How do you read a GMAT Table Analysis stem without skimming the data twice?

Table Analysis is one of the four question families inside the GMAT Focus Data Insights section, and it is the family where candidates who can read a graph fluently still drop the easiest points. The question looks harmless: a sortable table on the left, three to five statements on the right, a Yes/No binary for each statement. Yet the wording of each statement is engineered to exploit a specific reading mistake, and the table itself is built so that column headers do not always match the vocabulary in the stem. A Table Analysis item is not a comprehension test. It is a structured information-retrieval task with a strict scoring rule, and treating it like a reading passage costs candidates two to three questions per section, which on a 20-item Data Insights module is a real swing on the 60-to-90 scale.

This article walks through the exact protocol I would teach a candidate who is rebuilding Table Analysis from a starting score in the mid-60s. We will cover how the question is constructed, how the scoring treats Yes versus No, the three core moves that resolve almost every statement, and the column-shape patterns that decide which move to use first. We will also work through a complete item end to end so the protocol is visible in action, and close with a six-week practice plan tied to a target score band. The aim is for any candidate reading this to finish with a repeatable sequence of moves, not a vague sense that Table Analysis requires "careful reading."

What a Table Analysis question actually is on the GMAT Focus

Each Table Analysis item presents a single data table on the left side of the screen and a list of three to five binary statements on the right. The candidate's job is to evaluate every statement and decide whether the table supports it. The answer key records either True/False, Yes/No, or a similar binary, and the candidate's score on the item is the count of statements answered correctly. A statement judged correctly counts; a statement judged incorrectly counts against the candidate, but the partial-credit logic in Data Insights treats each statement as a separately scored micro-decision. That structure is the single most important fact about the item family, because it changes how you should pace and how you should triage.

The table itself typically contains between 12 and 25 rows and between 3 and 6 columns. The columns are almost always sortable; clicking a column header reorders the rows by that variable. Numeric columns sort ascending or descending depending on the click, and string columns sort alphabetically. The candidate can also type values into a small filter box beneath each column, which is what makes the question feel like a mini spreadsheet. Statements will refer to the table by vocabulary that is not always identical to the column header, so a candidate who skims the headers and jumps to the statements will misread at least one stem per item. Slow down on the column vocabulary first. Re-read every column header out loud or paraphrase it in a margin note. Then read the statements.

The statements themselves come in three structural shapes. A statement may ask whether a particular value meets a threshold, whether a count of rows satisfies a condition, or whether a comparison between two subsets of rows is true. The threshold shape tests whether you can locate one number. The count shape tests whether you can filter the table and tally. The comparison shape tests whether you can sort the table and then read the top or bottom of the ordered list. Most Table Analysis items contain at least two of these three shapes, and the hardest items mix all three inside a single statement, which is where candidates lose the most time.

How Table Analysis is scored inside Data Insights

Table Analysis contributes to the Data Insights scaled score, which on the GMAT Focus runs on a band that places most successful candidates in the 60 to 90 range. Within Data Insights, the four question families are not reported as sub-scores; the candidate receives a single composite. TestPrep İstanbul's diagnostic work suggests that a candidate aiming for a composite in the mid-80s should treat Table Analysis as a non-negotiable family: it is the family with the smallest vocabulary gap, the most mechanical solution path, and the highest floor. A candidate who masters Table Analysis typically protects 6 to 8 raw points in the section, which is enough to absorb one bad Data Sufficiency item and one bad Multi-Source Reasoning item without falling out of the target band.

The internal scoring rule is forgiving in one way and unforgiving in another. Forgiving: a statement that you judge incorrectly only affects that single statement, not the whole item. If an item has five statements and you misjudge one, you can still earn credit on the other four. Unforgiving: the scoring is binary at the statement level. There is no partial credit for "mostly true." If a statement says "Company X had revenue greater than 4.2 million in every quarter from Q2 through Q4" and Company X had revenue of 4.1 million in Q3, the answer is False, and any reasoning that arrives at True is wrong in full. This is why the protocol below insists on a literal reading of every quantifier inside every statement.

Two tactical points follow from the scoring rule. First, do not skip a statement because the first one was hard. The four remaining statements on the same item are usually easier than the first, and they are scored independently. Second, when you are stuck between two answers on a single statement, default to the answer that requires the smaller number of mental operations. The Data Insights scoring does not reward clever reasoning, only correct judgment. A 30-second read that yields a defensible answer is worth more than a 90-second reconstruction that yields the same answer.

The three core moves: sort, filter, ratio

Every Table Analysis statement reduces to one of three core moves, and identifying the move tells you which table interaction to use first. The moves are sort, filter, and ratio. Sort applies when a statement asks about an extreme value, a ranking, or a position in an ordered list. Filter applies when a statement asks about a count of rows that meet a condition. Ratio applies when a statement asks about a relationship between two columns across a defined subset of rows. Memorising the three moves in this exact form is the single most useful preparation step a candidate can take, because the moves map directly onto the table's interaction primitives: sort by column, filter by column value, and compute across columns.

Sort is the most under-used move. Candidates reach for filter by reflex, but filter is wasteful when a statement asks "which row has the maximum value of column C." Sorting column C descending puts the answer in the first visible row. The protocol I teach is: if the statement contains a superlative (largest, smallest, highest, lowest, most, fewest, earliest, latest) the move is sort. Click the relevant column header. Read the first row. Move on. The whole interaction takes fewer than 20 seconds for a sortable column with 20 rows. The same question attempted by scanning the table visually takes 60 to 90 seconds, and the error rate is higher because the eye picks the wrong row when two rows are close in value.

Filter is the move for any statement that asks how many rows meet a condition, what the sum of a column is over a subset, or whether any row in a subset satisfies a property. The filter box beneath each column accepts a numeric range or a categorical match, and the table updates instantly. The protocol is: identify the conditions in the statement, enter each condition into its column's filter, and read the filtered table. Be careful with conjunction. A statement of the form "rows where X is greater than 100 and Y is less than 50" requires both filters to be active simultaneously, and the table's row count must be read after both filters are applied. Applying only one filter and then mentally excluding the wrong rows is a common error pattern.

Ratio is the move for any statement that compares two columns or two derived quantities. Statements like "the ratio of column A to column B is greater than 2.0 for the majority of rows" require no sort and no filter; they require a small number of explicit computations. The protocol is: identify the columns, identify the comparison operator, identify the subset (which may be the whole table), and compute the comparison on enough rows to make a defensible judgment. For a statement about the "majority of rows," you need to look at a sample of five to seven rows and count. For a statement about "every row," you need to scan the entire filtered set. The sample size matters, and a candidate who computes on only one row will misjudge a statement about the majority.

Reading the statement vocabulary before touching the table

The most expensive error a candidate can make on Table Analysis is to read the statement loosely. Statements are written in a vocabulary that is paraphrased from the column headers, and the paraphrase is the trap. A column labelled "Net Revenue (USD mn)" will appear in a statement as "after-tax profit," or "total earnings," or "gross margin," and the candidate who assumes those phrases mean the same thing will answer at least one statement wrong on most items. The fix is a 15-second pass in which you underline every noun phrase in the statement and map it to a column header. If a noun phrase does not map, you have either missed a derived column (sum, ratio, difference) or you have misread the statement.

Quantifiers inside statements deserve the same treatment. Words like "all," "every," "some," "most," "few," "none," "exactly," "at least," and "at most" are not interchangeable, and a statement that flips from "most" to "all" moves from a defensible judgment to a brittle one. In practice, the safe move is to treat "all" and "every" as a demand for a full scan, "most" as a demand for a sample of at least five rows, and "some" as a demand for one matching row. A candidate who reads "most" as "all" will over-filter and may conclude False when the answer is True. A candidate who reads "some" as "most" will under-filter and may conclude True when the answer is False. The fix is mechanical: highlight the quantifier, decide which scan level it demands, and apply that level.

Temporal vocabulary is the second trap. Statements that refer to "the most recent year," "the earliest quarter," or "year-over-year change" require the candidate to identify the time axis in the table and sort by it before answering. A candidate who treats "the most recent year" as "the row at the bottom" will misread on tables where rows are not in chronological order. The protocol is: before answering any statement with a temporal adjective, sort the table by the time column and confirm the top and bottom of the ordered list. The sort costs fewer than 10 seconds, and it prevents an entire class of errors.

Worked example: a four-statement Table Analysis item

Consider a table of 18 rows, each row a public company, with columns for Sector (Technology, Industrials, Consumer, Healthcare, Financials), Employees (numeric), Revenue (USD bn, numeric), R&D Spend (USD bn, numeric), and Founded (year). The four statements are: (1) "More than half of the Technology companies were founded after 2000." (2) "The company with the largest R&D Spend has more than 100,000 employees." (3) "The ratio of R&D Spend to Revenue is greater than 0.20 for at least three companies." (4) "No Industrials company has Revenue below 5." The candidate has 2 minutes 30 seconds for the item. The protocol below shows how to spend that time.

First, 15 seconds to read every column header out loud and confirm the units. Revenue is in USD bn, R&D Spend is in USD bn, Employees is a count, Founded is a year. Second, 10 seconds to read all four statements and underline the quantifiers and the column references. Statement 1: "more than half" of Technology, "founded after 2000." Statement 2: "largest R&D Spend," "more than 100,000 employees." Statement 3: "ratio of R&D Spend to Revenue," "greater than 0.20," "at least three." Statement 4: "no Industrials," "Revenue below 5." Third, sort the table by Sector to count the Technology rows (filter move) and the Industrials rows. Fourth, execute each statement with the move the quantifier demands.

Statement 1: filter Sector equals Technology, sort Founded ascending, count the rows with Founded greater than 2000, and compare to half the Technology count. If the Technology subset is 4 rows and 3 of them were founded after 2000, the answer is True. Statement 2: sort R&D Spend descending, read the first row, read its Employees value, compare to 100,000. Statement 3: filter the whole table, scan the R&D Spend and Revenue columns, and compute the ratio for the rows where the ratio looks close to 0.20. A defensible judgment on "at least three" requires a count of three or more, so the candidate should confirm at least three rows satisfy the ratio. Statement 4: filter Sector equals Industrials, scan the Revenue column, confirm no row is below 5. If any row is below 5, the answer is False. The whole sequence should take under 2 minutes for a candidate who has practised the moves.

The diagnostic value of a worked example is to make the move-to-statement mapping visible. Statement 1 is a count move (filter then count). Statement 2 is a sort move (sort then read). Statement 3 is a ratio move (filter then compute). Statement 4 is a filter move (filter then scan). A candidate who internalises the mapping will not waste time choosing the wrong interaction, which is the most common time sink on Table Analysis.

Common pitfalls and how to avoid them

Five errors account for the majority of Table Analysis mistakes. The first is vocabulary mismatch: the statement uses a paraphrase of the column header and the candidate assumes it means the same thing. The fix is the 15-second mapping pass before any table interaction. The second is quantifier drift: the candidate reads "most" as "all" or "some" as "most." The fix is highlighting the quantifier and matching it to a scan level. The third is filter sequencing: the candidate applies one filter and mentally excludes the rest of the rows, missing a row that satisfies the statement's conjunction. The fix is to apply every filter the statement demands before reading any row. The fourth is sort direction: the candidate sorts descending when the statement asks for the smallest value, and reads the wrong row. The fix is to confirm the direction of the sort before reading the first or last row. The fifth is rounding: a ratio of 0.198 is not greater than 0.20, and a Revenue of 4.99 is below 5. The fix is to treat every threshold as strict unless the statement says "at least" or "approximately."

Two additional pitfalls are time-related. The first is over-investment: a candidate spends 90 seconds on one statement and rushes the remaining statements, picking answers on incomplete reasoning. The protocol is to cap any single statement at 45 seconds, mark it for revisit, and finish the item. A 30-second revisit at the end of the item is more productive than a 90-second struggle in the middle. The second is under-reading: a candidate reads only the first row of a sorted table and assumes the rest follow the same pattern. On a sortable column, the first row is the only row whose position is guaranteed, and the candidate must read at least the first two or three rows to confirm a statement about a subset.

The pitfall block is worth treating as a checklist. Before answering a statement, the candidate should ask: have I mapped the vocabulary? Have I identified the quantifier? Have I applied every filter? Have I confirmed the sort direction? Have I checked the threshold for strictness? A candidate who runs this five-question checklist on every statement will eliminate the most common Table Analysis errors, and the time cost is roughly 5 seconds per statement, which is recovered many times over by the reduction in misreads.

A six-week preparation plan for Table Analysis mastery

For a candidate whose Table Analysis accuracy is below 80% on practice items, a six-week plan is sufficient to reach a 90% accuracy band, which is the threshold at which Data Insights scoring becomes predictable. The plan divides into three two-week blocks: vocabulary, move discipline, and pacing. In the first two weeks, the candidate should solve 20 to 25 Table Analysis items with no time pressure, annotating every statement with the column mapping and the quantifier. The goal of this block is not accuracy; it is to make the mapping pass automatic. In the second two weeks, the candidate should solve the same 20 to 25 items under a soft time cap of 2 minutes 30 seconds per item, focusing on move selection. In the third two weeks, the candidate should solve 30 fresh items under a hard time cap of 2 minutes 15 seconds per item, and review every wrong statement with the five-question checklist.

Weekly targets should be set in terms of accuracy bands, not raw question counts. A candidate in the mid-60s on Data Insights should target 75% Table Analysis accuracy by the end of week two, 85% by the end of week four, and 90% by the end of week six. The candidate should also track the time per statement, not just the time per item. A 2 minutes 30 seconds item with four statements is 37 seconds per statement on average, and a candidate who is over 50 seconds on any single statement is signalling a move-selection error. The fix is to revisit the move-to-statement mapping for that statement type and drill five fresh statements of the same shape.

For candidates at a higher starting point, the same six-week structure applies, but the per-block targets shift. A candidate in the mid-70s on Data Insights should target 85% Table Analysis accuracy by the end of week two, 92% by the end of week four, and 95% by the end of week six. The third block should also include mixed Data Insights items, where Table Analysis is interleaved with Data Sufficiency, Multi-Source Reasoning, and Graphics Interpretation, to confirm that the move discipline survives a section-level workload. The aim at this level is to make Table Analysis a guaranteed floor, freeing mental bandwidth for the harder families.

Comparative table: the four Data Insights question families

The table below summarises how Table Analysis differs from the other three Data Insights families on the dimensions that matter for preparation: the dominant move, the typical time budget, the most common error pattern, and the floor a candidate can reliably protect with disciplined practice. The numbers are typical bands for a candidate working at the 75 to 85 Data Insights level, and they should be read as planning targets, not guarantees.

Reading the table row by row, Table Analysis has the highest accuracy floor of the four families when practiced with discipline, and the second-lowest time budget. Data Sufficiency has a similar time budget but a lower floor, because the dominant error (over-solving) is harder to detect in review. Multi-Source Reasoning has the highest time budget and the lowest floor, because the candidate must integrate information across multiple tabs. Graphics Interpretation sits between the two on both dimensions. For a candidate building a Data Insights preparation plan, the implication is that Table Analysis should be the first family mastered, because it offers the most accuracy per minute of practice, and mastering it frees mental bandwidth for the harder families later in the plan.

Conclusion and next steps for the candidate

Table Analysis is the Data Insights family that rewards protocol over talent. The protocol is: read the column headers first, map the statement vocabulary to the headers, identify the quantifier, choose the move (sort, filter, or ratio), execute the move, and check the threshold for strictness. A candidate who runs this sequence on every statement will protect a high accuracy floor, and the floor is what determines whether a Data Insights composite lands in the 70s, the 80s, or the 90s. The six-week plan above is a working template; the per-week targets should be adjusted to the candidate's starting band, and the move discipline should be drilled until it is automatic.

The natural next step for a candidate who has finished this article and wants to convert the protocol into score movement is a focused Table Analysis diagnostic: 12 items, timed at 2 minutes 15 seconds each, with every wrong statement reviewed against the five-question checklist. TestPrep İstanbul's Table Analysis diagnostic is built around that exact protocol and is a natural starting point for candidates building a sharper preparation plan on this family.

Frequently asked questions