear as a chart legend, axis label, or title. If the chart is about "total revenue" and the stem asks about "contribution from the cloud segment," the segment lives in the table. Second, the stem includes a qualifier that the chart cannot resolve — "adjusted for acquisitions," "on a constant-currency basis," "excluding the divested unit." These qualifiers almost always live in a footnote, a text tab or a supporting table. Third, the answer options have a precision that the chart cannot produce. If the chart can only be read to two significant figures but the options carry three, the precise value came from a table or a footnote, not the visual.
For most candidates reading this, the second signal — the qualifier — is the one that gets missed most often. A quick mental check is to ask: "Does the chart as drawn actually answer this question on its own?" If the honest answer is no, switch to the supporting display before calculating anything.
Common pitfalls and how to avoid them
- Reading the display that is open first. Position on the screen is not a hint. The stem decides which display is relevant. Train yourself to read the stem first, every time, even when the chart looks inviting.
- Trusting the legend over the footnote. Legends summarise; footnotes and text tabs qualify. The qualifier is usually the part of the question that breaks a candidate's first calculation.
- Skipping the unit check. A chart in millions and a table in thousands will produce answers that differ by a factor of a thousand. A quick unit scan prevents a long recalculation.
- Re-reading the stem after calculating. Re-reading is good; re-reading to second-guess a correct extraction is bad. If your extraction matches the stem and the unit, trust it and move to the answer options.
- Sinking far more than the average time into one item. GMAC advises answering every question, even with an educated guess, so enter your best answer, bookmark the item and keep going; at the end of the section you can review bookmarked items and change up to three answers if time remains.
How identifying relevant information interacts with the GMAT scoring model
The Data Insights section is scored on a 60 to 90 scale and counts equally with Quantitative Reasoning and Verbal Reasoning towards the 205 to 805 total. The section adapts question by question, so the score reflects both how many questions you answer correctly and how difficult they were. The relevant-information skill is therefore not just about getting individual items right — it is about freeing up minutes for the rest of the section, including Two-Part Analysis and Data Sufficiency, and about reaching the end of the section with an answer for every question, since GMAC counts the number of questions you leave unanswered in the score calculation.
The scoring also rewards consistency. A candidate who reliably solves the chart, table and multi-source items through clean triage contributes a stable block of correct answers, which keeps the adaptive estimate moving upward rather than swinging back after careless misses.
Why preparation strategy should mirror the section's mix
Candidates preparing for the GMAT often over-invest in Data Sufficiency because the format is unique to the exam. A balanced preparation plan also gives reading-protocol practice to Multi-Source Reasoning, Table Analysis and Graphics Interpretation, because the relevant-information skill is shared between them. A candidate who drills Data Sufficiency algebra for many hours but never practises the two-pass reading protocol leaves avoidable points on the table in the display-based items.
Drilling the skill: how to practise identifying relevant information efficiently
Practice for this skill has a poor return if it is done as full-section timed tests from day one. The relevant-information habit is built in short, focused drills, ideally 20 to 30 minutes per session, three to four times a week for two to three weeks. The drill format matters more than the source of the items.
Drill 1: stem-tagging only
Take ten chart, table or multi-source items and write down, for each, only the variable the stem is asking about and which of the seven families it belongs to. Do not calculate. Do not extract. The goal is to train the first-pass triage. After ten items, review: did you mis-tag any of the families? A common slip is tagging threshold and conditional readings as absolute values, because the words "what was the value" appear in both. The difference is the presence of a cutoff in the stem.
Drill 2: relevant-display only
For the same ten items, write down which display — chart, table or text tab — is the relevant one, and the single line in the stem that told you so. This drill builds the second-signal habit: when the chart cannot resolve the question on its own, the table or text is relevant. With practice you should be making the correct call within seconds of reading the stem.
Drill 3: extraction speed
Now time the second pass only. Open the relevant display, find the value, and write it down. The unit check is part of the timed portion. After ten items, calculate your average and track it across sessions. A suspiciously fast extraction is usually a sign that the unit check has been skipped.
Worked example: applying the protocol to a representative item
Consider a Multi-Source Reasoning set with two tabs. Tab 1 is a line chart showing annual revenue (in $ billions) for a company from 2020 to 2024, with one line for total revenue and a second line for revenue excluding the divested hardware unit. Tab 2 is a small table showing the same five years broken out by segment: cloud, services, hardware, and other, in $ billions, with one decimal place. The question reads: "By approximately how many $ billions did the cloud segment's revenue increase from 2022 to 2024?"
Pass one: the stem names one variable — cloud segment revenue — for two periods, which puts the question in the period-over-period change family. The relevant display is the table, because the table is the only place where cloud is broken out as a segment; the chart only shows total and total-ex-hardware. The chart is decorative for this question.
Pass two: extract cloud revenue in 2022 and 2024 from the table and take the difference. Check the unit: the table is in $ billions, so the difference is in $ billions. The option that is the negative of the correct value is the designed trap; confirm it is present and reject it.
With the protocol in place, this is a short solve. Without it, candidates often read the chart first, notice the two lines, and try to derive cloud from the gap between them — a calculation that does not work because the gap is total minus hardware, not total minus cloud. That path costs dead-end effort before the candidate switches to the table.
When the relevant information is hidden in a footnote or qualifier
Some items hide the relevant slice of information in a footnote attached to one of the displays or in a sentence on a text tab. The note may say "all values in constant 2020 dollars," or "hardware divested in Q3 2023, values shown pro forma," or "segments reclassified in 2022; prior years restated." Each of these qualifiers changes the answer, and none of them is visible in the chart's title or legend.
The reading protocol for footnote-loaded items is the same two passes, with one addition: in pass two, after extracting the value, re-read the footnote and confirm that the value you extracted is on the basis the question is asking about. The unit check and the basis check are siblings. A candidate who does both will rarely lose a point to a footnote on a well-designed item.
Three footnote patterns to memorise
First, currency and inflation adjustments: "in constant 2020 dollars," "nominal," "adjusted for a currency devaluation." Second, scope adjustments: "excluding the divested unit," "pro forma for the acquisition," "on a like-for-like basis." Third, classification restatements: "segments reclassified in 2022," "prior years restated under the new standard." Each of these patterns interacts with the question in a predictable way. Currency and scope adjustments change the value; classification restatements change which row you read from. A practiced reader recognises the pattern from two or three words and adjusts the extraction accordingly.
Comparative read: identifying relevant information across the display-based question types
The relevant-information skill is shared across Multi-Source Reasoning, Table Analysis and Graphics Interpretation, but it shows up in slightly different forms. A short comparison helps you transfer the protocol between them.
| Question type | Typical prompt structure | Where the relevant information usually lives | Most common reading error |
|---|---|---|---|
| Multi-Source Reasoning | Two or three tabs (text, charts, tables), several questions | The tab that holds the named variable; context and qualifiers often sit on the text tab | Staying on the tab that is open because it is open |
| Table Analysis | One sortable table, several statements to judge | The column the statement names; the row the statement names | Sorting the table by the wrong column after a misread |
| Graphics Interpretation | One graph or image, statements completed with drop-down menus | The part of the graph each blank refers to | Reading the wrong axis or series for one of the blanks |
The pattern across all three is the same: read the stem, identify the named variable, jump to the display that holds it, extract, check the unit and the basis. The differences are mechanical. Graphics Interpretation forces a check of each drop-down blank. Table Analysis forces a sort-and-extract. Multi-Source Reasoning forces a tab switch. Once the protocol is in place, switching between them is a small adjustment, not a new skill.
Final preparation: building the protocol into your GMAT section timing
By the final two weeks of preparation, the two-pass reading protocol should feel automatic. A useful final drill is a ten-item set of chart, table and multi-source items, untimed, where you speak the protocol aloud for each item: stem read, variable named, family identified, relevant display chosen, value extracted, unit and basis checked. The verbalisation slows the process enough that you can audit each step. Once the spoken version is clean, switch to timed practice and let the protocol run silently.
On test day, the protocol is your safety net. A clean first-pass triage prevents one of the section's most expensive errors — spending minutes on a question because the relevant information was in the second display and you did not switch. A clean second-pass extraction prevents one of its most common errors — picking a designed trap because the unit or the basis did not match.
Conclusion and next steps
Identifying relevant information in GMAT Data Insights is a learnable protocol, not an innate talent. Read the stem first, name the variable, tag the family, jump to the relevant display, extract the value, check the unit and the basis, and invert the question to spot the designed trap. Drill the protocol in short focused sessions for two to three weeks, then let it run silently under timed conditions. Candidates who build the habit before test day walk into the Data Insights section with a time surplus and a stable extraction routine. TestPrep Istanbul's Data Insights diagnostic walk-through is a natural next step for candidates ready to convert this protocol into a section-level score.