How to analyse a NEET mock test (and actually raise your score)

What is a NEET mock test actually for?

A NEET mock test is for finding out why you lose marks, not for finding out how many marks you have. The total is a summary of 180 separate events, and a summary cannot tell you which of those events to fix.

The real paper is fixed. According to the NTA Information Bulletin for NEET (UG) 2026, it has 180 compulsory multiple-choice questions, 45 in Physics, 45 in Chemistry and 90 in Biology, answered in 180 minutes for a maximum of 720 marks. A good mock copies that structure so that the only variable left is you. Once the structure is controlled, every wrong answer becomes evidence about your preparation.

Students who treat mocks as verdicts tend to take more of them and analyse fewer. Students who treat them as diagnostics tend to take fewer and get more from each. The second habit is the one this guide is about.

Why does the total score mislead you?

The total score misleads you because two very different problems can produce the same number. A student who scores 540 because of weak organic chemistry and a student who scores 540 because they ran out of time in Physics need completely different next weeks.

The total also moves for reasons unrelated to you. Different mock papers have different difficulty, so a ten-mark drop can mean a harder paper rather than a worse student. This is the problem Item Response Theory was built to solve: it weighs each answer by how difficult and how discriminating the question was, so that getting a hard question right counts as stronger evidence than getting an easy one right.

You do not need a statistical model to take the lesson. Look past the total to the pattern underneath it: which subjects, which chapters, and above all which kinds of mistakes.

What are the five types of NEET mistakes?

Almost every lost mark in NEET belongs to one of five types, and naming the type is the most useful thing you can do in analysis. The type tells you the fix; the chapter alone does not.

Concept gaps are the only type that more reading fixes. The other four are skills, and they improve with deliberate practice under timed conditions. Many students keep rereading chapters when most of their lost marks are misreads and slips, which is why their scores stay flat despite hard work.

How do you analyse a mock test, step by step?

A useful post-mock analysis follows the same six steps every time, so that results are comparable from one mock to the next. Do it the same day while you still remember what you were thinking.

• Step 1. Mark every question as correct, wrong, or unattempted, and separately flag any correct answer that was a guess. A lucky guess is a hidden wrong answer.
• Step 2. For every wrong and guessed question, write one line: the chapter, the concept, and the error type from the table above.
• Step 3. Before reading the solution, try the question again with no time limit. If you now get it right, the error was not a concept gap.
• Step 4. Read the full explanation, including why each wrong option is wrong. Distractors are written to catch specific misunderstandings, and knowing which one caught you is the point.
• Step 5. Count errors by type and by subject. Write the three largest counts at the top of your error log.
• Step 6. Schedule a re-attempt of every wrong question for three to seven days later, from a blank page, without looking at your notes.

The whole process usually takes as long as the test itself. That feels expensive until you notice that the alternative, sitting another mock without fixing anything, costs the same three hours and teaches less.

When should you skip a question in NEET?

You should skip a NEET question only when you cannot eliminate any option; once you can eliminate even one, attempting has a positive expected value. This follows directly from the marking scheme, which the NTA bulletin sets at plus four for a correct answer, minus one for an incorrect answer and zero for an unanswered question.

The arithmetic is short. With four options and no knowledge at all, a guess is right one time in four: 0.25 times 4 marks, minus 0.75 times 1 mark, gives an expected +0.25 marks. Eliminate one option confidently and you are choosing among three: one third of 4, minus two thirds of 1, gives about +0.67 marks. Eliminate two and the expected value rises to +1.5 marks.

Two caveats matter more than the arithmetic. First, expected value is an average over many questions, and a single guess can still cost you a mark. Second, the calculation assumes your elimination is genuinely correct. If your 'eliminated' option is often the right answer, your mock data will show it, and that is a distractor-trap problem to fix rather than a reason to guess more.

How should you use time data from a mock?

Time data tells you where the paper is spending you, rather than where you are spending the paper. Note roughly when you finished each subject and how many questions you left for a final pass.

In the real exam, 180 questions in 180 minutes works out to one minute per question on average, and the average hides the spread: many Biology questions take seconds while some Physics numericals take several minutes. The NTA notice of 12 June 2026 extended the window for the re-examination to 2:00 to 5:15 PM to cover formalities such as signing attendance sheets, and added four rough-work pages in place of two. Neither change alters the fact that pacing is a skill you have to practise.

If time pressure is a top-three error type for you, experiment with attempt order across mocks: one mock starting with Biology, one starting with Chemistry, and so on. Keep the order that produces the fewest time-pressure errors, not the one that feels most comfortable.

What should an error log contain?

An error log should contain one row per lost mark, with enough detail that you could recreate the mistake a month later. A notebook works; a spreadsheet makes counting easier.

• Date and mock name, so you can see trends over time.
• Subject, chapter and the specific concept, as narrow as you can make it.
• Error type from the five-type taxonomy.
• One sentence on what you will do differently next time.
• The date you re-attempted it and whether you got it right.

The log becomes most valuable after four or five mocks, when patterns appear that no single test could show. A concept that has caught you three times is a priority regardless of how the chapter is weighted in the syllabus.

How many mock tests should you take before NEET?

Take as many mock tests as you can properly analyse, and no more. There is no evidence-based magic number, and we will not invent one.

A practical rule is to leave enough gap between mocks to act on what the last one told you. If the next mock arrives before you have re-attempted the last one's wrong answers, you are collecting scores instead of improving. Early in preparation that might mean one full mock every week or two alongside chapter tests; closer to the exam, many students move to more frequent full-length papers to build stamina.

How does Lumi approach mock-test analysis?

Lumi does the counting part of this protocol automatically and leaves the thinking to you. Lumi publishes the methods it uses, so here they are.

Ability is estimated with a two-parameter logistic Item Response Theory model, fitted by maximum likelihood, so that a harder, sharply discriminating question counts for more than an easy one. Inside an adaptive exam, the next question is the one with the highest Fisher information at your current estimate, meaning the question whose answer would tell the system the most. Across a study plan, a simpleKT knowledge-tracing model predicts whether you would answer each concept correctly now, and FSRS schedules re-attempts of what you got wrong. All of this runs over roughly 2,700 concept atoms across Physics, Chemistry and Biology, which is what lets a wrong answer point at something smaller than a chapter.

None of that replaces the reflection in steps 2 and 3 of the protocol above. A model can tell you which concept failed; only you know whether you misread the question.