Why candidates who finish GMAT Data Insights early usually score lower

The GMAT Data Insights section asks candidates to answer 20 prompts in 45 minutes, which works out to 2 minutes and 15 seconds per prompt on average, but the actual clock pressure is much more uneven than that arithmetic suggests. Some prompts reward a 90-second solve; others demand a full 3 minutes, and a small group of multi-part stems quietly swallow five or six minutes if a candidate treats every option as equally urgent. The candidates who score in the high 70s and 80s on the GMAT Focus edition are almost never the ones who finish fastest. They are the ones who allocate clock time by prompt family, recognise when a question is not worth a second pass, and protect a small emergency reserve for the last three prompts. This article walks through the timing architecture that turns a chaotic 45-minute sprint into a controllable sequence of micro-decisions, and explains how to build that architecture across a 6-to-10-week preparation cycle.

The clock architecture of GMAT Data Insights: why a flat 2:15 budget fails

Most candidates arrive at the GMAT Data Insights section carrying a single timing rule in their head: 2 minutes and 15 seconds per prompt, derived from dividing 45 minutes by 20. That arithmetic is correct, and it is also the reason so many test-takers run out of clock in prompt 17. A flat per-question budget assumes the 20 prompts are interchangeable, and the section is engineered specifically to punish that assumption. The GMAT Focus test writers deliberately interleave quick Graphic Interpretation prompts with slow Multi-Source Reasoning tab pairs, drop a Data Sufficiency stem next to a Table Analysis prompt that demands genuine numerical work, and occasionally position a Two-Part Analysis prompt near the end of the section where fatigue has already eroded accuracy. A flat budget cannot absorb that variance, so the candidate ends up borrowing time from prompts 18, 19, and 20 to repay a debt accumulated on prompts 4, 5, and 12.

The right way to think about the section is as five micro-segments of four prompts each, with each micro-segment holding its own timing target. The first four prompts should be a warm-up: the test taker is settling into the rhythm, the screen, and the on-screen calculator, and the GMAT almost always opens with a slightly easier cluster. The middle two micro-segments are where the bulk of the score is won or lost, and the final four prompts are where pacing discipline is tested hardest. A candidate who plans for this shape rather than for 20 independent prompts will rarely run out of clock, because the plan is built around the variance the section is designed to produce.

A second reason flat budgets fail is that they confuse effort with return. Spending an extra 45 seconds on a hard Multi-Source Reasoning prompt can move a candidate from a guess at 50/50 to a confident answer; spending an extra 45 seconds on a Graphic Interpretation prompt that is already solved rarely changes the answer at all. A timing protocol that asks "how much clock is this prompt worth?" rather than "how much clock does this prompt type usually take?" produces a higher score at the same overall pace. The sections that follow break that question down prompt family by prompt family.

Prompt-family minute budgets: a worked table for all five item types

GMAT Data Insights contains five prompt families, and each one carries a different expected solve time, a different cost of a wrong first pass, and a different threshold beyond which further work is wasted. The table below is the timing map I teach, and it should be internalised before a candidate sits a single timed section. Numbers are anchored to the published 45-minute, 20-prompt structure and assume a target score in the 80 to 84 band; a 76-scoring candidate should add roughly 10 seconds to each ceiling, and an 88-scoring candidate should subtract 10.

The soft target is the time a candidate should aim to spend before committing to a first answer. The hard ceiling is the time beyond which the candidate should flag the prompt, pick a defensible answer, mark it, and move on. The typical count column is a planning anchor, not a guarantee: the GMAT Focus test adapts, so a candidate should expect the distribution to flex by one prompt in either direction. The cost-of-extra-minute column is the most important column for pacing decisions, because it tells the candidate which prompts reward a long solve and which ones punish it.

Data Sufficiency and Graphic Interpretation are the two families where a careful first pass almost always pays off, because the structure of the prompt reveals the answer pathway quickly. Table Analysis and Multi-Source Reasoning are the two families where a second pass is often needed, so the soft targets sit closer to the ceilings and the cost of an extra minute is correspondingly high. Two-Part Analysis is its own category: a single prompt carries two correct answers, and the screen layout tempts candidates into two independent solves. The minute budget for Two-Part must include the second solve, which is why the soft target starts at 2:30 even though the surface question looks similar to a Data Sufficiency stem.

Three tactical notes on using this table. First, the ceilings are not aspirations, they are guardrails: a candidate who hits 2:40 on a Data Sufficiency prompt and still has not isolated a statement should flag and move. Second, the soft target on Multi-Source Reasoning is a total across the tab pair, not per tab, because the tabs are designed to be read together. Third, on Table Analysis, the soft target assumes the candidate has already located the relevant rows in the data; candidates who skim the table twice in a row should treat 2:00 as the upper bound of the first pass, not the lower bound.

The three-pass protocol: open, decide, recover

Every prompt in Data Insights should be handled with the same three-pass micro-protocol, regardless of family. The protocol has nothing to do with reading the question carefully the first time, which is a habit that wastes clock; it has everything to do with making the clock a tool rather than a threat. Pass one is the open pass: read the stem, scan the data, form a one-sentence prediction of the answer shape, and either commit or flag. Pass two is the decide pass: if the prompt was flagged, return to it with the remaining clock budget visible on screen, and spend the hard ceiling, no more. Pass three is the recover pass: a 60-second sweep of flagged prompts in the final three minutes of the section, used to convert careless flags into confident answers or, when recovery is impossible, into a strategic guess anchored to the strongest remaining option.

The open pass is where most candidates lose clock. They read the entire stem, scan the entire data layer, identify the question type, and only then start solving. That sequence is correct for the first two prompts of the section; it is wasteful by prompt eight, when the candidate has already seen three Data Sufficiency stems and two Table Analysis prompts and knows the formats cold. The faster version of the open pass is: read the last two lines of the stem (which always contain the actual question), glance at the data layer for the variable names or categories referenced, and either commit or flag within 30 seconds. For most prompt families, that 30-second open pass is enough to separate the prompts that will yield to a 90-second solve from the prompts that will not.

The decide pass is where minute budgets become real. When a candidate returns to a flagged prompt, the screen still shows the original stem and data, but the cognitive state is different: the candidate has already failed the open pass once, so the decide pass must be a different kind of work. On a Multi-Source Reasoning flag, the decide pass usually means re-reading the second tab with the first question in mind, not re-reading the first tab. On a Data Sufficiency flag, the decide pass usually means testing statement (2) before statement (1), because the GMAT places the easier statement first about 60% of the time and the decide pass is the candidate's last chance to exploit that pattern. On a Two-Part Analysis flag, the decide pass means looking for a shared variable that constrains both answer slots simultaneously; if no such variable is visible, the candidate should commit to a paired guess and move.

The recover pass is the part most candidates skip, which is one of the main reasons 78-scoring candidates stay at 78 instead of climbing to 84. The recover pass takes the final 60 to 90 seconds of the section and turns them into a checkpoint: a sweep of every flagged prompt, a quick re-read of the stem, and a commit. The recover pass does not solve hard prompts. It converts panicked guesses into reasoned guesses, and it prevents the candidate from leaving two or three high-value prompts unanswered because the clock ran out mid-flag. A 78 scorer who runs the recover pass reliably tends to land at 81 or 82 within a single practice block; the score gain comes almost entirely from flagged prompts that were actually solvable once the candidate's nervous system had 60 seconds to settle.

Why the last four prompts behave differently: a tactical sub-protocol

The final four prompts of GMAT Data Insights are not the same animal as the first 16. Fatigue is real, the on-screen calculator is now familiar, and the test writers know that clock pressure is at its peak, so they tend to place prompts in this segment that are slightly easier on the surface but psychologically heavier. A 2-part analysis prompt at prompt 18 feels harder than the same prompt at prompt 8, even though the math is identical, because the candidate is watching the clock and reading the screen faster. A candidate who treats the last four prompts with the same flat budget they used for the first 16 will under-perform on those prompts by 1 to 3 points of accuracy, which is the difference between a 78 and an 81.

The tactical sub-protocol for the last four prompts is simple: shift the minute budget down by 15 seconds on the soft target and up by 20 seconds on the hard ceiling, and add a single explicit checkpoint at the 9-minute-to-go mark. The checkpoint is a 5-second glance at the clock and a count of how many prompts are still flagged. If three or more prompts are flagged with 9 minutes remaining, the candidate should drop the soft target on the next prompt by 30 seconds and accept a slightly lower confidence answer in exchange for clearing the backlog. If zero or one prompt is flagged, the candidate can spend the soft target freely because the section is now under control.

For most candidates, the psychological effect of that 9-minute checkpoint is larger than the time-management effect. Knowing that a checkpoint is coming converts the final four prompts from a sprint into a sequence, and the candidate stops reading the clock on every prompt. A 60-percentile scorer who installs the checkpoint reliably tends to add 2 points of accuracy on the final four prompts within two weeks, because the checkpoint prevents the panic-reread loop that costs 30 to 40 seconds per prompt once anxiety spikes.

One more tactical note: the last prompt of the section is almost always solvable in 90 seconds if the candidate reads it calmly, but it is also the prompt most often left blank by candidates who mismanage the section. Treat prompt 20 as a free 90-second solve, not as a rushed 45-second guess. The GMAT rarely puts the hardest prompt of the section at position 20, and even when it does, a calm 90-second attempt is worth more than a panicked 30-second guess. If the candidate has followed the protocol, prompt 20 should arrive with 90 to 120 seconds on the clock, which is exactly enough to read the stem, scan the data, and commit to an answer.

Common pitfalls and how to avoid them

Five timing pitfalls account for the majority of clock-related score loss on GMAT Data Insights. None of them are visible until the candidate starts a timed section, and most of them are invisible to the candidate even then, which is why I ask every diagnostic student to log their solve time per prompt for the first three practice sections before I look at their accuracy. The clock tells the truth before the answer key does.

• The re-read loop. A candidate reads the stem, scans the data, forms a tentative answer, then re-reads the stem to check it. The re-read costs 30 to 45 seconds and almost never changes the answer on a well-trained candidate's first read. The fix is a one-line prediction discipline: before scanning the data, write a single sentence describing the answer shape you expect. If the prediction and the data agree, commit. If they disagree, scan again, not the stem.
• The skip-and-stare. A candidate opens a prompt, decides it looks hard, and stares at the screen for 20 to 30 seconds trying to decide whether to attempt it. The stare burns clock without producing information. The fix is the 15-second rule: if a prompt has not yielded a starting move within 15 seconds, flag it and move. The 15-second rule sounds aggressive, and on a 90-prompt practice test it costs about 5 minutes of flag time. That flag time is recovered twice over in the decide pass, where the prompt is now approached with a clear head and a visible clock.
• The perfect-answer trap. A candidate commits to an answer, sees another answer that also looks defensible, and spends 60 to 90 seconds comparing the two. The comparison is rarely productive, because the comparison is happening on a stem the candidate has already solved. The fix is the 10-second rule: if a committed answer is challenged within 10 seconds of being marked, the candidate may re-evaluate; if more than 10 seconds have passed, the committed answer stands and the candidate moves on. The 10-second rule protects accuracy without inviting second-guessing.
• The calculator detour. A candidate reaches for the on-screen calculator on a prompt that does not require numerical work. The detour is small per prompt (10 to 20 seconds) but it happens on 30 to 40% of prompts for some candidates, which is a 5 to 8-minute section-level cost. The fix is a hand-on-mouse discipline: the cursor stays away from the calculator button until the candidate has decided that arithmetic is required. The decision to use the calculator is a separate cognitive step from the act of using it, and the discipline preserves the step.
• The end-of-section panic. A candidate sees 5 minutes remaining and 4 prompts unfinished, and reads faster and less accurately. The panic costs 1 to 2 prompts of accuracy in the final segment. The fix is the 9-minute checkpoint described in the previous section, which surfaces the backlog 4 minutes before panic sets in and gives the candidate room to act.

Building the timing protocol across a 6-week preparation cycle

A timing protocol is not installed by reading about it. It is installed by running it on a sequence of timed practice sections and adjusting the per-prompt-family ceilings until the soft targets feel natural and the hard ceilings feel like guardrails rather than threats. Most candidates I work with need 6 to 8 weeks of structured practice before the protocol becomes automatic, and the cycle below is the one I assign to a candidate whose diagnostic lands between 76 and 80 on Data Insights.

Weeks 1 and 2 are diagnostic and family isolation. The candidate takes one full timed section under realistic conditions, logs per-prompt solve time, and tags each prompt by family. The first week's data usually reveals one or two families where the candidate is over-budgeting (Multi-Source Reasoning is the most common offender) and one or two where the candidate is under-budgeting (Graphic Interpretation is the most common). Weeks 1 and 2 are spent on family-specific drills: 10 prompts of one family, timed, with a hard ceiling 15 seconds below the section-level soft target. The drills build muscle memory for the per-family pace.

Weeks 3 and 4 are protocol installation. The candidate takes two timed sections per week, runs the three-pass protocol explicitly (writing "open / decide / recover" on the scratch pad for the first two sections, then dropping the labels by section three), and logs both per-prompt solve time and per-prompt confidence. Confidence logging is the hidden tool of weeks 3 and 4: it forces the candidate to distinguish between prompts solved quickly with high confidence, prompts solved quickly with low confidence, and prompts that took the full budget. The third category is where the next 5 points of score live.

Weeks 5 and 6 are consolidation and edge-case handling. The candidate takes one full timed section per week, runs the protocol without labels, and reviews the previous week's section in detail. The review focuses on prompts that were flagged and recovered, prompts that were committed and wrong, and prompts where the soft target was exceeded by more than 20 seconds. By week 6, the candidate's per-family distribution of solve times should cluster within 10 seconds of the soft targets in the table above, and the hard ceilings should trigger fewer than 3 times per section. A candidate who reaches that state is ready for the live test, because the protocol is no longer something the candidate runs consciously; it is something the candidate's hands run on the keyboard while the brain handles the prompts.

One final tactical note on the cycle. The diagnostic section in week 1 should be taken cold, with no warm-up, on a day the candidate did not study the day before. The cold start reveals the candidate's true baseline pacing, and the baseline is what the rest of the cycle adjusts. A candidate who warms up before the diagnostic will over-pace by 15 to 25 seconds per prompt, which understates the family-level soft targets and produces a protocol that is too aggressive for the live test. The 45 minutes of the diagnostic is the only honest 45 minutes the candidate will have before the real test, and it is worth protecting.

Adapting the protocol to a non-native English reader

Non-native English readers face a specific timing problem on GMAT Data Insights that native readers do not: the stems and the on-screen data layer both carry English-language load, and that load is not evenly distributed across the five prompt families. A Table Analysis prompt with a 12-row table and 8 columns of mostly numeric data is light on English; a Two-Part Analysis prompt with a 6-line scenario and a two-column table is heavy on English. The protocol above works for every reader, but the minute budgets need to flex by family for a candidate whose English throughput is below the section's median.

For a non-native reader, the soft target on Table Analysis should drop by 10 to 15 seconds (because the data layer carries less English weight) and the soft target on Two-Part Analysis should rise by 20 to 30 seconds (because the scenario carries more). Multi-Source Reasoning is the family where this adjustment matters most: the three tabs of a Multi-Source prompt are heavy on prose, and a non-native reader who treats the soft target as a fixed 2:30 will under-perform on the second tab, where the prose density is highest. A 2:45 to 3:00 soft target on Multi-Source is realistic for a non-native reader, and the additional 30 seconds is recovered on Table Analysis, where the data layer is friendlier.

Graphic Interpretation is the family where the adjustment is smallest, because the data layer is visual rather than textual. A 1:40 soft target is achievable for most non-native readers, provided the candidate has practiced the on-screen calculator enough that the cursor-to-calculator move is automatic. The 10-second calculator detour described in the pitfalls section is the single largest hidden cost for non-native readers, and a week of dedicated calculator drills usually pays for itself twice over across the live section.

Conclusion and next steps

A timing protocol on the GMAT Data Insights section is not a clock-watching exercise. It is a per-prompt-family minute budget enforced by a three-pass micro-protocol, with a checkpoint in the final nine minutes that converts a sprint into a sequence. The protocol protects accuracy on the prompts that reward it, prevents clock loss on the prompts that punish it, and converts flagged prompts from a liability into a recoverable asset. A candidate who installs the protocol over a 6-week cycle typically gains 4 to 7 points on Data Insights, and the gain is stable across retakes because the protocol is procedural rather than content-dependent. TestPrep İstanbul's per-prompt-family timing diagnostic is a useful starting point for candidates who want to map their current distribution against the soft targets in the table above, and to identify the two or three families where the minute budget is leaking the most accuracy.

Frequently asked questions