FRAMING THE PROBLEM

30
1 FRAMING THE PROBLEM T he ability to frame business problems to make them suscepti- ble to rigorous fact-based analysis is one of the core skills of a McKinsey consultant. More than that, it is the hallmark of a McKinsey-ite: if you can’t solve problems in a structured, hypothesis-driven manner, you’re unlikely to make it through the door of the Firm. 1 Data Intuition Managing • Team • Client • Self Presenting • Structure • Buy-in Analyzing • Framing • Designing • Gathering • Interpreting

Transcript of FRAMING THE PROBLEM

1

FRAMING THE PROBLEM

The ability to frame business problems to make them suscepti-ble to rigorous fact-based analysis is one of the core skills of

a McKinsey consultant. More than that, it is the hallmark ofa McKinsey-ite: if you can’t solve problems in a structured,hypothesis-driven manner, you’re unlikely to make it through thedoor of the Firm.

1

Data

Intuition

Managing• Team• Client• Self

Presenting• Structure• Buy-in

Analyzing• Framing• Designing• Gathering• Interpreting

01 (001-030B) chapter 01 8/27/2001 3:33 PM Page 1

2 The McKinsey Mind

The McKinsey problem-solving process begins with the use ofstructured frameworks to generate fact-based hypotheses followedby data gathering and analysis to prove or disprove the hypothesis.A hypothesis greatly speeds up your quest for a solution by sketch-ing out a road map for research and analyses that will guide yourwork throughout the problem-solving process, all the way to thepresentation of your solution. Given the value of this methodologyto Firm alumni in their post-McKinsey careers, we begin with anexamination of ways to adapt that process to businesses beyondthe Firm.

In this chapter, we will show you how to apply structure toyour business problems and how to go about devising initialhypotheses that will speed up your own decision making. Becausestructure is the basis for the McKinsey problem-solving process,let’s start there.

STRUCTURE

Although McKinsey & Company often uses the term fact-based

to describe it, the McKinsey problem-solving process begins notwith facts but with structure. Structure can refer to particularproblem-solving frameworks or more generally to defining theboundaries of a problem and then breaking it down into its com-ponent elements. With either approach, structure allows McKinseyconsultants to come rapidly to grips with the issues facing themand enables them to form initial hypotheses about possible solu-tions. The benefits of structure transfer readily beyond the confinesof the Firm, as our alumni have shown. The facts, as we will see,come later.

01 (001-030B) chapter 01 8/27/2001 3:33 PM Page 2

Framing the Problem 3

THE McKINSEY WAYLet’s start by summarizing the ways McKinsey consultants applystructure to their business problems.

Feel free to be MECE. Structure is vital to McKinsey’s fact-based problem-solving process. For McKinsey-ites, structure is lessa tool and more a way of life. One Firm alumnus summed up hisMcKinsey experience as “Structure, structure, structure. MECE,MECE, MECE.” The concept of MECE (pronounced “mee-see”and an acronym for Mutually Exclusive, Collectively Exhaustive),is a basic tenet of the McKinsey thought process. Being MECE inthe context of problem solving means separating your probleminto distinct, nonoverlapping issues while making sure that noissues relevant to your problem have been overlooked.

Don’t reinvent the wheel. McKinsey has leveraged its experi-ence with structured problem solving through numerous frame-works that help its consultants rapidly visualize the outlines ofmany common business situations. Your organization may have itsown frameworks, and you should take advantage of them if pos-sible. Otherwise, develop your own problem-solving tool kit basedon your experience.

Every client is unique. Frameworks are not magic bullets.McKinsey-ites know that every client is unique. Simply trying tosqueeze every organization’s problems through the appropriateframeworks will only get you so far. If anything, this lesson is dou-bly true for McKinsey-ites once they leave the Firm.

LESSONS LEARNED AND IMPLEMENTATIONILLUSTRATIONSHow does McKinsey’s structured problem-solving approach farebeyond the specific conditions of the Firm? Extremely well. Our

01 (001-030B) chapter 01 8/27/2001 3:33 PM Page 3

4 The McKinsey Mind

discussions with McKinsey alumni have led us to several specificconclusions about the suitability and adaptability of structuredthinking:

• Without structure, your ideas won’t stand up.• Use structure to strengthen your thinking.

Let’s see what these lessons look like in practice.Without structure, your ideas won’t stand up. Think about

your company and the way you and your colleagues formulate andpresent business ideas. Do you use a consistent structure or at leastemphasize the need for internal coherence and logic in your prob-lem solving? Or do people usually arrive at decisions ad hoc,without a recognizable structure or factual support? WhenMcKinsey-ites exit the Firm, they are often shocked by the sloppythinking processes prevalent in many organizations.

Most of us are not blessed from birth with the ability to thinkin a rigorous, structured manner; we have to learn how. Unfortu-nately, that skill is not part of most university curricula, and fewcompanies have the resources or the inclination to teach it to theiremployees. McKinsey and some other strategy-consulting firms areexceptions to this pattern. Even some of the most highly regardedcompanies in American business don’t always stress structuredproblem solving, as Bill Ross learned when he joined the Trans-portation Division of General Electric:

GE people move quickly when new situations arise. It’s partof the culture. The mind-set seems to be “once we have iden-tified an issue, let’s wrestle it to the ground and movequickly,” and they’re great at doing it. Rarely do people takethe time to examine the issue and develop a clear plan ofaction. The structured approach really surprises a lot of peo-

01 (001-030B) chapter 01 8/27/2001 3:33 PM Page 4

ple. I think just focusing people on that has allowed me toadd value.Many highly successful organizations don’t apply structured

thought even to their core competencies, as Paul Kenny describesat GlaxoSmithKline:

From a scientific point of view, a lot of the research organi-zation is rather serendipity led: you invest in research, youmay have a direction, but often that direction will changeas a result of information you find. Some of the best drugson the market today were found more by luck than bydesign. Then, thinking back, we realize that we could haveredesigned these clinical trials in a way to shape the productmore appropriately for the market. There are concrete exam-ples of ways to increase value by making more-commercialmarketing decisions earlier on in the pipeline, and designingproducts from the very beginning to have the right charac-teristics, rather than just letting them evolve from the R&Dpipeline however they emerge.

If structured thinking is hard to find at GE and GlaxoSmith-Kline, two of the world’s most respected and successful companies,one can imagine that it may be a pretty rare coin in manyorganizations.

Further complicating matters, the corporate cultures of someorganizations have been imbued with the wrong types of structure.In another example from GlaxoSmithKline, a linear, deductivethought process got in the way of sound decision making:

We have a project leader who wants to switch his drug fromits current twice-a-day formulation to a once-a-day formu-lation. The drug is at an early stage in research, and it’s a

Framing the Problem 5

01 (001-030B) chapter 01 8/27/2001 3:33 PM Page 5

standard rule that once a day is better than twice a day. It’seasier for people to take, so ultimately there’s a market-driven push to develop the once-a-day dosage. He has pre-sented this as a binary decision: either we invest in it, or wedon’t. But the idea of thinking through the various optionsthat might really be possible in a MECE way, opening out allthe possibilities and then considering or rejecting them inde-pendently, hasn’t really occurred to him.

In fact, there are a number of options, including launch-ing as a twice-a-day formulation, getting through a lot of thedevelopment risk that way, then moving to a once-a-day for-mulation once the drug has proved efficacious and mar-ketable. Taking the all-or-nothing approach may notnecessarily be the best way to create value; the incrementalsales may not be worth the incremental costs and risks.

Between inappropriate thinking processes and the completeabsence of structured thought, there appears to be a lot of room forsomeone with a McKinsey Mind to add value.

Use structure to strengthen your thinking. In all sorts ofplaces—whether huge corporations, new economy start-ups, oreven nonbusiness organizations such as nonprofits and govern-ment agencies—McKinsey-ites have been able to apply structuredthinking in ways that allow them to add value to their organiza-tions. For example, making strategic decisions requires under-standing the capabilities of your organization and how to utilizethose capabilities to maximize performance. That’s what Jim Ben-nett did during his tenure as chairman of retail banking at KeyCorp.:

I became chair of the retail bank at a time when we reallyneeded to grow our operation. It was a third of the company,and we had to grow at 10 percent per year for the rest of thecompany to do well. I had to determine whether that was

6 The McKinsey Mind

01 (001-030B) chapter 01 8/27/2001 3:33 PM Page 6

possible or not. Of course, this depended on understandinghow good we were. The only way I could come to grips withthat was to lay out an issue tree.* By the time I was done, Ihad a MECE issue tree with all the branches covered byyes/no questions. That proved very useful to me as the linemanager and chief strategist of Key Corp.’s largest businessin making sure that we were on the right track with our per-formance improvement program.

I did it myself and then exposed others to it and the gen-eral idea behind it. The issue tree in and of itself probablystrikes people as a bit “consultanty,” but when I’ve been ableto translate it into a communicable message, it’s never failedme in any setting, anywhere.

Another example of the successful application of McKinseyframeworks in a large organization comes from Bill Ross, who wasthen at GE:

The biggest “framing the problem” issue I found involvedthe big question, “Do we know where we are going in thelong term and have we developed our growth strategy?” Theanswer in many cases was no. I worked individually withsome of the other general managers and then actually usedMcKinsey to put together a workshop with the senior lead-ership team to talk explicitly about our growth strategy. Thisallowed me to start feeding them information and exposethem to some of the previous frameworks that I learned atMcKinsey. When they saw those, it triggered light bulbs intheir heads.

Large, cash-rich corporations might seem the ideal place toapply McKinsey techniques. After all, most of McKinsey’s clientsfit that description. What might surprise you, however, is how

Framing the Problem 7

* We will discuss issue trees later in this chapter.

01 (001-030B) chapter 01 8/27/2001 3:33 PM Page 7

effective these same techniques can be in the short-of-cash, short-of-time, short-of-people environment of a New Economy start-up,as Omowale Crenshaw discovered at Africa.com, a Web portal forthe African continent:

We had to survey the marketplace and decide how todevelop products and services for our particular target mar-kets: African ex-pats and the “Africa-interested.” Thatmeant analyzing a number of industries such as Africanwine, or African home-decorative accessories, furniture, andart, and making a decision as to which of them would be suf-ficiently attractive to our target markets. By allowing us tocome to grips quickly with the market sizes, the competitiveenvironment, the key players, etc., the structural frameworksthat I learned at the Firm helped us decide which of thesemarkets made sense for us.

Structuring your thinking can add value outside the confines ofthe business world. Sylvia Mathews was deputy chief of staff toPresident Clinton, so she should know:

Problem solving at the federal government level tends to be alittle more complicated than in the business world, in that itinvolves things that are less tangible than valuation of com-panies, profit, loss, etc. But the same techniques still apply.When I was in charge of the State of the Union production in1996, in August (the President delivers the State of the Unionaddress in January), I started by doing something that Icalled the Pillars Project. It covered every area of the Stateof the Union and put all our policy examples together in thesame framework and with the same approach to show whatwe were going to try to achieve over the second four years.We then assembled them into documents that the Presidentand Vice President could respond to over their vacations.

8 The McKinsey Mind

01 (001-030B) chapter 01 8/27/2001 3:33 PM Page 8

We framed the issue very clearly: What is the problem?What are its dimensions? What are we going to do? And itlaid out a number of things that we could try in a limitedway to increase our chances of success. We covered the var-ious issues stemming from each problem: you might wantto do something, but is it achievable, do we have theresources financially, can we get the congressional support,and what are the political ramifications?

Now that you’ve seen how useful structured thinking can bein almost any sort of organization, let’s discuss how you can applyit in your own business and career.

IMPLEMENTATION GUIDANCEAs we’ve seen, structured thinking is an important element in anybusinessperson’s problem-solving arsenal. How should you use thisweapon? First, you must understand that structure doesn’t existin a vacuum; you have to wield it with a goal in mind. In the con-text of framing and solving business problems, your goal is tobring order out of chaos.

Today’s executives and entrepreneurs have access to far moreinformation than they can possibly use. They can manage this sur-feit of data only by filtering out all but the most relevant facts. Theappropriate structured framework will allow you to do this muchmore efficiently, thereby increasing the probability that you willarrive at a solution in a reasonable amount of time—and add valueto your business. As Omowale Crenshaw observes:

One of the things that was very clear from my McKinseyexperience—and this definitely applies in an entrepreneur-ial environment—was that the skill set that I have allows meto make some sense of ambiguity, of all the possible paths we

Framing the Problem 9

01 (001-030B) chapter 01 8/27/2001 3:33 PM Page 9

could take. We have limited resources and limited funds, sowe can’t go everywhere; we have to start following thesepaths one at a time. A framework helps you prioritize youroptions. We save so much time and energy by not goingdown the wrong path. That’s the key. Not necessarily know-ing what the right path is, but not going too far down thewrong one.

The role of senior management in this is to structure “reality”in order to make it easily graspable. Executives do this by defin-ing the scope of the problem at hand in order to see all its ramifi-cations—the links to other factors and the whole scope ofconsequences. They can then disregard unimportant factors andconcentrate on prioritizing the options available to the organiza-tion. This allows them to communicate the (potentially complex)problem and its solution in easily understandable terms, to make itclear to those who need to execute management’s directives.

We will examine gathering the data and communicating thesolution in later chapters, so let’s turn now to defining and simpli-fying the problem. In the generic approach to framing the problem,McKinsey-ites put this concept into practice by breaking the prob-lem before them into its component elements. Why? In most cases,a complex problem can be reduced to a group of smaller, simplerproblems that can be solved individually. The problems McKinseyhandles are either extremely complex (“How can we maintainshareholder value in the face of competitive pressure and uniondemands when our core market is shrinking?”) or stated sobroadly as to be insoluble without further clarification (“How dowe make money in our industry?”). Separating out the individualpieces of the problem will make it easier for you and your team toidentify the key drivers of the problem (see Chapter 2) and focusyour analysis accordingly.

10 The McKinsey Mind

01 (001-030B) chapter 01 8/27/2001 3:33 PM Page 10

Framing the Problem 11

This technique works not only for business problems, but alsofor complex problems in other realms, such as politics. Forinstance, Francesco Grillo, formerly with McKinsey’s Rome office,is now a public-sector consultant and policy adviser to the Italiangovernment. He used these same techniques with great success onproblems such as unemployment in the European Union, reform ofthe Italian electoral system, and the evaluation of the economicimpact of programs funded by the European Commission.

The most common tool McKinsey-ites use to break problemsapart is the logic tree, a hierarchical listing of all the components ofa problem, starting at the “20,000-foot view” and moving pro-gressively downward. As an illustration, let’s look at that fine oldblue-chip firm Acme Widgets. Let’s suppose that Acme’s board hascalled your team in to help answer the basic question “How canwe increase our profits?” The first question that might pop intoyour head upon hearing this is, “Where do your profits comefrom?” The board answers, “From our three core business units:widgets, grommets, and thrum-mats.”

“Aha!” you think to yourself, “that is the first level of our logictree for this problem.” You could then proceed down another levelby breaking apart the income streams of each business unit, mostbasically into “Revenues” and “Expenses,” and then into progres-sively smaller components as you move further down the tree. Bythe time you’ve finished, you should have a detailed, MECE mapof Acme Widgets’ business system, along the lines of Figure 1-1(page 12).

Remember, when you are drawing a logic tree, there may beseveral ways to break apart a problem. Which one you choose willaffect the way you view the problem and can either reveal orobscure critical issues for your team. For instance, instead of draw-ing your logic tree of Acme Widgets with an organizational hier-archy (by business unit), you might want to look at it from afunctional perspective (production, sales, marketing, research, ful-

01 (001-030B) chapter 01 8/27/2001 3:33 PM Page 11

fillment, etc.). This perspective could point your team in other,potentially useful directions. Just make sure, whichever view youtake, that your logic tree is MECE, so that you miss nothing andavoid confusion.

In a real-life example of a logic tree at work, Naras Eecham-badi, when he went to First Union after leaving McKinsey, had toput together the business case for his customer information man-agement unit in order to get funding approved by the president ofthe corporation:

The question boiled down to, “If we are going to produce areturn on investment for the company based on how webuild and leverage customer information, where are thesources of revenue and profit? Where is the money going tocome from?” I came up with a MECE breakdown showinghow we could make money by adding or selling more prod-ucts and generating more revenue from existing customers,cutting costs to serve the existing customers, reducing theattrition of existing customers, or being much more effec-tive and efficient in bringing on new customers. And I was

12 The McKinsey Mind

Figure 1-1. Acme Widgets Logic Tree

AcmeWidgets

Grommets

Revenue Europe

Asia

Expenses

Widgets

Thrum-mats

NorthAmerica

WidgetSales

WidgetLeasing

WidgetService

01 (001-030B) chapter 01 8/27/2001 3:33 PM Page 12

Framing the Problem 13

able to bounce around the problem and say for each of thepieces, “How much more money can be expected? What isthe economic benefit? And at the end of the day, what is thecost of doing all these things?” That’s how I built my busi-ness case, by breaking down and reconstructing the problem,figuring out where the pieces were.

The logic tree is one framework among many that McKinseyconsultants use and an especially popular one for them to takewith them when they leave the Firm. Like any framework, it helpsyou clear away the clutter of a complex problem and bring orderout of chaos by building a simplified representation of the realworld. Jeff Sakaguchi, who left McKinsey’s Los Angeles office tobecome a partner at Accenture, sums up the usefulness of theframeworks he learned at the Firm:

The whole framework-driven approach is really trying tothink about, “How could you organize this?” Every frame-work—all the way down to the simple two-by-two matriceswe use day in, day out—are an attempt to frame the problemaround some nifty set of three, four, or five balls or boxesor triangles or whatever you need to create a simple repre-sentation of a complex problem. McKinsey was masterfulat that. I’ve really tried to adapt that for my work.

When employing logic trees or any framework, bear in mindyour eventual audience. Tailor your presentation of that frame-work accordingly. As Bill Ross discovered at GE:

I find that, although frameworks work great internally atMcKinsey, when you go outside McKinsey you have to becareful about their use. Many people will see a frameworkand automatically start getting defensive. We heard it a lot atMcKinsey: “Oh, you’re taking an approach that you used on

01 (001-030B) chapter 01 8/27/2001 3:33 PM Page 13

14 The McKinsey Mind

somebody else’s problems and trying to apply it to me. Myproblems are different.” We knew that wasn’t the case; wewere just trying to get the thoughts started, giving ourselvesa systematic checklist for what the key issues are and howto present those key issues. You have to be careful aboutintroducing frameworks because they can carry a fairly neg-ative connotation, especially if they are overused. Instead ofreusing an old framework, use the concepts from the frame-work to generate new ideas that help solve the problem athand.

Finally, remember that structure is only the beginning. You stillneed to develop a strong hypothesis, devise and perform the rightanalyses to draw your conclusions, and communicate those con-clusions effectively. We will address these issues further on in thebook, beginning in the next section of this chapter, with formulat-ing the initial hypothesis.

EXERCISES• If you can, think of some frameworks that are commonly

used in your business or that you learned elsewhere. Canyou apply them to your current work? If not, how couldyou apply them?

• Look at your organization. Can you lay out your sourcesof profit in a MECE logic tree? How about the process bywhich you generate products or deliver services?

• Think of a common, but complex, nonbusiness process,say, a wedding or a vacation. Can you come up with aMECE structuring of all the tasks that need to be done inorder for this process to work? What are the key elementsof the process (e.g., for a wedding, getting the guests there

01 (001-030B) chapter 01 8/27/2001 3:33 PM Page 14

on time, making sure the groom shows up)? Write themout in the form of a logic tree. Can you come up with a dif-ferent grouping—say, by responsibility—that is stillMECE?

HYPOTHESIS

Having reduced the problem to its essential components throughthe use of appropriate frameworks, you are ready to embark on thenext step in the process of framing it: forming a hypothesis as to itslikely solution. McKinsey believes, and the experiences of McKin-sey alumni demonstrate, that using an initial hypothesis to guideyour research and analysis will increase both the efficiency andeffectiveness of your decision making.

THE McKINSEY WAY As we did with structure, let’s begin our exploration of usinghypotheses by recapping the relevant principles espoused byMcKinsey.

Solve the problem at the first meeting. McKinsey-ites learn thatit is much more efficient to analyze the facts of a problem with theintent of proving or disproving a hypothesis than to analyze thosefacts one by one to determine which answer they will eventuallyprovide. For a start, a hypothesis provides you and your team witha problem-solving road map that will lead you to ask the rightquestions and perform the correct analyses to get to your answer.A good hypothesis will also save you time by pointing out poten-tial blind alleys much more quickly and allowing you to get backto the main issues if you do go down the wrong path.

Framing the Problem 15

01 (001-030B) chapter 01 8/27/2001 3:33 PM Page 15

You generate your initial hypothesis by drawing conclusionsbased on the limited facts that you know about the problem athand without doing a lot of additional research. For a consultantnew to the industry in question, this might mean spending a fewhours reading press articles and annual reports; someone withplentiful industry experience might just jot down a few preliminarythoughts. Ideally, you would then spend an hour or two meetingwith your teammates and hashing out some likely answers to theproblem.

Your next step is to figure out which analyses you have to per-form and which questions you have to ask in order to prove or dis-prove your hypothesis. One way to lay out these questions is withan issue tree. The issue tree, a species of logic tree in which eachbranch of the tree is an issue or question, bridges the gap betweenstructure and hypothesis.* Every issue generated by a frameworkwill likely be reducible to subissues, and these in turn may breakdown further. An issue tree is simply the laying out of issues andsubissues into a MECE visual progression. By answering the ques-tions in the issue tree, you can very quickly determine the validityof your hypothesis.

Proper prior preparation. McKinsey teams rely on brainstorm-ing to develop and test their initial hypotheses. BrainstormingMcKinsey-style, however, requires that all the team members cometo the meeting prepared, having absorbed all the facts currentlyknown to the team and having spent some time thinking abouttheir implications. Sometimes, especially for team leaders, it helpsif individuals have their own initial hypotheses already developed,so that the team can bat them around, but it’s not essential. Justdon’t come into the meeting thinking you know the “answer.” Beprepared to learn.

16 The McKinsey Mind

*We will detail the distinctions between the logic tree and the issue tree later in this chapter.

01 (001-030B) chapter 01 8/27/2001 3:33 PM Page 16

In a white room. Brainstorming is about generating new ideas.Check your preconceptions at the door. Everyone in the meetingmust be able to speak his mind and share his knowledge. For yourbrainstorming sessions to succeed, you should follow these rules:First, there are no bad ideas. Second, there are no dumb questions.Third, be prepared to “kill your babies” (i.e., to see your ideas getshot down, and to pull the trigger yourself if necessary). Fourth,know when to say when; don’t let brainstorming drag on past thepoint of diminishing returns. Last and most important, get it downon paper.

The problem is not always the problem. Every consultant facesthe temptation of taking the client’s diagnosis of his problem atface value. Resist this temptation. Just as a patient is not alwaysaware of the meaning of his symptoms, so are managers sometimesincorrect in their diagnoses of what ails their organizations.

The only way to determine whether the problem you have beengiven is the real problem is to dig deeper, ask questions, and get thefacts. A little skepticism early on in the problem-solving processcould save you a lot of frustration further down the road. What’smore, you will be doing your client a service by getting to the realproblem, even if, sometimes, your client doesn’t like it.

LESSONS LEARNED AND IMPLEMENTATIONILLUSTRATIONSFor McKinsey alumni, hypothesis-based decision making hasproved extremely portable. It doesn’t require a lot of resources toimplement; it can be done in teams but, if need be, also on one’sown; and it is applicable across a wide spectrum of problems. Ourquestioning of former McKinsey-ites has produced two good rea-sons why you should rely on an initial hypothesis in your ownproblem-solving efforts:

Framing the Problem 17

01 (001-030B) chapter 01 8/27/2001 3:33 PM Page 17

• An initial hypothesis will save you time.• An initial hypothesis will make your decision making more

effective.

An initial hypothesis will save you time. Most people, whenfaced with a complex problem, will start at the beginning andwade through all the data until they come to the end—the solution.This is sometimes referred to as the deductive approach: if A, thenB; if B, then C; . . . if Y, then Z. When you form an initial hypoth-esis, you leap all the way to Z, and it’s easier to work your waybackward from Z to A. One simple example of this is a pen-and-paper labyrinth or maze, the kind you sometimes see in the Sundaycomics or in puzzle books. Anyone who plays with these can tellyou that it is easier to solve the maze by tracing the route from thefinish to the start rather than starting at the beginning. One rea-son for this is that by already knowing where your solution is, youeliminate a lot of paths that lead to dead ends.

Forming an initial hypothesis will allow you to work throughthe labyrinth of your business problem more quickly. It saves youtime partly because it allows you to start drawing conclusionsbased on limited information—which, at the beginning of theproblem-solving process, is usually what you have. This holdsespecially true when you are trying to break new ground wherenobody has the information you seek, as Omowale Crenshaw dis-covered when figuring out how to open up Africa to E-commerce:

Sometimes at McKinsey we had the luxury of leveraging somuch data—the whole analysis paralysis—that we didn’t doanything, nor did our clients. When we were starting ourWeb portal, we had to figure out what mattered when wereally didn’t have enough data on one side or the other. Wejust had to say, “OK, realistically, what do we know aboutthe largest three or four or five markets? What are our

18 The McKinsey Mind

01 (001-030B) chapter 01 8/27/2001 3:33 PM Page 18

guesstimates about them?” We were figuring it out on theback of an envelope, trying to be mostly right versus pre-cisely wrong, and making some hypotheses from that. Wewould say, “OK, if we assume that the market size is X,what do we have to believe?”

Then the process became iterative: “We think the marketsize is X, and if the market size is X, then Y must be true,” sowe went and looked at Y. As we started doing that, itbecame much more apparent that we were on the righttrack. We’re still struggling with the actual size of the mar-ket, but we feel much more comfortable that we’ve done theactual due diligence as it relates to tapping into any andevery resource we could think of.

Finally, an initial hypothesis saves you time by forcing you andyour team to focus only on those issues that can prove or disproveit. This is especially helpful for those who have trouble focusingand prioritizing. You may even know of a few such individuals inyour own organization.

An initial hypothesis will make your decision making moreeffective. Not only does the hypothesis-driven approach make yourproblem solving faster and more efficient, it allows you to assessmultiple options quickly. As a result, your decision makingbecomes more flexible and therefore more effective. As CEO of abrand-name consumer goods manufacturer, McKinsey alumnusBob Garda, now a member of the marketing faculty at the FuquaSchool of Business at Duke University, used a strong initial hypoth-esis that went against his company’s conventional wisdom to turnaround its core business:

We were selling products that had been around for 20 yearsand faced a lot of price pressure from Wal-Mart, Kmart, andTarget—our three biggest customers. They kept threatening

Framing the Problem 19

01 (001-030B) chapter 01 8/27/2001 3:33 PM Page 19

to go to suppliers in China or India unless we lowered ourprices. We had four alternatives: (1) reduce costs to matchChina and India, (2) buy from China and India and resell toour customers, (3) introduce new products (one of whichwas close at hand), or (4) do a combination of the above. Myhypothesis was that we could best minimize the price pres-sure by introducing new products. Sure enough, when wewalked into the Big 3 to introduce the new product, theywere excited; we could practically charge whatever wewanted. After that, they were much less concerned aboutbeating us up on price, even on the established products, aslong as we kept generating new products. Therefore, thehypothesis worked out.

Bob compared his hypothesis against other options available tohis company:

We could have taken one of the other approaches and triedto cut costs to beat India and China. In fact, several of ourkey managers thought cost reduction was the only answer, asit had been in the past. Well, good luck; you can’t beat Chinaand India on costs with U.S.–made products. Naturally, costreduction was part of the answer for the long term; welaunched a cost reduction effort, but we never got down toChinese or Indian costs.

The other option, purchasing from China and India andreselling to Kmart, Wal-Mart, and Target, while popularwith a small faction of management, made no sense to me.All we would be doing was setting up a distribution systemfor the Asian manufacturers and, once established, theywould go directly to the buyers and cut us out. Because ofconstant price pressure, this option will continue to be on the

20 The McKinsey Mind

01 (001-030B) chapter 01 8/27/2001 3:33 PM Page 20

table, but as long as we have other values to offer the Big 3,the better off we are going to be.

The increased effectiveness of hypothesis-based decision mak-ing stems from the lesson that “the problem is not always the prob-lem.” That’s what Dominic Falkowski found out when he movedto Egon Zehnder’s Warsaw office:

My client was looking for a CFO because the current man-ager was not coping well enough with reporting and invest-ment analysis, and he was having problems with his team.We weren’t so sure this was the case. After an analysis of thesituation, including an assessment of the CFO, we realizedthat it was the CEO who was not structured, changed opin-ions and processes too often, and did not communicatechanges throughout the organization. The CFO was partlyto blame, however, as he had poor interpersonal skills anddid not cope well with any form of feedback.

We suggested some internal reengineering to be con-ducted by a strategic consultancy, and we ourselves coachedthe CFO and CEO. The result: solved problem, happy clientand CFO, and a prospering organization. Further, we provedthat an external search would not have brought the value theclient wanted.

IMPLEMENTATION GUIDANCEForming an initial hypothesis will make your problem solvingmore efficient and more effective, but to reap these benefits, youneed to be able to generate and test robust hypotheses. Since youshould form your hypothesis at the start of the problem-solvingprocess, you have to rely less on facts (you won’t have done most

Framing the Problem 21

01 (001-030B) chapter 01 8/27/2001 3:33 PM Page 21

of your fact gathering yet) and more on instinct or intuition. Takewhat you know about the problem at hand, combine it with yourgut feelings on the issue, and think about what the most likelyanswers are. This does not mean that the most likely answer is nec-essarily the correct one, but it’s a good starting point.

If something leaps out at you immediately, congratulations;you’ve just formed a hypothesis. At this point, whether you arealone at your desk, standing under the shower (alone or other-wise), or in a brainstorming session with your team, you should dothe Quick and Dirty Test (QDT) of your hypothesis. The QDT issimply this: what assumptions are you making that need to be truein order for your hypothesis to be true? If any of these assumptionsis false, then the hypothesis is false. Much of the time, you canknock out false hypotheses in just a few minutes with the QDT.This is especially useful when you need to choose from a fewoptions quickly, as Ciara Burnham, now a venture capitalist atEvercore Partners, can attest:

So much of my job involves triaging potential investmentopportunities to figure out which ones are worth spendingtime on. At the outset of any deal evaluation, I ask, “Whatdo I need to believe in order for this to be a good investment,and what are the ways in which this investment could blowup? Therefore, what analysis do I need to do to support/reject the investment and to dimension the risks?” Soundslike a simple approach, but frankly one that many peopletrained in the deal execution side of the business don’t take.

As an example, let’s go back to Acme Widgets. Yesterday, theboard told you and your team to figure out a way to lower themarginal cost of Acme’s venerable line of thrum-mats. Today, inthe first few minutes of your brainstorming session, you’ve comeup with a few options that might make the cut: (1) You might pres-

22 The McKinsey Mind

01 (001-030B) chapter 01 8/27/2001 3:33 PM Page 22

sure your suppliers to lower your raw-materials costs. (2) Youcould cut the workforce at your thrum-mat manufacturing facili-ties while maintaining production levels. (3) You might reduce thetime that the thrum-mats spend in the curing process, therebyincreasing throughput. Now you are going to put each option tothe QDT.

Pressuring your suppliers would be great, but can it be done?What needs to be true for that option to work? Well, you mightsay, raw materials should be a significant factor in total thrum-matcost; otherwise, reducing the cost of raw materials wouldn’t makemuch difference in the total marginal cost of a finished thrum-mat.As it happens, one of your team knows that raw thrums make upabout 35 percent of the total cost of a thrum-mat, so there shouldbe some mileage there. Next, you would need to have some pricingleverage with your suppliers. Unfortunately, on page A2 of thismorning’s Wall Street Journal, there is a story about the newlyannounced takeover of General Thrums by Allied Thrums andBezels. Analysts expect the merger to result in a significant reduc-tion in total thrum production capacity, with correspondingupward pressure on wholesale thrum prices. So much for that idea.

What about reducing manufacturing head count? Labor is alarge component of the total cost of thrum-mat production, so thatwould seem a fruitful area for exploration. The key question thenis whether Acme’s production facilities are overstaffed. One way todetermine this is by learning whether Acme’s per-worker produc-tivity is low relative to the industry. You recall seeing the resultsof a recent benchmarking study on thrum-mat production. Thatstudy put Acme significantly ahead of its competitors in per-worker output. Another dead end.

That leaves reducing the time that thrum-mats spend in thecuring process. Traditionally, Class A thrum-mats spend at leasttwo weeks in the curing locker—an expensive proposition that not

Framing the Problem 23

01 (001-030B) chapter 01 8/27/2001 3:33 PM Page 23

only uses a lot of energy, but also ties up inventory, thereby unnec-essarily swelling Acme’s balance sheet. A reduction in curing timecould thus have a double benefit. Not only would it boost Acme’sprofit line, it also would reduce work-in-progress inventory. Whatneeds to be true to make this feasible? As a first cut, the questionwould seem to be whether it is possible to make Grade A thrum-mats without a full two-week cure. Coincidentally, one of yourteam has just read an article in Thrum-mat Manufacturing Weekly

on a new curing process that utilizes special temperature, humidity,and atmospheric controls and that produces the same or betterresults as traditional curing methods.

Excellent! Your team has now found an initial hypothesis thatpassed the QDT.* Your next step will be to test your hypothesismore thoroughly and, if necessary, refine it. To accomplish thesegoals, you should now put together an issue tree.

An issue tree is the evolved cousin of the logic tree. Where alogic tree is simply a hierarchical grouping of elements, an issuetree is the series of questions or issues that must be addressed toprove or disprove a hypothesis. Issue trees bridge the gap betweenstructure and hypothesis. Every issue generated by a frameworkwill likely be reducible to subissues, and these in turn may breakdown further. By creating an issue tree, you lay out the issues andsubissues in a visual progression. This allows you to determinewhat questions to ask in order to form your hypothesis and servesas a road map for your analysis. It also allows you very rapidly toeliminate dead ends in your analysis, since the answer to any issueimmediately eliminates all the branches falsified by that answer.

24 The McKinsey Mind

*This example is provided for illustrative purposes only. It is not meant as a specific pre-scription for any company, partnership, sole proprietorship, or any other organizationinvolved in the manufacturing, curing, delivering, or servicing of thrum-mats. If you areactually in the thrum-mat business and are reading this, please do your own research.

01 (001-030B) chapter 01 8/27/2001 3:33 PM Page 24

Dan Veto found issue trees especially useful when defining a setof initiatives for the E-business arm of Conseco, one of America’slargest diversified financial services companies:

In problem solving, a lot of people try to be uniformly com-plete. The fact is, you don’t always have to be. You want tothink of everything as MECE, but you don’t need to investi-gate everything to the same depth. For example, we werethinking about our E-business strategy, as we had justformed a new business unit, eConseco. This would be astand-alone business with a real P&L, so we had to ask our-selves, “What are the key levers of profitability and growth?Which things matter, and which things don’t?”

We were inundated with ideas. Somebody on the rev-enue side would say, “Well, we could sell books.” Being ableto figure out quickly that that option is never going to makeyou any money is critical. Being able to snip from the tree thebranches that don’t matter so that we’re focused on thebranches that do matter is just an unbelievable problem-solving and problem-framing skill that is not necessarilyintuitive but which can really speed the problem-solvingprocess.

Going back to your team room at Acme Widgets, what wouldan issue map look like for reducing the thrum-mat curing process?As you and your team discuss it, several questions arise: Will itactually save money? Does it require special skills? Do we havethose skills in the organization? Will it reduce the quality of ourthrum-mats? Can we implement the change in the first place?

When laying out your issue tree, you need to come up with aMECE grouping of these issues and the others that arise. As a first

Framing the Problem 25

01 (001-030B) chapter 01 8/27/2001 3:33 PM Page 25

step, you need to figure out which are the top-line issues, the issuesthat have to be true for your hypothesis to be true. After a bit ofbrainstorming, you isolate three questions that address the validityof your hypothesis: Will shortening the curing process reduce ourcosts? Can we, as an organization, implement the necessarychanges? If we implement this change, can we maintain productquality? Put these issues one level below your hypothesis, as shownin Figure 1-2.

Unfortunately, the answer to each of these questions lies in sev-eral more questions. You will have to answer those in turn beforeyou can come up with a final yes or no. As you take each questiondown one or more levels, your analysis road map will begin to takeshape. Let’s drill down on one of these issues and see where itleads us.

26 The McKinsey Mind

Issue: Can weincrease thrum-matprofitability withthe new productionprocess?

Subissue: Can ourorganizationimplement thenecessary changes?

Subissue: Will itreduce our costs?

Subissue: Can wemaintain productquality whileimplementing the process?

Figure 1-2. Issue Tree for Acme Widgets

01 (001-030B) chapter 01 8/27/2001 3:33 PM Page 26

The issue “Can we implement the necessary changes?” throwsoff numerous subsidiary questions (see Figure 1-3). Some of themcame out in the initial brainstorming, while others will arise whenyou spend more time thinking specifically about the issue. Just asyou did for the main issue, you need to figure out the logical pro-gression of these questions. For the sake of this exercise, let’s saythere are two top-line questions for this issue: (1) Does the new,shorter process require special facilities that we don’t have? (2)Does it require special skills that we don’t have? For both of thesequestions, the ideal answer, “no,” shuts down any further inquiry.If, however, the answer to either of these questions is yes, thehypothesis is not immediately invalidated. Rather, this answerraises additional questions that must be answered. For example,in the case of facilities, you would ask, “Can we build or buy

Framing the Problem 27

Issue: Can weincrease thrum-matprofitability withthe new productionprocess?

Subissue: Can ourorganizationimplement thenecessary changes?

Does the processrequire specialfacilities?

Does the processrequire specialskills?

Subissue: Will itreduce our costs?

Subissue: Can wemaintain productquality whileimplementing the process?

Figure 1-3. Issue Tree for Acme Widgets with Subissues

01 (001-030B) chapter 01 8/27/2001 3:33 PM Page 27

them?” If the answers to these questions lower down the tree turnout to be no, then your hypothesis is indeed in jeopardy.

At this point, the process of laying out an issue tree should beclear to you. If you’ve done it, then you have sketched out the tasksthat you will need to fulfill in terms of research and analysis—sub-jects that we will address in the following chapters.

The McKinsey technique of hypothesis-driven problem solv-ing—solving the problem at the first meeting—has proved to bean excellent decision-making skill beyond the confines of the Firm.If you spend some up-front time combining your initial fact basewith your gut instinct, you will enable yourself to come to a morerobust solution sooner. A little bit of time spent weeding outinvalid hypotheses at the outset and then determining the scope ofyour analysis with an issue tree will save you time and improveyour results.

EXERCISES• Think about a nonbusiness issue about which you hold a

strong view (e.g., gun control, evolution, global warming).List the assumptions that you are making with regard toyour position. Are they all true? What information oranalyses would you need to support your view?

• If you haven’t already, come up with a couple of likelyhypotheses for whatever issue you are currently workingon in your job. Can you come up with one or two thingsthat must be true for each hypothesis to be valid? Nowsubject each hypothesis to the QDT.

28 The McKinsey Mind

01 (001-030B) chapter 01 8/27/2001 3:33 PM Page 28

CONCLUSION

By using structured frameworks to create an initial hypothesis, youwill enable yourself and your team to select the analyses and areasof research that will allow you to reach a robust conclusion in theshortest possible time. In the next chapter, we will look at how toplan an analysis to prove or disprove your hypothesis in the short-est time possible.

Framing the Problem 29

01 (001-030B) chapter 01 8/27/2001 3:33 PM Page 29

01 (001-030B) chapter 01 8/27/2001 3:33 PM Page 30