An error log on the GMAT Focus is the single piece of preparation infrastructure that most candidates set up and then abandon within a fortnight. The idea is sound: a deliberate record of every question you got wrong, and every question you got right for the wrong reason, so that you can target the recurring patterns instead of grinding random problem sets. In practice, the log either becomes a graveyard of half-finished entries, a list of screenshots with no analysis, or a sprawling spreadsheet that the candidate never opens again. The reason is rarely laziness. It is almost always a design failure: the log asks for too much information, too little information, or the wrong kind of information, and the candidate quietly gives up after the third evening of writing essay-length post-mortems.
This article is the walkthrough I would give a candidate sitting at my desk. It covers the columns that earn their keep, the entries that should never make it into the log, the cadence for revisiting old entries, and the small tactical decisions that decide whether the log becomes a score-raising tool or a decorative notebook. The advice applies across the GMAT Focus format and across all three sections: Quantitative, Verbal, and Data Insights. The examples lean on Quantitative and Data Insights because that is where most candidates leak the most points, but the same columns work for Critical Reasoning, Reading Comprehension, and Sentence Correction.
What a GMAT error log is actually for
The first mistake candidates make with an error log is treating it as a record of wrong answers. It is not. A list of wrong answers is a score report; you already have one of those, and it tells you that you missed six Data Sufficiency questions in module two. What it does not tell you is why. A working GMAT error log exists to capture the cognitive move that failed, not the outcome. The output of every entry is a single sentence describing the move, written in the candidate's own voice, and re-readable in six weeks without context.
Think of the log as a delayed feedback loop. On test day, you see a question, commit to an answer, and move on; the scoring engine has no interest in why you missed the question. During preparation, you see a question, commit to an answer, and then have an opportunity to inspect the move in slow motion. The log is the surface on which that inspection happens. If the inspection is shallow, the log teaches nothing. If the inspection is precise, the log becomes the most efficient tutor a candidate can build at home.
In my experience the candidates who climb from a 655 to a 705 in a single preparation cycle almost always share one habit: they have a log, they use it within twenty-four hours of every practice block, and they revisit old entries on a fixed schedule. Candidates who plateau in the high 600s usually have a log too, but it sits in a Google Drive folder and gets opened once a fortnight. The mechanism is the same; the cadence is the difference.
The two kinds of entry the log must capture
Most candidates log only the questions they answered incorrectly. That is half the job. The other half is the question they answered correctly for a reason that did not survive contact with the explanation. I call these lucky rights: a guess that happened to land on the right choice, a textbook answer reached by arithmetic that would have collapsed on a slightly harder prompt, a Critical Reasoning answer chosen by gut feel and confirmed only after the explanation reframed the argument. Logging lucky rights is uncomfortable, because the score column says Q01 = correct and most candidates want to move on. The log exists precisely to interrupt that reflex. A lucky right on a hard Critical Reasoning stem is more diagnostic than a clean miss on an easy Data Sufficiency question, because the lucky right is the entry that is most likely to repeat as a miss on test day.
For the GMAT Focus specifically, the temptation to skip lucky rights is stronger than it was on the classic GMAT, because the adaptive scoring means a single lucky right can be the difference between an easy module two and a hard module two. The candidate who guessed their way through the last three questions of a section at 70 percent accuracy has not demonstrated a 70 percent skill level. The log is the only place that distinction gets recorded.
The five columns that earn their keep
A blank log invites procrastination. A log with twenty columns invites a different kind of failure: the candidate spends forty minutes on a single entry and then stops logging altogether. The right design is the minimum viable log: a small number of columns, each pulling its weight, each answerable in under ninety seconds per question. Anything beyond that is feature creep.
Here are the five columns I recommend. They map directly to the three GMAT Focus sections without modification, and they survive a switch from a paper notebook to a spreadsheet to a Notion database without redesign.
- Date and block. A short label that lets you re-find the original question bank, section, and module. Format it as YYYY-MM-DD plus a one-word block tag such as “Q-morning” or “DI-evening.” This column takes five seconds and saves ten minutes of hunting later.
- Question type and topic. Two short tags. For Quantitative: Data Sufficiency, Problem Solving, Algebra, Number Properties, Word Problems, Geometry, Arithmetic. For Verbal: Critical Reasoning, Reading Comprehension, Sentence Correction, plus the CR subtype — strengthen, weaken, evaluate, assumption, inference, boldface, flaw, method of reasoning. For Data Insights: Data Sufficiency, Multi-Source Reasoning, Table Analysis, Graphics Interpretation, Two-Part Analysis. Granularity matters because the log only works if you can sort it later by the column that matters.
- What I chose vs. what was right. The letter, not the text. Two characters. This column is not for analysis; it is for triage. If a candidate chose C and the right answer was A, that is the only fact the candidate needs to re-trigger the memory of the question six weeks later.
- Why I chose it. The single most important column. One to three sentences, written in the candidate's own voice, describing the cognitive move that led to the wrong answer (or the lucky right). The most useful prompts to answer this column are: What did I think the question was asking? What assumption did I make? What step did I skip? Did I misread a number, a quantifier, or a polarity word?
- What the right move would have been. One to two sentences describing the correct approach in the abstract, not the correct arithmetic. The phrasing “divide before subtracting” is useful; the phrasing “the answer is 14” is not.
Anything beyond these five columns should be treated as a project, not a log. A sixth column for “time spent in seconds” can be useful for a four-week diagnostic phase and then removed. A column for “mood” is decorative. A column for “linked forum post” is a research tool, not a log column. The discipline of the five-column log is that every entry takes two to three minutes to write. Anything longer is essay-writing, and essays are why candidates abandon logs.
How to log across the three GMAT Focus sections
The five columns are constant, but the entries themselves look different across Quantitative, Verbal, and Data Insights. The reason is that the failure modes are different. A Quantitative miss is usually a single-step arithmetic or algebraic slip. A Verbal miss is usually a misread of the argument's structure. A Data Insights miss is usually a misread of the visual or the prompt's two-task structure. The log needs to capture the failure mode in language that the candidate will still understand a month later.
Quantitative entries
For Quantitative, the why I chose it column is almost always an admission of one of three moves. The first is arithmetic slip: the algebra was right, the numbers were right, the sign flipped on the way to the answer choice. The second is wrong abstraction: the candidate modelled the question as a system of equations when the right move was a number-property argument, or modelled it as a number-property argument when the right move was a system of equations. The third is unread stem: the candidate solved a different question from the one that was asked, usually because of a quantifier like “must be true” versus “could be true” or a constraint that was skipped on first read. Logging which of these three moves fired is the difference between fixing one habit and grinding three hundred practice questions that do not address the habit at all.
For Data Sufficiency specifically, the most useful tag in the why I chose it column is the statement on which the candidate committed. A candidate who consistently commits on statement one before testing statement two is a candidate with a habit problem, not a knowledge problem. The log surfaces the habit in a way that a score report cannot.
Verbal entries
For Critical Reasoning, the why I chose it column should record the question subtype the candidate believed they were answering. A candidate who reads a weaken question as a strengthen question is a candidate who has a stem-reading problem, and the log will show that problem appearing on four different passages in three different weeks. The fix is targeted stem drills, not more practice sets.
For Reading Comprehension, the most useful entry in the why I chose it column is the location of the answer choice the candidate picked. If the candidate picked an answer that required material from a different paragraph than the one referenced in the stem, the log will accumulate a cluster of cross-paragraph errors. If the candidate picked an answer that paraphrased a single sentence in the passage, the log will accumulate a cluster of detail-trap errors. Each cluster has a different fix.
For Sentence Correction, the column that earns its keep is what I chose vs. what was right combined with a one-word tag for the grammar concept: modifier, parallelism, subject-verb, idiom, pronoun, verb tense, comparison, quantity. Sentence Correction is the only Verbal question type where a single tag in the log can reliably point to the grammar chapter the candidate needs to revisit.
Data Insights entries
Data Insights rewards a slightly different log discipline. The question type column should always be one of the five: Data Sufficiency, Multi-Source Reasoning, Table Analysis, Graphics Interpretation, Two-Part Analysis. The why I chose it column should record whether the failure was in reading the visual, reading the prompt, or executing the arithmetic. The three failure modes have very different fixes, and a log that lumps them together will not produce a useful sort.
For Two-Part Analysis, the most useful single tag in the why I chose it column is whether the candidate treated the two parts as independent when they were linked, or treated them as linked when they were independent. That single tag, repeated across a month of entries, is more diagnostic than any score report.
Common pitfalls and how to avoid them
Most error logs fail for predictable reasons. The fixes are also predictable. The trap is treating the log as a content project — a place to record information — rather than a habits project, a place to record the cognitive move that failed. Below are the four failure modes I see most often in candidate logs and the discipline that addresses each one.
- The screenshot graveyard. The candidate takes a screenshot of every missed question, drops it into a folder, and never returns. The screenshot captures the question, not the reason. The fix is to write the reason in the log within twenty-four hours, in the candidate's own voice, before the emotional charge of the miss has faded. The screenshot is a backup, not the entry.
- The novel-length entry. The candidate writes a three-paragraph post-mortem for each question and burns out by week three. The fix is the two-to-three-minute rule: every entry is one to three sentences in the why I chose it column, full stop. If the entry needs more space, the candidate is writing a solution, not a log entry, and should move the analysis to a separate notebook.
- The never-revisited log. The candidate writes entries diligently but never opens the log again. The log has no delayed-feedback value if it is never revisited. The fix is a fixed revisit cadence — every Sunday, every two weeks, whatever the candidate can sustain — and a single rule: every revisit, the candidate writes one new line on the oldest entry, describing whether the same habit appeared in the new week.
- The log that records only wrong answers. The candidate logs misses and skips lucky rights. The fix is to treat every practice block as a logging session in which both misses and lucky rights get an entry. The lucky right is the higher-value entry on a hard question; skipping it is the single most expensive logging mistake.
These four pitfalls are not independent. A candidate who falls into the first pitfall usually falls into the third within a month, because the folder of screenshots is harder to revisit than a structured log and easier to ignore. The fix is to choose a log medium that the candidate can actually revisit — paper, spreadsheet, Notion, Anki, whatever — and to commit to a cadence before the first entry is written.
Spaced repetition and the revisit cadence
The error log earns its keep not when the entry is written but when the entry is re-read. A log that is written but never revisited is a journal. A journal has its uses, but score gains come from revisit, not from writing. The revisit cadence is the second piece of the log's design, and it is the piece most candidates skip.
The cadence I recommend is a three-pass rotation. The first revisit is twenty-four to seventy-two hours after the entry is written, when the topic is still warm. The second revisit is one to two weeks later, when the topic is cool but the question is still familiar. The third revisit is one month later, when the topic is cold and the candidate should be able to describe the right move without looking at the original question. Each revisit is short — five to ten minutes for a stack of twenty entries — and each revisit produces a single new line in the log: still happens, fixed, partially fixed, not relevant anymore.
This rotation has a particular shape. The first revisit surfaces habits that were hidden in the original entry. The second revisit surfaces habits that survived a week of new practice. The third revisit surfaces habits that survived a month of new practice. A habit that survives the third revisit is a habit the candidate cannot fix with more practice; it requires a different study method, a tutor, or a curriculum change. The log is what tells the candidate when that threshold has been reached.
For candidates using a digital log, the medium that supports this rotation most cheaply is a spreadsheet with a last revisited date column and a sort by date. For candidates using a paper log, a small notebook divided into weekly sections works, with the rightmost column reserved for revisit notes. The medium is less important than the rule: every entry gets three revisits, on a schedule, with a one-line note each time.
How the log interacts with section-level preparation
The error log is not a substitute for a curriculum. It is a feedback loop that runs in parallel to a curriculum, and the two have to be designed to talk to each other. A candidate studying Data Insights using a topic-by-topic curriculum — graphics interpretation this week, table analysis next week — needs the log to be sortable by question type so that the cluster of misses this week is visible before the next week begins. A candidate running a mixed review block needs the log to be sortable by date so that the cluster of recent misses is visible before the next review block begins.
For the GMAT Focus in particular, the adaptive scoring means that the question mix inside any given practice block is not under the candidate's control. A block of twenty Data Insights questions might include three Two-Part Analysis questions, four Graphics Interpretation questions, two Multi-Source Reasoning tab pairs, and the rest Data Sufficiency. The log is the only way to see, in the candidate's own voice, which subtype is leaking the most points across blocks. A score report can tell the candidate the section score; only the log can tell the candidate which subtype of Two-Part Analysis is producing the misses.
Using the log to choose the next study block
The single highest-leverage use of the log is the choice of the next study block. The candidate finishes a practice block, logs the entries, sorts the log by question type, and looks for the largest cluster. If the largest cluster is Data Sufficiency — committing on statement one, the next study block is a stem-reading drill, not a Data Sufficiency problem set. If the largest cluster is Critical Reasoning — reading weaken as strengthen, the next study block is a stem-identification drill, not a Critical Reasoning passage set.
This is the move that converts the log from a record into a steering wheel. The candidate is no longer choosing the next study block from a syllabus or a chapter list; the candidate is choosing it from the log. The syllabus becomes the menu; the log becomes the order in which the menu items are consumed.
From the log to the score: a worked example
To make the cadence concrete, walk through a single entry. A candidate logs a Quantitative Data Sufficiency question on a Tuesday evening. The question concerns a system of two equations in two unknowns. The candidate chose C (both statements together are sufficient) and the right answer was E (neither statement is sufficient, even together). The why I chose it column reads: “I saw two equations, I assumed they had a unique solution, I did not check that the equations might be dependent.” The what the right move would have been column reads: “When a system has the same number of equations as unknowns, check whether the equations are linearly dependent before committing.” The entry takes two minutes.
On Friday, the candidate revisits the entry. The new line reads: “Same habit appeared on a Problem Solving question; I assumed two constraints gave a unique solution without checking.” The candidate has now confirmed that the habit is a habit, not a one-off slip. The next study block becomes a number-property drill on dependent systems, not a generic Data Sufficiency set.
Two weeks later, the candidate revisits the entry again. The new line reads: “Did not fire on today's block. I am consciously checking dependency now.” The habit is on the way to being fixed, and the log has the receipt. One month later, the third revisit: “Did not fire this month. Closing the entry.” The entry is now a four-line record of a habit that was identified, confirmed, targeted, and closed. That four-line record is what a score report cannot produce.
The same pattern works for Verbal. A candidate logs a Critical Reasoning evaluate question that the candidate answered as if it were an assumption question. The first revisit surfaces a second evaluate-as-assumption miss from a different passage. The candidate now knows the habit is real. The next study block is a stem-identification drill on evaluate versus assumption stems, not a Critical Reasoning passage set. Two weeks later, the third revisit: the habit is fading. One month later: closed.
What to do with the log in the final two weeks
The final two weeks before the GMAT Focus are not the time to expand the log. They are the time to use it. The candidate should be running timed full-length practice blocks, logging as usual, and then spending the time between blocks reading the log, not writing in it. The reading has two functions. The first is to confirm that the habits closed in earlier months have not reopened under timed conditions. The second is to identify any new habit that has appeared under the pressure of full-length timing.
A useful discipline in the final two weeks is the last-week filter: at the end of each timed block, the candidate reads only the entries from the previous seven days, and writes one new line summarising the most common habit. If the most common habit is the same one that appeared in week one, the candidate has regressed and the next study block has to address that regression. If the most common habit is new, the candidate has discovered a habit that the early-preparation log could not have surfaced, because the habit only appears under fatigue.
The final day before the test is not a logging day. It is a reading day. The candidate should re-read the closed entries — the habits that are fixed — as a form of self-confirmation, not a study session. The log's job on the day before the test is to remind the candidate of the work already done.
Conclusion and next steps
A GMAT error log is a feedback loop, not a record. It earns its keep by capturing the cognitive move that failed, in the candidate's own voice, on a cadence that the candidate can sustain. The five columns — date and block, question type and topic, what I chose vs. what was right, why I chose it, what the right move would have been — are the minimum viable design. The three-pass revisit rotation is what converts entries into score gains. The log is what tells the candidate which study block to run next, and it is the only preparation tool that improves on a curve, because the log itself gets more useful as the candidate's habits get more visible.
The next step for a candidate reading this is to build the log medium — paper, spreadsheet, or database — and to log the next practice block within twenty-four hours. A diagnostic walkthrough of an existing log, with a tutor reading the entries and pointing out the highest-leverage habit, is a natural starting point for a candidate who has been logging but not steering.