The introduction of the Data Insights section was one of the major changes brought about in the GMAT Focus Edition in 2023. Combining the Integrated Reasoning and Data Sufficiency elements of the former GMAT exam, the GMAT Data Insights section is designed to test your ability to analyze and synthesize data from multiple sources and make informed decisions from that data.
While data sufficiency questions are slightly less complicated, the other four types are particularly complex and abstract. These questions are difficult and expensive to produce because the tables, graphs, and scenarios are so intricate and elaborate. The precision in wording and clever data tricks that you find in official Data Insights questions are almost impossible for test prep companies to replicate. Trying to improve your Data Insights skills with unofficial questions is futile because you will be practicing for a much less complex and far easier game.
Remember that the skills being tested in Data Insights are simply a mix of what is assessed in the verbal and quant sections, but the presentation varies dramatically. Below you will find two sections: one example for Data Sufficiency and one Graphics Interpretation example to represent the other four types you see in Data Insights.
Data Sufficiency—Find the Reward in Your Answer!
Data Sufficiency Best Practices
Data Sufficiency is the most important question type in the Data Insights section. Of the 20 questions you encounter in Data Insights, 8 are usually Data Sufficiency. Additionally, it is probably easiest to gain a competitive advantage on Data Sufficiency compared to the other 4 question types (MultiSource Reasoning, Table Analysis, Graphics Interpretation, and TwoPart Analysis). All of this means one thing: spend time learning the right approaches and strategies for Data Sufficiency!
In our curriculum we start by making sure that everyone has learned exactly what constitutes sufficiency in a given statement, and we make sure people avoid the following common rookie mistakes:
 In a “yes or no” question, a “no” is just as good as a “yes.” Don’t overlook the cleverly created “no” answer.
 Don’t treat a “yes or no” question as if it is a “what is the value?” question. A statement can allow for many values but still be sufficient in a “yes or no” question.
 Don’t accidentally carry information between statements. On every DS question, ask yourself consciously: have I carried a tiny bit of information between the two statements?
 Don’t accidentally treat the question as a statement or vice versa.
 Don’t be lazy and don’t make assumptions. If a statement seems insufficient, do everything you can to prove it is sufficient or vice versa. If the answer you are about to pick does not differentiate yourself from smart people, leverage harder or be more critical. Find the dopamine response!
 Recognize the common math setups used in data sufficiency. If you fall for a trap once, never fall for it again.
 Don’t do unnecessary work but make sure you do enough to prove sufficiency or insufficiency.
After ironing out those common mistakes, we focus on the “4level strategy” for avoiding cons and finding the reward in your answer. In data sufficiency there are two basic mistakes you can make: 1. You can undervalue the information given. In this scenario, you think a statement is not sufficient, when it really is. 2. You can overvalue the information given. In this scenario, you think something is sufficient when it really isn’t. To prevent these two mistakes in data sufficiency—the undervaluing or overvaluing of information—always consciously consider the level(s) beside the one you are going to pick. This may sound like a simple strategy, but people don’t consciously do it on every question, and you will be amazed how often it will keep you from picking the “sucker” answer. The Four Levels of Sufficiency are outlined below:
Level 1: (D) Each Statement Alone is Sufficient
Level 2: (A) or (B) One Statement Alone is Sufficient
Level 3: (C) Both Statements Together Are Sufficient, But Neither Alone Is
Level 4: (E) Both Statements Together Are Not Sufficient
As an example: at level 1, you are picking (D), which means you believe that each statement alone provides enough information to answer the question. If you are incorrect at level 1, there is only one mistake you could have made: you overvalued the information in one of the statements and the correct answer is really at level 2 (or lower). When you miss a data sufficiency question, it is almost always by picking an answer that is one level away from the correct one: you have just slightly overvalued or undervalued information in one or both of the statements. These questions are cleverly designed to make you do just that!
NOTE: if you are more than one level away from the correct answer in a data sufficiency question, it means you either don’t understand the underlying math concept or you are misreading the question or statements (and thus it is not a “data sufficiency” mistake).
If you learn how to avoid common rookie mistakes and actively leverage the “4level Strategy” to find the reward in your answer, your accuracy and efficiency will go up dramatically in Data Sufficiency.
GMAT Data Sufficiency Example Question
A gourmet cheese shop sold several orders of English Stilton and Spanish Manchego yesterday. Customer A purchased 15 pounds of English Stilton and 3.75 pounds of Spanish Manchego for a total of $438.00. If the price for each of these cheeses is proportional to its weight, what is the price of 1 pound of Spanish Manchego?
(1) Customer B purchased 5 pounds of English Stilton and 4 pounds of Spanish Manchego for a total of $214.75.
(2) Customer C purchased 6 pounds of English Stilton and 1.5 pounds of Spanish Manchego for a total of $175.20.

 Statement (1) ALONE is sufficient, but statement (2) alone is not sufficient.
 Statement (2) ALONE is sufficient, but statement (1) alone is not sufficient.
 BOTH statements TOGETHER are sufficient, but NEITHER statement ALONE is sufficient.
 EACH statement ALONE is sufficient.
 Statements (1) and (2) TOGETHER are not sufficient.
As you just learned in the Best Practices section, data sufficiency questions involve an interesting mix of math skills, critical reasoning, reading carefully, and spotting the con. This official example question perfectly captures all of these elements. As a first step, you should carefully organize all the given information provided in the question stem and isolate clearly what is being asked. In the question stem, you are given a revenue equation for customer A as follows, using E as the price per pound for English Stilton cheese and S for Spanish Manchego cheese:
Additionally, you are told that the price for each of these cheeses is proportional to its weight, which simply means that there is a constant and unchanging price per pound. The question is asking for the value of S and you can now evaluate each statement and apply the given equation to both:
(1) This statement gives you a second revenue equation for the purchase of Customer B as follows: Since this gives you a second, unique equation using the two variables provided in the question stem, you know you can solve for each variable and find the value for S. NOTE: you should not actually figure out what the value is (would be a lot of wasted time), but you do need to make absolutely sure you can find it. This statement is indeed sufficient, so you know the answer must be (A) or (D) and there are thus only two levels to consider from the 4level strategy discussed in the Best Practices section.
(2) This statement appears to also give you another unique equation for customer C that will allow you to solve for both unknowns. The question is designed for you to pick (D) and many people do just that!! Before you pick (D), you are supposed to consider any possible answer one level away, in this case (A). Also, your “bull&^% detector” should also be going off as you go to select (D): why would they make this question if (D) is the answer and how am I differentiating myself from smart people by picking it? The answer: they would not make this question as (D) is too simple of an answer.
Since you are sure the first statement is sufficient, you need to figure out what kind of trick might exist in the 2nd. This point is important: you need to know mathematically when a 2nd equation with two unknowns might NOT allow you to solve for the two unknowns. When might that happen? When the 2nd equation is actually the same, often cleverly, as the first one. If you are looking for this con (and it has been used dozens and dozens of times on previous official Data Sufficiency questions) then you will notice that the equation given in statement (2) is actually the same as the equation given in the question stem. Let’s compare them:
If you take the 2nd equation and multiply each term by 2.5 or , then you see it is identical to the given equation. By using complex numbers, the test makers make it much harder to see that the two equations are the same,but you will see it if you are aware of this common con. Since statement (2) is not giving you new information then it cannot be sufficient and the correct answer is (A).
Graphics Interpretation—Answer the Proper Question!
Best Practices for Data Insights
For the four types of questions in the Data Insights section other than Data Sufficiency—MultiSource Reasoning, Graphics Interpretation, Table Analysis, and TwoPart Analysis—there is an extremely wide range of presentations for the different scenarios, graphs, tables, etc. Because of that, these questions are a little harder to prepare for! Sometimes the questions address very particular math concepts—weighted average, venn diagram, probability, data analysis—and sometimes the issues relate more to reading comprehension and critical reasoning. Overall, regardless of the question type, there are a few important strategies and best practices when attacking Data Insights questions:
 Read carefully and answer the proper question. By far, these are the two most important skills when solving Data Insights questions, particularly in MultiSource Reasoning. The wording and interpretation tricks on Data Insights questions are quite sophisticated so they require an obsessive attention to details and wording.
 Learn how to sort through unnecessary “garbage” and navigate red herrings. A big part of what is being tested in Data Insights is your ability to figure out what matters in a sea of information. Essentially, these problems are designed to differentiate between those who can efficiently sort through data to answer particular questions and those who get sidetracked by irrelevant or purposefully misleading information. With this in mind, it is important that you always examine questions and prompts before diving headfirst into complex graphs, tables, and sets of information. Get a general sense of provided information but then immediately go to the prompts so you know exactly what you need to do.
 Know the traps and cons associated with certain types of graphs, sortable tables, and particular question types. Whenever there is percentage and absolute number data, don’t accidentally conflate the two; know what a positive and negative correlation look like; don’t get your answers reversed in confusing twopart analysis questions; etc.
 Remember that you must get all prompts correct to get credit for a question. If you have spent a lot of time to get two of the three prompts correct, make damn sure you don’t get sloppy or lazy on the remaining one!
Because Data Insights questions are so varied and contain so much abstract and often irrelevant information, you need a set of best practices and guiding principles as outlined above. Whenever you miss a Data Insights question, always ask yourself the following questions to improve on future problems:
 Did I miss this because I don’t understand a core math concept or core principle of logic and critical reasoning? If so, make sure you study the concept and principle so you are better prepared the next time it is tested.
 Did I miss this question because I misinterpreted tricky wording or answered the wrong question? If so, figure out exactly how you were tricked into that wording mistake and be on the lookout for this type of difficulty in the future.
 Did I not properly understand a certain graph type or did I not sort a table of data in the most effective way? If so, learn the specifics of that graph type or get more skilled at sorting tables in the most effective way for a given prompt.
 Did I waste time on irrelevant information that was intellectually engaging but not relevant to the particular prompts in the question? If so, realize that a lot of the provided information is unimportant in these questions and use prompts and questions actively to quickly find what really matters in the provided information and data.
With lots of practice on each of the four question types, important patterns emerge for how difficulty is created and how to answer prompts more accurately and efficiently. Data Insights is a challenging section that requires a primary focus on strategy and approach rather than on memorizing or learning particular math skills/content knowledge.
GMAT Data Insights Example Question
Students in 65 education systems worldwide took a global exam in reading, science, and mathematics. On the scatterplot, each of the 65 data points displays the average mathematics score (M) and the average reading score (R), both rounded to the nearest integer, for one of the education systems. The line represents all points where M and R are equal.
Based on the information provided, select from each dropdown menu (selections represented as bullets) the option that creates the most accurate statement.
Prompt #1
The percent of the 65 education systems for which the value of M exceeds the value of R is between …
 0 and 25 percent
 25 and 50 percent
 50 and 75 percent
 75 and 100 percent
Prompt #2
The value of R exceeds the value of M by the greatest amount for the education system for which the value of R is in the interval from …
 380 to 400
 400 to 420
 440 to 460
 500 to 520
 560 to 580
With that basic understanding of the graph, you should then look at the prompts to see in particular what tasks you need to perform:
Prompt #1 is relatively straightforward. The question is simply asking what percentage of the 65 data points lie to the right of the middle line, since points lie in that zone when M > R. For this type of question, you can always eyeball the answer and it would be unnecessary to actually count points…the range of the answers will be such that you can do it visually. It is clear from the graph that more data points lie to the left of the line, but not by that much. If you had to give a figure without using answers, most would guess around 4045% are to the right and thus the answer of 2550% in the drop down menu is correct. NOTE: as discussed in the best practices section, you should double check that you are answering the proper question and not accidentally looking at points in which R>M instead of M>R
When a prompt is relatively straightforward in a question, you should always double check that you have not fallen for a con and then look to the other prompt(s) to be clever or tricky. Prompt #2 is exactly that: it is amazing how many students miss it!
In terms of what this prompt is asking, you need to find the point above the line that is farthest to the left. The farther left you are, the lower the math score is for a particular reading score and the difference between the two would be the greatest in which R >M. Imagine a point at 380 in the xaxis that is at 500 on the yaxis…that point would be WAY left of the center line and the amount of difference would be a very large 120. In searching for the point that is furthest left, you have a few candidates at first glance. By looking carefully, you can see that the point straight up from around 405 on the math xaxis is farthest left. It goes a full block + ¾ of a block to the left and none of the others do that.
Interestingly, almost everyone does that task correctly. The mistake they then make is that they see it is between 400 and 420 on the xaxis and incorrectly pick that range in the dropdown menu, when the prompt is asking for the range of r on the yaxis! Because you naturally look at this graph in the xaxis it is VERY easy to make this mistake, and this trick has been used on many scatterplot graph questions. If you look at the R interval on the yaxis in which that point lies, you see that the correct answer is between 440 and 460. Even if you do the most important tasks correctly on this question, it is still easy to miss it and I watch a lot of smart students do exactly that by accidentally focusing on the xaxis instead of the yaxis.
Caution: Avoid Unofficial GMAT Data Insights Questions
As you can see from these two examples, Data Insights questions are highly variable and test a wide range of skills.
Data Sufficiency mixes core quant and data skills with critical reasoning: you must be highly critical of the information provided and find the reward in your answer. These questions are made to reward both skills (quantitative reasoning and critical thinking) but unofficial questions tend to focus primarily on the underlying math skills without the clever wording and critical reasoning you see in official problems.
The other four types—MultiSource Reasoning, Table Analysis, Graphics Interpretation, and TwoPart Analysis—are simply so intricate and so hard to create that unofficial questions border on useless for improving your skills on these difficult questions.
Even more so than on the questions found in the verbal and quant sections, the importance of using official GMAT Data Insights questions in your preparation cannot be overstated.
To fully prepare for the complexity and variety of Data Insights questions on the GMAT, consider enrolling in our GMAT prep course, which uses exclusively official material throughout its curriculum.
Questions about this article? Email us or leave a comment below.