About TestLumo

What we are trying to do

Good GMAT practice is expensive and bad GMAT practice is everywhere. That combination pushes a large number of candidates towards free material that looks plausible and is quietly wrong: questions that test content the current exam removed, timing that does not match the real clock, interfaces that let you skip freely when the real one does not, and scores generated by a formula nobody will explain.

The consequence is not that those candidates learn nothing. It is that they cannot tell how they are doing. A practice score that is inflated by twenty points, or produced from a test that gave them 60 minutes for a 45 minute section, is worse than no score at all, because decisions get made on it: whether to book the exam, whether to keep studying, which section to prioritise, when to stop.

TestLumo exists to make the accurate version free. Not a stripped-down sample with the real product behind a card, but the whole thing: full length practice tests, section tests for Quantitative Reasoning, Verbal Reasoning and Data Insights, and the same score report after every one of them.

Who this is for

Anyone sitting the GMAT Focus Edition, but three groups in particular get disproportionate value from what we have built.

Candidates who are self-preparing

If you are working through the Official Guide on your own, the thing you are missing is not questions, it is measurement under real conditions. A book cannot enforce a 45 minute clock, cannot cap your answer edits at three, and cannot show you a scatter of your response times against outcomes. That is exactly the gap the practice tests fill, and it costs nothing to use them alongside whatever book you already own.

Candidates who have already sat the exam once

A retake is a targeting problem rather than a learning problem. You usually know roughly where the score went; what you need is a way to confirm it and track whether the fix is working. Section tests are the efficient tool here, because they give you a format-level breakdown every 45 minutes rather than every two and a half hours.

Candidates working with old material

A great deal of GMAT preparation in circulation predates the Focus Edition. If your materials include geometry in the quantitative section, Sentence Correction in the verbal section, or Data Sufficiency outside Data Insights, a meaningful share of your study time is aimed at content that is no longer tested. Sitting one of our tests is the fastest way to see what the current exam actually contains.

We are less useful if you are looking to be taught mathematics from scratch. TestLumo is a practice and diagnosis platform, and it assumes you have or can get the underlying content from somewhere. What it does is tell you precisely which parts of it you are losing marks on.

How the questions are built

This is the part we care most about and it is worth explaining properly. Every practice question on TestLumo is original, and every one is written against a specification derived from measurement rather than from opinion.

The benchmark

We took the 797 questions in the current GMAT Official Guide and measured them. Not their content, which is copyright and which we do not reproduce in any form, but their shape. Stem word counts by format and by difficulty band. Answer choice lengths. The spread between the longest and shortest choice in each set. How often five choices share an opening word. Passage lengths and how many questions attach to each. Concept distribution across each section. The result is a set of numeric bands that describe what a real item looks like.

Those numbers do a specific job. A question that sits far outside the band for its format will feel wrong to an experienced candidate even when it is mathematically or logically sound, and feeling wrong is enough to make practice unrepresentative. A 140 word Problem Solving stem, a 40 word Critical Reasoning argument or a Data Sufficiency pair where one statement is five times the length of the other are all technically possible and none of them behave like the exam.

Distractors

The rule we apply most strictly is that every wrong answer has to be traceable to a specific candidate error. The official explanations set this standard: they never dismiss a choice as merely wrong, they name the misstep that produces it. If we cannot name the error a distractor catches, it is filler, a strong candidate will eliminate it instantly, and the question is easier than its difficulty label claims.

• Answering the wrong question: solving for x where the item asked for 2x, or the part where it asked for the whole. The most common quantitative error there is.
• One step short: the value at the penultimate stage of the correct chain.
• Sign, direction or complement flips: an increase for a decrease, 40 per cent for the remaining 60.
• Right relation, wrong pair: the correct operation applied to the wrong two quantities.
• True but irrelevant, in verbal: something the passage supports that does not answer the question asked.
• Too strong: the right idea overstated with always, only or proves.
• Out of scope: plausible real-world reasoning the text never licenses.
• Reversed causation: cause and effect swapped inside an otherwise correct reading.

Answer key balance

The published corpus is not evenly distributed across the five positions. We measured A at 13.8 per cent and D at 23.8 per cent, which is an artefact of how the book was assembled rather than a property of the exam. We do not copy it. Our correct answer positions are randomised to a flat 20 per cent each, because a bank that leans towards any letter is exploitable and a candidate who learns to exploit it has learned nothing transferable.

Data Sufficiency is the deliberate exception. Its natural distribution genuinely favours the both-statements-needed and still-insufficient cases, because those are the ones worth testing, and flattening it would misrepresent the format. We keep it close to its observed shape.

Difficulty and coverage

The Official Guide splits its practice set into near-equal thirds of easy, medium and hard items, and our bank matches that, so no paper is quietly easier or harder than the exam it stands in for. Concept coverage follows the measured distribution too, which means Applied Problems is the largest quantitative bucket on our tests because it is the largest bucket on the exam, and combinatorics is a small one for the same reason.

Every question carries three tags drawn from the official score report vocabulary: content domain, question type and fundamental skill. That is why our reports can be read alongside a real score card without translation.

How the tests behave

Accuracy in the questions is only half of it. The exam is also a set of rules, and we reproduce them rather than approximating them.

• Three sections of 45 minutes each: 21 Quantitative Reasoning, 23 Verbal Reasoning, 20 Data Insights.
• A section order chosen from six permutations before the first section, with one minute to decide and a default if the timer lapses.
• One optional 10-minute break, offered after the first section and again after the second, with the first taking forfeiting the second.
• No skipping. You answer and advance; revisiting happens only at the review screen.
• Three answer edits per section, each confirmed through a dialogue, on a counter that does not reset.
• The review screen is skipped entirely if the section clock expires first.
• An on-screen calculator in Data Insights and in no other section.
• Bookmarking on any question, surfaced on the review screen.

The last few of those are the ones other free tests tend to drop, and they are the ones that most affect how a section is played. A candidate who has never had to decide how to spend three edits, or who has always assumed the review screen will be waiting, is going to make those decisions for the first time on the day that counts.

What our scoring claims, and what it does not

Section scores are reported on the 60 to 90 scale and totals on the 205 to 805 scale in 10 point steps, matching the official reporting. Our estimator maps your performance onto those scales taking account of the difficulty and format mix of the questions you actually saw, rather than converting a raw count of correct answers.

What it does not claim is prediction. We are not the test maker, we do not have access to the official scoring model, and no third party does. Our estimate is calibrated to be honest rather than encouraging, which occasionally means it reads lower than a number another free test handed you. We think that is the more useful error to make.

• Treat a single practice score as a data point with error bars, not as a forecast.
• Three tests across six weeks tell you far more than any one of them does. The trend is the signal.
• Percentiles are indicative. They are derived from our own population, which is not the official test-taking population.
• The section and format breakdowns are the genuinely reliable part of the report. Use them.

We would rather be trusted on the diagnosis than believed on the number. The panel we would point any candidate at first is not the score at all; it is the scatter of response time against outcome, which shows you where your time went and what it bought.

Why it is free

A fair question, and one worth answering plainly rather than with a slogan. TestLumo is part of a small network of exam preparation sites that has been running for several years, and the practice tests here are genuinely free with no card, no trial period and no locked score report.

The economics are straightforward. People who prepare seriously for one exam often want structured tuition, and some of them ask us for it. A free, accurate practice platform is how they find us, and it is the part of what we do that scales without a person attached to every candidate. The tests are not a demonstration of a paid product; they are complete on their own, and the large majority of people who use them will never buy anything, which is fine.

The other half of the answer is that free is what makes the site work at all. The value of a practice platform to any individual candidate depends on how many people have used it, because that is where our percentile estimates and difficulty calibration come from. Putting a card in front of that would make the product worse for everyone, including the people who paid.

Independent, and clear about it

TestLumo is not affiliated with, endorsed by or connected to the Graduate Management Admission Council in any way. GMAT is their trademark and we use it only to describe what our practice material is for.

Every question on this site is written by us. We do not reproduce official questions, and the benchmark work described above exists precisely so that we can match the shape of real items without copying any of them. Where we quote a measurement from the Official Guide corpus, it is a statistic about that corpus, not an extract from it.

We also try to be careful about claims. Where a figure on this site is a measurement, we say what it was measured from and how many items it covers. Where something is an estimate, we say that too. If you find a factual error anywhere in our content, particularly about exam structure or scoring, we want to hear about it, because a preparation site that is wrong about the exam is worse than useless.

Things we have deliberately not done

A short list, because what a preparation site refuses to do says as much as what it offers. None of these are missing because they were hard; they are missing because we think they make practice worse.

• No score guarantees. Nobody outside the test maker can promise a point gain, and every such promise we have seen rests on conditions that quietly exclude most people who fail to hit it.
• No invented testimonials. Where we quote a candidate, it is a real person who agreed to it.
• No inflated question counts. Our 192 questions serve 12 tests because each full-length mock is assembled from three of the section papers — we say so rather than advertising 768.
• No timer that pauses when you switch tabs. The exam does not do that, so neither do we.
• No pre-Focus content left in the bank because removing it would shrink the number. Geometry and Sentence Correction items are gone.
• No score report behind a paywall, and no account to create. We ask for an email address so we can send you the report and keep your attempts together — the full diagnosis, not just the number.

The last one is the one we feel most strongly about. A total score with no breakdown underneath it is the least useful output a practice test can produce, and it is the standard output of most free tests precisely because it creates an incentive to register. Our reports open in full immediately: section scores, format breakdowns, per-question response times, the time-against-outcome scatter and the edit history, whether or not you have an account.

Where to start

If you have not sat a GMAT practice test recently, start with a full length one and give it the honest 135 minutes. If you already know which section is costing you, go straight to the quantitative practice test, the verbal practice test or the data insights practice test. Everything is free, nothing needs a card, and the report is the same either way.