# GMAT Data Insights: Score 80+ with Better Timing | topin

On this page

## What each DI score means in the latest official percentiles

The GMAT (formerly GMAT Focus Edition) reports Data Insights scores from 60 to 90 in one-point steps. GMAC’s official percentile data, labelled 2021–2026 and checked on 4 October 2026, shows DI percentiles as whole numbers. The full DI table, alongside Quant, Verbal and total percentiles, is on our [GMAT percentiles page](/articles/gmat-percentiles); this guide is about reaching 80+.

The steps around an 80+ target are substantial: DI78 is at 69, DI79 at 76, DI80 at 83, DI81 at 88 and DI82 at 93\. DI75 is now at 46, not the 47 still quoted in some guides, and DI85 is at 98\. Those changes in relative rank help explain why one scaled point can feel meaningful, but they do not reveal how many extra correct answers you need.

A score of 80 is not 80% accuracy. The percentile is also not a percentage of questions correct or a prediction of MBA admission. Your task is to improve the performance that generates the section score, not to reproduce a raw-accuracy number from an online anecdote.

DI contributes equally with Quant and Verbal to the [current total score](/articles/gmat-score-converter). Treat it as a full scoring priority when setting your study schedule. If it is your weakest section, practising only Quant because you consider DI “mostly maths” leaves information interpretation and multi-part decision-making unaddressed.

## Why there is no reliable “80 equals this many mistakes” rule

GMAC describes section scoring in terms of the questions answered, correct responses, question difficulty and other question parameters. That is not a fixed count-to-score chart. Two practice attempts with the same number of errors do not provide enough information to infer that they should receive the same official DI score.

For questions with multiple parts, the official scoring description requires all responses to be correct for credit. That makes partial confidence a poor stopping signal. You may understand the first statement and still lose credit because you misread the second statement’s scope. In review, distinguish a completely wrong method from an otherwise sound solution with one unchecked response.

Avoid trying to identify question difficulty from how much text is displayed. A long source can lead to a narrow lookup, while a short sufficiency problem may require careful logical reasoning. Guessing that the adaptive system is “punishing” you after one unfamiliar item distracts from the question you can actually answer.

Track decisions you control: the information selected, the conclusion supported, the time spent and the final responses entered. Your error record should describe what you did and what a better action would have been. “DI is unpredictable” is not a useful diagnostic. “I compared totals when the question asked for rates” identifies something you can practise and later verify under timing.

## Diagnose the obstacle before increasing practice volume

Use a fresh timed section or full mock, then review every wrong answer and every right answer that consumed disproportionate time. A correct answer reached through six minutes of trial and error can still explain why later questions were rushed. Separate reasoning quality from the cost of reaching the answer.

Classify each problem into a small number of actionable causes. Did you choose the wrong source? Did you overlook a unit or population? Did you infer something the data did not establish? Did you perform the wrong calculation correctly? Did you know the method but delay committing to it? These distinctions stop you prescribing more arithmetic practice for a reading or navigation problem.

For each mistake, write one sentence beginning “Next time I will…”, followed by an observable action. Examples include checking whether values are annual or monthly, writing the comparison denominator before calculating, or testing whether a statement establishes uniqueness rather than merely one possible value. Avoid vague notes such as “be more careful”.

Then redo the reasoning untimed without copying the explanation. Explain why the rejected answer fails. Finally, use a fresh problem to check whether the improved action transfers. Remembering the answer to a reviewed question proves less than selecting the right information in unfamiliar material. Keep that difference visible when you decide whether the weakness is improving.

* Selection error: wrong tab, row, variable or relevant population.
* Interpretation error: conclusion stronger than the evidence permits.
* Calculation error: wrong denominator, unit, sign or operation.
* Response error: one part unchecked or an answer entered incorrectly.
* Timing error: too much exploration before choosing a method.

## A 45-minute pacing plan that leaves time for review

The official DI section contains 20 questions in 45 minutes, giving an overall arithmetic average of two minutes and 15 seconds per question. That average includes any end-of-section review. It is a budget, not an instruction to spend exactly the same time on every item. Information-heavy sets can require an initial investment that later questions partly share.

The example below reserves three minutes for review and allocates 42 minutes to first-pass answers, an average of two minutes and six seconds. These are our suggested checkpoints, not GMAC’s pacing rules. Test whether they work for you in practice. A reserve that consistently forces rushed final answers should be adjusted; protecting completion is more useful than meeting an arbitrary spreadsheet schedule.

At each five-question checkpoint, look at the remaining time and the cause of any delay. If you are behind, simplify the next decision: identify exactly what is being asked, select the necessary information and choose an efficient method. Do not try to recover several minutes by rushing every easy question, which can turn one slow problem into several preventable errors.

Use a decision point on individual questions too. If you have spent substantial time and still cannot state a viable method, make the best available choice and bookmark it. That is a practice strategy for preserving the section budget, not a claim that any particular guessed question has a known score cost. Record whether the decision helped you reach all 20 questions.

__Illustrative 42-minute first pass plus 3-minute review, based on the official 20-question/45-minute format; last checked 4 October 2026__
| After answering | Suggested elapsed time | Suggested time remaining |
| --------------- | ---------------------- | ------------------------ |
| Question 5      | 10:30                  | 34:30                    |
| Question 10     | 21:00                  | 24:00                    |
| Question 15     | 31:30                  | 13:30                    |
| Question 20     | 42:00                  | 3:00 for review          |

## Use the calculator after choosing the operation

DI allows the on-screen calculator; Quant does not. GMAC does not allow a personal calculator to replace it. Practise with the permitted workflow so that you can judge when entering a calculation is helpful. The calculator can reduce arithmetic burden, but it cannot select the relevant values or decide what the question means.

Before entering numbers, write or mentally state the operation and units. For percentage growth, identify the original value as the denominator. For a share, identify the whole. For an average, check whether the groups have equal weights. This short pause prevents a fast, precise answer to the wrong calculation.

Use estimation first when it can eliminate options or expose an implausible result. If an increase from 100,000 to 125,000 produces “20% growth”, the denominator deserves checking: 25,000 divided by the original 100,000 is 25%. By contrast, 25,000 is 20% of the final 125,000\. Both calculations are arithmetically sound; only the wording determines which is relevant.

For awkward division or a calculation that needs precision, the calculator can be the efficient choice once the setup is secure. After calculating, check whether the answer should be a proportion, percentage, currency value or quantity. Do not treat every displayed decimal as directly comparable with the answer options. If the problem is about sufficiency rather than a requested numerical value, establish whether the information determines a unique answer before spending time on arithmetic.

__Original arithmetic illustrations: choose the question and denominator before calculating; checked 4 October 2026__
| Task                                     | Setup                         | Result         | Decision to practise                 |
| ---------------------------------------- | ----------------------------- | -------------- | ------------------------------------ |
| Growth from 100,000 to 125,000           | (125,000 − 100,000) ÷ 100,000 | 25%            | Use original amount as denominator   |
| 25,000 as share of 125,000               | 25,000 ÷ 125,000              | 20%            | Use the stated whole                 |
| Weighted average: 20 items at 4, 30 at 6 | (20 × 4 + 30 × 6) ÷ 50        | 5.2            | Weight each group before dividing    |
| Compare 18 of 60 with 25 of 100          | 18 ÷ 60 versus 25 ÷ 100       | 30% versus 25% | Compare rates rather than raw counts |

## Practise the workflow behind each DI question type

The [official five Data Insights types](/gmat/data-insights) are Data Sufficiency, Multi-Source Reasoning, Table Analysis, Graphics Interpretation and Two-Part Analysis. For an 80+ goal, use the type label to diagnose where time or credit is being lost. You do not need five disconnected study plans; you need a repeatable way to select evidence and verify the requested responses.

For [Multi-Source Reasoning](/gmat/data-insights/multi-source-reasoning), practise reading the question before exploring every detail. Locate the source that can answer that particular request, then check whether another tab changes the conclusion. For Table Analysis, inspect headers and units before relying on a sorted row. Sorting changes presentation; it does not change the meaning of the population or measure.

For Graphics Interpretation, identify the axes, legend and units before extracting a value. A plausible visual impression is not a substitute for checking what the plot actually measures. For Two-Part Analysis, keep both requested outputs visible while solving, and check each selected response against its own condition before moving on.

For [Data Sufficiency](/gmat/data-insights/data-sufficiency), focus on whether information determines the answer rather than whether you can produce one convenient example. Practise separating the statements’ individual effects from their combined effect. If your notes show repeated logical errors, slow down in untimed practice until you can explain why the evidence is or is not sufficient; then bring that reasoning back into timed work.

Across all five types, stop reading once you have the necessary evidence and a supported answer. The aim is purposeful selection, not speed-reading every word or memorising a fixed order of clicks. Review examples where extra exploration consumed time without changing the answer.

## Review and edit: earn the opportunity before relying on it

The GMAT review screen appears only after you have answered every question in the section and only if time remains. You can revisit as many questions as time allows, but edit up to three answers per section. A bookmark does not give you extra time or permission to skip an unanswered item and return freely during the first pass.

Bookmark a question for a specific reason. You may want to recheck a calculation, test a logical edge case or verify a response you entered quickly. A vague feeling that the question was difficult is less useful when the review clock is short. If you bookmark nearly everything, choosing what to review becomes another time-consuming task.

Prioritise changes that have a concrete justification. A unit error, a wrong denominator or an overlooked qualifier gives a reason to reconsider. Nervousness alone does not. During practice, record whether each edit moved from wrong to right, right to wrong, or left the result unchanged; that pattern can help you improve your review decisions.

Do not reserve review for completing the original reading of several unsolved questions. Three minutes may support a targeted check but not a second full attempt at every difficult item. Build a first-pass process that produces a considered answer, then use any remaining time to verify the most promising uncertainties. The official edit limit is a ceiling, not a target you must reach on every section.

## A two-week practice cycle for a repeatable 80+ attempt

Use this as a practice cycle rather than a promised score timetable. On the first day, take a fresh timed diagnostic and create the error categories above. Over the next few sessions, work on the two most frequent causes of lost credit or excessive time. Keep some work untimed so that you can explain the reasoning, then use short timed sets to test the improved process.

In the middle of the cycle, practise the pacing checkpoints and calculator decisions together. Record the time remaining after five, ten and fifteen answers, whether you reached the review screen, and why any edit was made. Review the section’s final questions as carefully as its first ones; a repeatable late rush can reveal a budgeting problem rather than a late-question content weakness.

In the second week, take a fresh full-length mock to check whether the DI improvement survives the rest of the exam. The GMAT allows any section order and one optional ten-minute break after the first or second section. Test a sensible order under realistic conditions instead of assuming DI must always be first or last.

Then compare the new performance with the diagnostic. Did the same error category recur? Was the section completed with considered answers? Did your review become more selective? A better scaled score is useful evidence, but it is stronger when it comes with a process improvement you can describe and repeat. If the score did not improve, change the identified cause rather than merely adding another mock.

## Decide when you are ready, using fresh evidence

Do not treat one remembered practice result as proof that 80+ is secure. Look for fresh timed work showing that you can select information correctly, finish the section and use the calculator without turning routine tasks into long detours. Repeated questions can test understanding during review, but they overstate readiness if used as a supposedly fresh score forecast.

Your official report can support this assessment after a real attempt: GMAC provides section performance, question-type and timing insights. Use those details to form a narrower retake plan. “Improve DI” is too broad; “reduce repeated rate-versus-total errors and preserve time after question fifteen” is something you can test.

topin offers GMAT practice questions with explanations and a [free full-length mock](/gmat/practice-test) timed like test day and scored on the official scale. Use explanations to understand why an answer works, then verify the method on unfamiliar questions. Reading an explanation fluently is not the same skill as reaching that explanation’s conclusion under a clock.

Finally, connect DI to your total-score and application goals. An 80+ section target may be appropriate for you, but a particular DI percentile is not a universal school minimum. Build the strongest repeatable performance across all three equally weighted sections, with your chosen programme’s current profile and application deadline providing the context.

## Try a full GMAT mock free

Timed like test day, every section scored, every answer explained.

[Take the free mock ](/gmat/login)

## FAQs

What percentile is GMAT Data Insights 80?

DI80 is the 83rd percentile in GMAC’s official percentile data, labelled 2021–2026\. DI81 is 88 and DI82 is 93.

Is Data Insights 75 the 47th percentile?

No longer. GMAC’s official data, checked on 4 October 2026, puts DI75 at the 46th percentile.

How many mistakes can I make and still score DI80?

There is no official fixed error-count conversion. Scoring accounts for performance and question characteristics, so the number wrong alone does not determine the score.

How much time do I get per DI question?

The section has 20 questions in 45 minutes, an overall average of 2 minutes 15 seconds each. Any review time must come from that same 45-minute budget.

Can I use a calculator in Data Insights?

Yes, the on-screen calculator is allowed in DI. You cannot use a personal calculator, and the Quant section does not permit the calculator.

Can I skip a DI question and return later?

You must answer all questions before the review screen becomes available, with time remaining. Bookmark uncertain answers and use the end-of-section review if you reach it.

How many DI answers can I change?

You can edit up to three answers in the section during review. You can revisit more questions if time allows.

Does DI count towards the GMAT total?

Yes. Data Insights, Quantitative Reasoning and Verbal Reasoning contribute equally to the current total score.

## Sources (checked 4 October 2026)

* [GMAC: DI percentiles, scoring and section content](https://www.gmac.com/gmat-other-assessments/about-the-gmat-focus-edition/exam-scores)
* [mba.com: official question counts, timing and review/edit rules](https://www.mba.com/exams/gmat-exam/about/exam-structure)
* [mba.com: calculator and note-taking policies](https://support.mba.com/hc/en-us/articles/14076264362907-GMAT-Calculator-and-Scratch-Paper-Note-Taking-Policies)
* [mba.com: official score-report diagnostics](https://www.mba.com/exams/gmat-exam/scores/official-score-reports)

## Related articles

[All GMAT study guides ](/gmat)

* [GMATScores & percentilesGMAT percentilesCheck the latest official GMAT percentiles for 2026–27: all total and section scores, the August 2026 update, and what your ranking means for MBA targets.Updated 4 Oct 2026](/articles/gmat-percentiles)
* [GMATScores & percentilesGMAT score conversionConvert every classic GMAT score from 200–800 to the current 205–805 scale using GMAC’s August 2026 concordance, with target ranges and worked examples.Updated 4 Oct 2026](/articles/gmat-score-converter)
* [GMATAdmissionsISB and IIM GMAT targetsCompare official current-scale GMAT profiles for ISB and IIM Ahmedabad PGPX, with dated INR fees, GST treatment, centre-only test rules and score targets.Updated 4 Oct 2026](/articles/gmat-score-isb-iim)
* [GMATAdmissionsM7 and top-20 GMAT scoresSee official current-scale GMAT scores for M7 and top-20 MBA comparisons, Class of 2027, with averages, medians, source conflicts, GRE shares and waivers.Updated 4 Oct 2026](/articles/gmat-score-m7)