Victor gave me a clear understanding of how to structure a case interview using a On my second attempt two years later, I followed everything Victor Cheng Figure 4: Profitability Framework Printable Version: A PDF of this framework. Case Interview Secrets A Former McKinsey Interviewer Reveals How to Get in Consulting by by Victor Cheng PDF File: ([PDF]) Case Interview 2 Secrets: A. Victor Cheng Case Interview Marathon Workshop. Victor Cheng's. Case Interview Core Frameworks v By Victor Cheng.

Victor Cheng Case Interview Secrets Pdf

Language:English, Portuguese, French
Published (Last):04.01.2016
ePub File Size:15.51 MB
PDF File Size:16.20 MB
Distribution:Free* [*Registration Required]
Uploaded by: MARIANO Case Interview Marathon Workshop. Overhead Slides v By Victor Cheng These materials provided on an. Former McKinsey case interviewer, reveals his insider secrets to passing the infamous case interview. Read Case Interview Secrets PDF - A Former McKinsey Interviewer Reveals How to Get Multiple Job Offers in Consulting by Victor Cheng.

My conclusion is that my initial hypothesis seems to have been proven false. It looks like the client is facing a cost problem more so than a revenue problem. My next hypothesis is Notice how this hypothesis-experiment-conclusion process repeats itself.

The first step is to state a hypothesis, and the second step is to decide how you will test your hypothesis using an issue tree, or framework. An issue tree lays out a set of logical conditions that, if proven correct, prove the hypothesis correct. The term issue tree comes from the way such a logical structure looks—like a tree on its side—when diagrammed on paper: Figure 1: Issue Tree Diagram Alternatively, you can think of an issue tree as a logical argument.

Attorneys use an issue tree structure when writing a legal brief. Your high school English or native language teacher used something similar when assigning an expository writing assignment in which you had to argue logically for a point of view.

The logic of an issue tree is as follows: A two-layer issue tree has two layers of conditions: A two-layer issue tree looks like this: Frameworks as Issue Tree Templates A framework is essentially a template based on a commonly used issue tree.

Because consultants tend to see similar problems among many different clients, some have developed frameworks to deal with these frequently occurring problems. For example, many clients experience declining profits. Revenues are further broken down into their component parts: A similar breakdown is used for costs. Figure 2: Profitability Framework Diagram In the profitability framework, the logical construct happens to be quantitative in nature and uses a mathematical equation to solve profitability problems.

For example: Eating Mexican food for dinner is the optimal choice Culinary benefit: We like the food Financial benefit: Assuming the structure is valid, the logic says that if all three conditions are proven true, then our hypothesis is true. Yes, we should eat Mexican food tonight. Consultants refer to the process of devising logical arguments or issue trees to test their hypotheses as problem structuring. So if an interviewer tells you that your problem structuring is weak, it means one of two things: I call this drill-down analysis because you start at the top of your issue tree and drill down through all the logical branches and sub-branches, gathering data to factually prove or disprove that branch or sub-branch.

When you do drill-down analysis properly, you reach a logical dead end—a point at which you have clearly proven all conditions within a branch and can logically conclude that a particular branch of your argument is valid. When the latter occurs, you have effectively disproven your hypothesis, so you must consider a new one. As part of this process, you will often go back to your revised hypothesis and use drill-down analysis to analyze the remaining branches of your issue tree assuming the branches are still relevant and have not been modified due to your revised hypothesis.

Drill-down analysis is essentially a process of elimination. You drill down one branch of analysis, disqualifying or qualifying its relevance, which very much simulates the iterative problem- solving approach typical of consulting and case interviews.

See a Problem?

You do this over and over, all day long, in both consulting and case interviews. Drill down, pull up, revise the hypothesis, restructure the issue tree, drill down again Your communication must be concise, integrate detailed analysis in view of the big picture, and be action-oriented. Many analytically strong candidates, including those with PhDs in engineering and the sciences, who are new to the case interview process can solve a case effectively up until the synthesis stage.

These candidates understand the logical and quantitative aspects of the case clearly, but they sometimes find that explaining their discovery concisely to an interviewer or client can be challenging. Most candidates close a case by summarizing the steps they took to arrive at the conclusion, which typically looks like this: What I learned 1 What I learned 2 What I learned 3 What I learned 4 What I learned 5 Essentially, case interview beginners instinctively want to list every fact they discovered during the case.

For example, someone using this structure might say: Sales are down 20 percent in the Northeast. The Fortune account customer segment is growing the fastest. This is not the preferred way to synthesize. The preferred method uses the following structure: What client should do Why client should do it—Fact 1 Why client should do it—Fact 2 Why client should do it—Fact 3 Restate what client should do based on these facts In the ideal synthesis, you start your communication by stating what the client should do.

The following are examples of this type of statement: You need to shut down the Eastern region factory. To achieve your financial goals, you must enter the XYZ market immediately. You should lay off 3, employees. After making your action-oriented conclusion statement, lay out the rationale for your recommendation by citing key facts to support it. Quite often the structure of your closing argument will mirror your final issue tree structure. So here they are again: In some cases, you may be asked to use the first three tools to prepare and then give a presentation to the interviewer, as in a presentation-based case interview.

Regardless of the format of the case interview, you will use these core tools repeatedly. In the next four chapters, I outline everything you need to master these four tools. Many case interview beginners find it unnatural to decide upon and verbally state a hypothesis at the beginning of a case.

In a client engagement, you have a set number of weeks to deliver a recommendation, but in a case interview you typically have only 30 to 40 minutes either to complete the case or to get within a step or two of the finish line.

This shortens the problem-solving process by reducing the range of possible conclusions. The consensus among interviewers is that a hypothesis should be used early in a case interview, though interviewers are divided regarding precisely how early. One school of thought says you should state your hypothesis immediately at the opening of a case. The minute the interviewer explains the case background, state your hypothesis and move on to the other problem-solving tools.

Interviewers in this camp argue that stating a hypothesis when you have no background information seems overly formulaic and robotic. As a candidate, I belonged to the latter school of thought, and out of more than 60 cases, I never stated a hypothesis within the first 30 seconds.

You can ask a few background questions before stating your hypothesis, but if you take this approach, you risk forgetting to state a hypothesis at all. Going through your prep materials, I realized that understanding the problem through some cursory data gathering was the key to laying out a sound structure for the case My trusted personal rule is to consider what works with clients and then do that with interviewers.

If you meet client team members for the first time and tell them your hypothesis as to what is wrong with their company before you ever ask them any questions, they look at you with suspicion and distrust. You want to delay stating a hypothesis long enough to establish some basis for it but not so long that you forget to state it at all.

Several candidates forgot to state a hypothesis during mock interviews in my Look Over My Shoulder program www. In the debriefing sessions, these candidates had difficulty determining where to draw the line between asking a few initial clarifying questions and asking too many. I asked these candidates about this. They all said they knew what a hypothesis was and that they were supposed to state one, yet two out of three did not do so.

Stress is to blame here. This is why I emphasize not just taking notes on how to do a case but also devoting as much time as possible to practicing your case interview skills. My discussion with these candidates prompted me to create a rule for a specific time within which you should state your hypothesis. I call it the Five-Minute Hypothesis Rule. Regardless of what you have discovered by the fifth minute of the interview, I suggest that you state your best-guess hypothesis right then, lest you forget to state a hypothesis at all.

But if you do forget, the interviewer can remind you of your original intention, and this incurs less of a penalty than does forgetting entirely. The Five-Minute Hypothesis Rule will not work in a certain type of case interview. The whole case interview consists of about five sections, each lasting about six minutes and focusing on a different aspect of the same case. This type of interview jumps from one part of the case to another in no particular order.

In one section of this type of interview, the interviewer will ask for your intuition about the case, which is basically just another way of asking for your hypothesis. The hypothesis and its corresponding issue tree are a distinct prescheduled step for this interview format. During the specific time allotted for the hypothesis and issue tree, you must define both items, and you will not have the chance to revisit this step later in the interview. Interviewers commonly call this step problem structuring.

When a candidate cannot find a clean, logical, definitive way to test a hypothesis, interviewers often say the candidate has poor problem-structuring skills. To test a hypothesis, you need to create an issue tree, which identifies the key issues that, once known, will conclusively determine whether a hypothesis is true.

The term framework is used more often than is issue tree in case interview circles. A framework is an issue tree template—an initial structure for common business problems. A framework serves as a starting point for an issue tree and should be customized for every case. Thus, one commonly used issue tree deconstructs profits into component parts, enabling you to analyze each component to identify the root, or underlying, cause of the profitability problem.

Because this is a somewhat standard analysis, we call this the profitability framework. Issue tree: A logical argument or structure designed to test the validity of a hypothesis Framework: An issue tree template used to solve common business problems must be customized on a case-by-case basis The Structure of an Issue Tree As noted in Chapter 7, the term issue tree comes from the way such a logical structure looks— like a tree on its side—when diagrammed on paper: Figure 3: Issue Tree Diagram An issue tree, which is also sometimes called a logic tree, is a logical argument, the validity of which can be tested via data.

This is the essence of problem structuring using issue trees—you make a logical argument based on the hypothesis that can be easily validated with concrete data. A well-structured issue tree passes the following three validity tests: Your Hypothesis Most candidates obsess over memorizing a few key frameworks and end up focusing on the wrong things. The framework is a commonly but not exclusively used tool. The more important skill is the ability to take your hypothesis and create an issue tree or customize a standard framework that will logically test your hypothesis in this specific case.

A framework is a commonly used issue tree to test common hypotheses. But a framework has limitations—it does not cover all hypotheses for all possible situations. Many candidates jump into a framework without having stated a hypothesis, which interviewers find ridiculous: Interviewers give this type of feedback all the time, and many of my readers email me asking how to interpret it.

The only reason you use a framework or issue tree is to test a hypothesis! Remember this. Issue Tree Validity Test 2: In considering a decision your hypothesis , the issue tree lists the most relevant factors you must consider in making that decision. In the context of consulting and case interviews, the MECE principle describes how decision- making factors should be categorized in order to minimize confusion and ensure problem-solving thoroughness.

Specifically, all the information should be grouped into discrete categories, with no overlap between categories mutually exclusive , and all the categories added together should cover all possible options collectively exhaustive.

We can group customers into categories which consultants call customer segments in accordance with the MECE principle or not. Customers grouped by hobbies, for example, do not follow the MECE principle because an individual customer can belong to more than one hobby category. In contrast, customers grouped by age pass the MECE test because no individual can belong to more than one age category mutually exclusive , and the age categories cover the entire population collectively exhaustive.

The MECE principle can be applied to groups of related items, such as customers in different age brackets, and can also be used to group quantitative data. The components of revenues typically unit pricing and units sold do not appear in the cost category and vice versa.

As such, this formula fulfills the mutually exclusive criterion of the MECE test. In addition, if you combine all the factors both revenues and costs , the combination of all categories fully explains the cause of a change in profits; no other factor is missing.

Thus, this categorization fulfills the collectively exhaustive criterion of the MECE test. The MECE principle can also be applied to conceptual data.

Every possible reason for or against this decision can be grouped into one of these categories. Items that are financial in nature cannot appear in the nonfinancial category, and vice versa, so the categorization structure is mutually exclusive. In addition, if you combine all the financial and nonfinancial factors, no other reason exists as to why entering the XYZ market is a good or a bad idea.

Both categories together cover all the possible factors to be considered, so this categorization is collectively exhaustive. Because both categories are mutually exclusive and collectively exhaustive, this categorization passes the MECE test. External factors market factors such as customers, competitors, industry regulation, etc. In addition, because all factors are either external or internal, both categories combined are collectively exhaustive.

As a result, this structure passes the MECE test. When you create an issue tree or modify an issue tree template, the structure should as closely as possible adhere to the optimal MECE structure. Let me give you some examples. In a math-driven issue tree such as the profitability framework, the issue tree can maintain its MECE structure through multiple layers: In practice, we interchange units manufactured and units sold because at a strategic level the difference is considered negligible unless information exists that suggests otherwise.

Issue trees can also categorize conceptual data where the relationship among categories can be described in a mathematical formula. For example, many clients ask consultants to advise them on whether they should introduce a new product.

PDF - Case Interview Secrets

So the issue tree structure looks like this: Introducing XYZ product makes sense. Sure, customer factors, competitor factors, and company factors are fairly mutually exclusive, but the product factors could overlap with competitor factors and company factors.

You could cover products twice—once under competitors and once under company. Issue Tree Validity Test 3: What I mean by this is that the following statement should be true about your issue tree structure: In addition, it forces you to remove any factors that do not conclusively prove or disprove your hypothesis or at least substantially improve the conclusiveness of your hypothesis. Here are some examples: Apple should enter the tablet market with its iPad product.

Customer—The future size of the market is potentially very large, with extremely strong growth prospects. Competition—No competition exists, providing a competitive vacuum. Company—We have a unique competitive advantage in product design and could bring to market a unique product in this segment. Return on investment—The potential financial benefit likely will outweigh the research and development costs. For our purposes, however, we could just as easily have used the following hypothesis and issue tree instead: Apple should not enter the tablet market with its iPad product.

Customer—Market demand does not exist today, and nothing indicates that demand will ever exist. Competition—Competition in this market is intense and saturated. Company—We have no competitive advantage in product design and would be incapable of bringing a unique product to market.

Return on investment—The research and development costs would massively outweigh aggregate future revenues. What is critical is that each branch has something concrete that could be challenged with factual data, each branch tests a distinct aspect of the argument separate from the other branches is mutually exclusive , and all the branches combined cover the entire range of key issues are collectively exhaustive.

The structure of the two issue trees above indicates that if future customer demand is favorable, if no competition exists, if Apple can produce a unique offering, and if the return- oninvestment profile is favorable, then entering the tablet market with the iPad would be a good idea.

But before we use this issue tree structure, we have to ask if this logical structure is conclusive. For example, maybe site has a higher-priced product, but what about HP or BlackBerry or some other competitor? Maybe the site product costs less and does less. In this case, Apple could still win. You want to run this conclusiveness test on each branch of your issue tree to see if perhaps that branch should be defined differently or perhaps excluded entirely and on your issue tree as a whole to see if you missed something that should be included in one of the issue trees.

The Importance of Issue Tree Validity and Strong Problem Structuring If your issue tree or your framework does not pass these three tests, then the validity of your issue tree declines dramatically.

How do we improve profits? Grow sales? Enter a new market? Because clients tend to face similar problems, consultants and by proxy, candidates address these same problems with many different clients.

Case Interview Secrets Summary

Some consultants have developed frameworks to deal with these frequently occurring problems. Case interview beginners use the term framework far more often than they do issue tree, and I believe this is a mistake.

An issue tree can be used to structure any business problem, so it provides the maximum flexibility to cover the widest range of problems. In contrast, a single framework covers just one somewhat common business problem.

I think the emphasis on frameworks, at the expense of issue trees, is misplaced, because during interviews a candidate commonly will be asked to deal with a business problem that is not covered by a standard framework or one that has an unusual twist. The Curse of the Framework Robot A candidate who has memorized a few frameworks but lacks an understanding of the underlying issue tree process will struggle with applying this tool effectively.

I call this type of candidate a framework robot because he, like a robot, mechanically uses a previously memorized framework but cannot think critically as to whether that framework is relevant to a particular case situation. The framework robot attempts to force a case to fit within a framework he knows, even though that framework was never designed to solve that type of case.

This mechanical approach probably is the single biggest complaint interviewers have about candidates. This assumes the candidate actually has a hypothesis. In fact, I cover several of my favorite frameworks in the next section of this book. Recognize the situation.

Notice where the framework falls short. Create a customized issue tree that perhaps incorporates parts of the standard framework to test the hypothesis logically.

Sometimes this involves using a framework, and other times it requires using a custom issue tree. The case interview is very much a logical and critical-reasoning evaluation process; it is not a memorization test. Keep these subtle but critical distinctions in mind as we cover frameworks in more detail throughout the rest of the book.

The Relationship Between Hypothesis and Issue Tree or Framework I receive many emails asking if I have a framework for some unusual kind of case—for example, one involving a nonprofit, human resources organization design, manufacturing process improvement, or IT vendor selection. Although I focus on strategy consulting cases, people want to know if I have a framework for these unusual cases, mostly because of a dearth of information online about how to solve them.

PDF - Case Interview Secrets

I can tell just by how someone phrases one of these questions whether she grasps something very fundamental about how to approach solving a case. This obsession is misplaced. What is your hypothesis? What data do you need to disprove it? Sometimes it makes sense to use a standard framework just to learn a little more about the client situation and then form a hypothesis five to six minutes into the case.

This often involves revising the framework to fit the hypothesis better. What I am about to say is worth writing down, circling, and marking with a few stars, so remember it: The hypothesis is more important than the framework. This sentence is worth rereading a few times too.

This is why the case interview process is not a framework-driven approach but rather a hypothesis-driven one. This is yet another point worth remembering, so mark it with some stars.

This point is fundamental to case interviews, yet it takes some people awhile to master. Everyone understands it intellectually, but not everyone can master the concept to the point of automatically using it in an interview. As a candidate, you want to let the hypothesis dictate whether you use a customized issue tree or a commonly used framework—whichever approach will best test the hypothesis is the one you want to use.

The candidate has spent a lot of time preparing incorrectly. The candidate starts the case well, structures it well, and analyzes within the branches well, but he misses key insights that a candidate with this skill level does not normally miss, so the conclusion is missing certain elements.

For years, interviewers have complained about candidates being framework robots. The complaint is similar to the story about a child with a hammer who suddenly thinks that everything is a nail. A case interview is a thinking game, not a memory recall game. The process of mindlessly memorizing and recalling a frame is something I call framework vomit. All you do is swallow a bunch of frameworks in preparation for a case interview and then vomit them all back up during the case, regardless of whether they actually work for that specific case.

Interviewers complain about the proliferation of framework vomit. They want candidates who can think in an interview, not just blindly recall the 15 questions associated with a particular framework. The key is to listen to the answers you get to the first few of those 15 questions, think about what is happening formulate a hypothesis , and then decide if continuing with a certain framework would be useful.

Many candidates mentally construct an issue tree with three branches but initially tell the interviewer about the first branch only—after all, candidates tackle that one first. But this is a mistake. It makes it hard for the interviewer to grasp your overall problem-solving approach or structure. Interviewers want to see the entire problem-solving structure, not just one branch, up front to determine whether your structure is a logical test of your hypothesis. This allows the interviewer to evaluate your problem-structuring skills before you even begin the analysis phase.

You need to use a process-of-elimination approach to analyze the components of your issue tree or framework. As you work through each branch of the issue tree, you must determine if the analysis of that branch is consistent with the hypothesis. If it is, the hypothesis might be correct, so you should continue analyzing the remaining branches of the framework.

If analysis of a branch disproves the hypothesis, you should revise your hypothesis, reconsider what framework or issue tree is needed to test the new hypothesis, and then and only then continue your analysis.

Client should enter the XYZ market. Consumer demand is strong, growing, and likely to stay that way. Client has a strong, sustainable competitive advantage in the market. Competition in the market is moderate and can be beaten with modest effort.

The structure of the hypothesis and the corresponding issue tree looks very similar to the structure of a logical argument. This is not an accident. Attorneys might call this type of argument a brief. In high school, my English teacher called this an expository essay. Once you have a hypothesis and its supporting argument structure, you need to test the validity of each branch of the argument by comparing it to relevant facts, thereby testing the validity of the hypothesis overall.

As such, you cannot use this branch to validate the hypothesis. For example, you could ask the following questions: Where are they located? What do they want to download? Products—What products does the client offer versus its competitors? How do they differ? At this step in the case, you want to ask yourself constantly if the revised statement is true and under what circumstances it could be wrong. In the example, do you know which half of the market is growing and which half is shrinking?

I started off with a hypothesis and an issue tree with three components. I analyzed Component 1 regarding market growth and found its support for my hypothesis to be mixed. What I did next was subtle—yet profoundly important: I stepped back from the branch-level analysis and integrated everything from the first two branches of my issue tree from a big-picture perspective. I then reconsidered the logic of my analysis to determine under what conditions my intended revisions to the hypothesis could be false.

Logically, I knew that only half the market was shrinking. This could support a variety of hypotheses, so I realized that I needed to understand that topic better before I could draw an accurate conclusion from it. Based on that one statement about the market, I could just as easily argue for the client to enter or not to enter the market.

Does the client have a slight but temporary advantage in the shrinking portion or the growing portion of the market? In this example, neither branch of the analysis alone would indicate that one segment of the market might be a better fit than another.

You must look at the big picture, taking into account everything learned from both branches of the issue tree, to notice a potential insight. I attempted to revise my hypothesis to fit the newly discovered facts.

I had to reconcile my hypothesis with two key facts: Client should enter XYZ market. Fifty percent of the market is shrinking. Client has a modest and temporary competitive advantage.

Client should not enter XYZ market. That argument is also somewhat weak. So how can we improve it? In the U. Do this by asking yourself if any facts support either of the two hypotheses: Video demonstrations of this process are available here: Our case data so far is imperfect and incomplete, so the two extreme hypotheses seem difficult to support.

At this point, it makes sense to revise both the hypothesis and the issue tree. Refining your hypothesis numerous times as you discover more information is typical of the case interview process. Continuing with our example, we know that the two hypotheses are a stretch to support, so we need to find middle ground. The client should enter one segment of the XYZ market.

One segment of the market is growing and attractive. Client has at least a modest competitive advantage in this segment that can be strengthened over time. Competition in this segment is weak or modest. If these three supporting statements are true, they provide reasonable support that the hypothesis is correct and thus should be considered a logical conclusion.

Before we can positively conclude this, we need additional data to validate the supporting statements. For example, in the first branch, we need to prove or disprove that one segment of the market is growing and attractive. Earlier analysis showed that half the market is growing, so we can validate this branch.

Is the growth driven by a certain type of customer? A certain geographical region? A certain product category? In other words, competitors have already flocked to the best parts of the market, and the client has nothing unique to add.

By process of elimination, we must conclude that the client should not enter this segment or the market at all. Tips for the Process-of-Elimination Process Tip 1: Start with the Branch That Eliminates the Most Uncertainty First Identify all the branches you intend to analyze, and then arrange them in the order in which you plan to analyze them.

The first branch you analyze should reveal the most relevant, critical information—the information most likely to prove or disprove your hypothesis. You need to begin your analysis this way to assure the interviewer that you intend to acknowledge the other branch es. Tip 2: Use Both Quantitative and Qualitative Analyses A common mistake many candidates make is to stick to the type of analysis they are most comfortable with instead of taking a more balanced approach.

To overgeneralize, engineers and scientists tend to err on doing only quantitative analysis. English majors tend to stick to qualitative assessments and avoid doing math. As you know by now, such a problem can be driven by two categories of underlying causes: An appropriately quantitatively oriented candidate would hear that same background and seek to quantitatively verify what she has heard qualitatively.

This candidate would ask for actual pricing, unit sales, and cost data going back several years. This candidate would also get comparable data for competitors. If all the competitors were facing pricing declines, then this candidate would conclude that indeed there is an industry-wide pricing problem. Otherwise, this candidate might look elsewhere. An overly qualitative candidate would be inclined to analyze changes in both sales and costs.

After all, both contribute to the profit decline problem. Based on this alone, this approach is a mistake. The balanced candidate would instead numerically measure what percentage of the total decline in profits is attributable to the decline in sales versus the increase in costs.

Perhaps of the total decline in profits, 75 percent comes from a decline in sales and only 25 percent comes from an increase in costs. The balanced candidate would focus on analyzing the decline in sales first and possibly exclusively because it is the primary driver of the problem.

Now, if such a candidate discovered that 50 percent of the profit shortfall came from a decline in sales and the other 50 percent from an increase in costs, that candidate would have no choice but to analyze both areas. The point here is that the decision of where to focus first is supported by the numerical facts, not just the qualitative intuition. He fails to ask for qualitative data. Typically applicants with engineering backgrounds or PhDs in quantitative fields are comfortable asking for numerical data and doing math.

This candidate might further compute that the profit problem the client faces is due entirely to a difference in how the competitors price their products versus how the client does.

Quantitative analysis is very useful in answering questions regarding what is causing a particular symptom in a business and by how much. Conversely, quantitative analysis is often not useful in determining why a certain decision in a client organization was made and how certain processes e.

Tip 3: Most people new to case interviews routinely make this mistake. In other words, upwards of 75 percent of the time spent on a case is on problem definition.

And the easiest way to define the problem is to isolate the problem by using the process of elimination to identify what the problem is not. Rather than sticking to the systematic process-of-elimination approach to prove what the problem is not, they typically guess what the problem is: Maybe the client could redesign the product so it could sell better.

The candidate proposes a solution before isolating and defining the problem, basically assuming the problem is a design problem and therefore incorrectly concluding that a redesign will fix the problem. Maybe the client could partner with a Fortune company with more resources, and they could provide a joint offering of some type. Again, the candidate proposes a solution without first defining and isolating the problem. Maybe the client could implement a more generous sales commission program to improve sales.

The candidate assumes that product sales are causing the profit problem and that sales performance could be improved by providing higher incentives. Then and only then can you propose a solution. If you isolate the problem exceedingly well, usually the answer is extremely obvious, so stating your recommendation is just a formality, because both you and the interviewer know the answer.

You set up an incorrect issue tree. So the trick is to drill down far enough into the branch of an issue tree to prove your hypothesis correct or incorrect, because either conclusion is useful.

Either the client should not enter the market due to the lack of market growth, or the client should enter the market due to other favorable factors that more than offset the lack of growth. You would pick either scenario as your revised hypothesis, reformulate your issue tree to test this new hypothesis, and move on to analyze a branch of this new issue tree. Tip 4: This represents a minimally necessary amount of data to reach this conclusion.

What we do not need to analyze is whether the market is shrinking at 5 percent per year or 20 percent per year. This level of detail exceeds the minimally necessary level of data to reach the conclusion. Anytime you analyze a situation beyond what is minimally necessary, the interviewer considers that highly inefficient and counts it against you during the case interview. Tip 5: I encourage taking notes, but I discourage taking notes in a list format.

I recommend taking notes on two different sets of paper. The first set focuses on the structure of the case and should be kept very neat. It should consist exclusively of issue tree diagrams and your hypotheses. Your hypotheses and issue trees will be neatly organized in your first set of papers, so both you and the interviewer need only look at the papers to determine where you are in the structure of the case. The second set of papers is solely for you to do computations; think of it as a scratch pad.

Use this set to determine whether a particular part of the issue can be eliminated mathematically as a possible cause of the problem at hand. Given this context, you can see why interviewers ask estimation questions. Computation-Level Estimates To effectively answer estimation questions based on a set of basic facts that provide a snapshot of a particular situation, you must be able to 1 do mental math with larger numbers, and 2 round numbers intelligently.

Estimation Skill 1: Ideally, you need to be able to do these computations in your head or, at most, with only a pen and a piece of paper. The trick is to simplify the problem before you attempt to solve it.

Let me illustrate using the example above. If we draw out the equation in the order it is given, it appears as follows: So I try to move decimal points only when the math is very simple.

Looking at the following equation, I think about which type of operation multiplication or division will be less confusing to tackle. I go with multiplication. The formula is currently: All I need to do is solve the first part of the formula and split it in half in order to get the second half: Often it is much easier to solve a long series of simple math problems than a short series of complicated ones.

You may not need to break down this problem into as many simpler parts as I did, but the process of doing so is important when it comes to passing quantitative assessment tests. In the example above, I simplified the problem enough that I felt comfortable doing the math computations in my head.

You can simplify a problem in any mathematically correct way you choose. The secret here is to become accustomed to rearranging a large-numbers math problem into a simpler format before you compute anything. As with any new habit or skill, you will want to practice this math-simplification skill. You can do so by using the large-numbers math practice tool available here: Estimation Skill 2: We need a directionally correct answer only.

Pay attention not only to the math but also to my thought process and rationale for why I make certain adjustments. I start with the following formula: I need to get it to a round number. I could round down to 50 million or up to 60 million. It is very important to keep track of whether your estimate will be too high or too low.

Back to the computation, we now have: I have a decimal in there, and 17 is a hard number to work with. Well, I could round to 15 percent or to 20 percent, and both are 2. That means my estimate so far is too low.

So, if possible, in my next step I want to round in the opposite direction, up. So now I have: This is the mental thought process you want to use when estimating numbers—round numbers intelligently in an offsetting fashion. If you round down to start, you want to round up next, and vice versa. The discreet hand signals I use are as follows: Estimation Skill 3: Finding a Proxy One of the big secrets to solving extremely complicated estimation questions is to find a useful proxy.

What is a proxy? As manufacturing costs and prices decline, what will sales for cellular phones be in ? When interviewers asked me questions like these, my instinctive response was to panic, but I successfully answered these questions, passed the interviews, and got offers from all three firms that asked me those questions—Oliver Wyman, McKinsey, and Bain.

The interview process for these firms today differs from when I went through it. The examples that follow illustrate the range of difficulty you can expect when tackling a question like this. However, you should not use my personal experience as an indicator of which firms ask these types of questions in which interview round.

As a candidate, I solved many of these estimation questions without explicitly realizing what I was doing. That step is finding the proxy. When I first learned how to tackle estimation questions, most of the examples I found were about estimating a market: How many X are sold in America?

Without realizing it, I used population as a partial proxy for market size. The key to solving estimation questions is not to base your estimates on population size automatically. Instead, base your estimates on a relevant proxy coincidentally, this is population much of the time. Example 1 How many gallons or liters of gasoline does a typical filling station pump each week?

What factors correlate with how much gasoline a typical filling station pumps on a given weekday? Any thoughts? Here are mine: The average number of pumps each station has on-site The average number of cars that drive by, based on time of day more cars during commuter hours, fewer cars during off-peak hours The average percentage of pumps being used The average volume of gasoline pumped per car The average pump time per car Example 2 Assume the year is , and Motorola just invented a new technology called the cellular phone.

I nearly answered this tough question wrong. First, let me provide the context. It was my Bain final-round interview. I had done well with all the other interviewers, and this was the last question, from the last interviewer, in the last round.

I heard the interviewer ask the question, and I panicked. Furthermore, based on personal knowledge, I knew the mobile phone would ultimately succeed. But how could I prove this based on what was knowable in as opposed to what we know today?

As I often do when I panic, I stalled for time! I smiled calmly on the outside, but inside I was scratching my head. Here was my thought process: Clearly, sales will be a function of the size of the U.

I also knew that sales of this technology would skyrocket and be significant, so clearly population alone was not the best proxy. But what was?

The main issue was the price; it was just so damned expensive. So, as time progresses, technology costs will go down, prices will go down, and consumers will respond by downloading more.

And how big will the unit sales increases be? Implicitly, I was trying to identify a good proxy for how quickly consumers would download cell phone technology as prices fell. If prices fell by 20 percent, how much would unit sales increase? What would be the closest proxy? That was the vague sentiment I had in my head, and I wish I had known enough to phrase it using these terms. We have the first three years of sales data for cellular phones, and assuming we can get Motorola to estimate manufacturing costs at various product run sizes, we can triangulate the unit sales growth with the price-drop ratio for cellular technology.

We could do this by comparing the first three years of sales for cellular technology with the first three years of sales for every other major new technology. As prices of fax machines dropped by 20 percent, for example, how much did unit sales increase? What about microwave ovens? In essence, the adoption curve for cellular phones might mirror the adoption curve for other major new technologies. She ended up offering me a job the next day. In hindsight, I got lucky. I realize now that finding the proxy is the critical step in solving this and every other estimation question.

Estimation Skill 4: For example, if we look at the gas station example, we know that how much gasoline a typical filling station dispenses is correlated to how many pumps the typical station has and what percentage of those pumps is used at any given time. If every pump is used 24 hours a day, seven days a week, then the total pumping volume will be determined by how many pumps that station has on the premises.

We can be confident that this proxy sets the upper limit. We know, however, that a gas pump will not be used all day long. Sometimes pumps sit idle, so the number of pumps is a useful but imperfect proxy. Here was my rationale: Car traffic peaks on local roads during commuter hours, which in the United States are roughly 7: The likelihood that most, if not all, of the pumps at a gas station are in use during those hours is quite high.

Estimation Skill 5: Segmenting Estimates to Minimize Proxy Imperfections Once you have a qualitative sense of what makes a proxy imperfect, segment your estimate into smaller, more precise sub-estimates. In the case of the gasoline-pumping example, I created three different estimates: Conceptually, my estimate looked like this: Estimation Skill 6: Solving the Sub-estimates via Assumptions aka Guesstimating Once you have segmented your estimates to minimize the imperfections caused by a particular proxy, solve each sub-estimate.

You will typically be able to use a pen and a piece of paper but not a calculator to solve these computations. Continuing with our prior example, we would start by estimating peak hour gallons pumped. Each island has two stations. Each station has two pumps one on each side of the island. The actual pumping takes about four to five minutes, and it takes another minute to put the pump back, grab the receipt, and move my car. In a minute time period, at six minutes per car, that means each pump can fill ten cars per hour.

I know some cars take a lot more, and some of the smaller fuel-efficient cars take less. We also have to factor in those cars whose tanks are not completely empty when being filled.

I mentioned earlier that the typical island has all eight pumps running at roughly 80 percent utilization during peak hours. So my guess is that the utilization rate is around 25 to 50 percent during off-peak hours, and we can simplify that by saying 40 percent of the pumps are utilized. This works out to be exactly half of the 80 percent utilization rate during peak hours. That works out to 4, gallons: Practice Makes Perfect!

I hope this chapter has demystified estimation questions and shown you a process you can use to tackle these questions in your interviews. To be extremely proficient, you must practice the component-level math of large numbers and tackle estimation questions from top to bottom. To practice the component-level math of large numbers, including rounding, go here: You can submit your answer and compare it to the answers of several hundred other people and to my answer key.

This sample question and its answer key can be found here: If you demonstrate that you have mastered the skills that consulting firms want, you will do well in every type of case interview. To appreciate why this is so and its implications for you, you must recognize why consulting firms do certain things during the recruiting process.

They often report back to me on what advice from me they found helpful in passing the case interview. The information that follows is that advice. Why Consulting Firms Do What They Do Candidates often send me feedback about my explanations as to why consulting firms do what they do.

Why do interviewers ask certain questions? Why do they use certain assessments? Why do they challenge your answers so aggressively? This is important because what consulting firms do in the recruiting process changes yearly and sometimes from one interviewer to another. But why they do what they do has not changed in decades. If a candidate has enough interviews, he will encounter an exotic question, a new twist, or an extremely unusual case. With that in mind, consider the following implications.

First, if you hate case interviews, you likely will hate the job. Being a consultant is like going through a case interview every day of your career. Second, case interviews involve estimation questions because clients ask estimation questions all the time.

I was asked plenty of estimation questions when I was a candidate, but once I started working at McKinsey, I ended up answering more estimation questions as a consultant than I ever did as a candidate.

Third, everything that happens in a case interview happens because it simulates some aspect of the on-the-job experience. When the interviewer asks you a random question in an interview, stop Instead, think like a consultant: Proving Yourself as a Consultant The consulting team and firm must prove themselves early in their relationship with a client.

When you start working with a new client, some individuals within the client organization may express skepticism about the value you and the consulting firm can bring. So how can you prove your worth and be taken seriously?

Develop data-supported conclusions especially counterintuitive ones that lead the client toward a different set of decisions. Often the client or certain members of the client organization are looking to discredit you so they can get back to running the company.

In that case, a semi-hostile client is looking for you to screw up somehow.

The two most common screwups are the following: Clients often interpret nervousness as a lack of conviction about a particular recommendation, which is why answering a case perfectly but nervously will get you rejected. For example, if a consultant were to recommend nervously that the client lay off 2, employees, the client would second-guess the recommendation. Even if the recommendation were percent correct, the client would sense some degree of hesitation, uncertainty, or reservation from the consultant based on how the message was delivered, not the content of the message itself.

As a result, consulting firm interviewers assess the level of confidence you project while solving a problem analytically. As I mentioned, every aspect of the interview process happens for a reason, and most often that reason is to simulate some aspect of the on-the-job experience.

This is the vital point to keep in mind as I cover in the next chapter the specifics of what interviewers are looking for and explain why they look for the things they do. Most candidates start the case interview learning process by seeking out rigid rules to follow. They think that if they can learn every interview format and every type of case question, they will be prepared. But in addition to preparing for your case interview, keep in mind a simple principle: Interviewers look for candidates who seem like colleagues already.

The tone of the interview switched from evaluative to collaborative. In essence, the candidate who stands out the most in an interview is the one who acts like a consultant already.

To me, a case interview is no different than a team meeting with the partner. Clients demand certain things of consulting firm partners, and partners expect their consultants to offer what the client demands. Consultants, who also serve as case interviewers, in turn demand these skills from the candidates they interview. How the Consulting Business Works At the heart of every consulting firm are two groups of people: At McKinsey, we never discounted fees; we over-delivered, did extra work for clients for free, and worked harder.

How Client Billing Works Each consultant on the team has two billing rates: These rates are rarely published internally or if published, not published very widely , but they exist to keep track of whether a client engagement is profitable.

When partners and managers look to staff their teams, they tend to look for the strongest contributors at every cost point— who provides the most value per dollar—in order to deliver higher-quality work at lower costs and maximum profit. An interesting problem arises with brand-new first-year consultants, who often contribute negative value to an engagement team. More-experienced consultants must double-check at an extremely detailed level anything these consultants work on.

Thus, the additional time required to manage new consultants largely offsets their contributions. In these situations, the cost of the first-year consultant on his or her first assignment is billed to the training department.

This entire process, convoluted as it seems, exists in large part because first-year consultants have not yet proven their ability to solve problems independently. The Value of an Independent Problem Solver Let me share a story with you that illustrates why consulting firms value independent problem solvers. In my third month at McKinsey, when I was only 22 years old, I was assigned to a client based in New York that had a small division in Cleveland, Ohio.

The entire team, including me, was based in New York. All the other consultants on the team were married and had kids. My manager went to Cleveland weekly, at most, for just one day, mostly to build client relationships and attend key meetings. The rest of the time I was on my own. So, what was the project? See you in a few months. Plus, I had just 90 days of consulting experience then—not much more skill than what I had during the interview process.

When a consultant interviews you, she is wondering, Can I drop you off with a division of a Fortune company by yourself, with little to no supervision? Can you handle the client, solve its problems, and in the process make the firm look good?

She is also thinking, Do I want you on my team right now? Phrased differently, that interviewer is asking herself, Will you be an independent problem solver fairly quickly, or will I have to babysit you for the next two years of your career?

My manager on the Cleveland project became a partner at McKinsey. Initially he tried to His ability to do partner-level work was directly related to my ability and the ability of the other consultants on the team to be an independent problem solver. This is exactly why managers need independent problem solvers—their own career progress very much depends on it. Doing as Little as Possible vs. Boiling the Ocean As I mentioned previously, a routine problem in consulting is not having enough consultants to do all the work the client wants done.

Clients want you to do as much work as possible for the lowest possible cost, whereas consulting firms want to charge as much as possible while doing as little work as possible and still delighting the client.

You end up having to make difficult choices in how you manage your time. In this context, many consultants use the phrase boiling the ocean to describe how not to prioritize your client work. As the metaphor goes, if you want one cup of hot water, there are two ways you can get it: The first is math questions — problems that involve some sort of arithmetic, percentage calculation or interpretation of data. But the key is to answer them as quickly as possible, which takes lots of practice.

No matter how much of a math whiz you are, your brain is a muscle that needs regular training to perform optimally. But how? One way is to practice old interview questions. For instance, you can become faster by searching for the types of questions asked by firms like McKinsey or the Boston Consulting Group. Case Interview Secrets Key Idea 2: Complicated problems can be solved quickly by breaking them down and using rounding.

So what exactly is a computational question? Answering such problems successfully requires precise arithmetic using large numbers — no easy task. But as a matter of fact, performing complex mathematical equations in your head can be done precisely and quickly if you break them up. What is your total revenue? This is helpful because rough estimations will let you perform equations much more quickly. So, take a variation of the above question in which the market is 42 million customers, the market share is Then, round In fact, arriving at the correct answer is more or less irrelevant.Through this promotion I learned how consulting firms work and how consultants think.

Once you have a hypothesis and its supporting argument structure, you need to test the validity of each branch of the argument by comparing it to relevant facts, thereby testing the validity of the hypothesis overall. You can use the exact same tools to tackle any type of case interview: If you demonstrate that you have mastered the skills that consulting firms want, you will do well in every type of case interview. Clipping is a handy way to collect important slides you want to go back to later.

My manager went to Cleveland weekly, at most, for just one day, mostly to build client relationships and attend key meetings. This gives you a great view of how to create a discussion on-the-fly, which is what you have to do in a case interview. At McKinsey, he was rated in the top 10 percent of consultants worldwide in his cohort.