should you build strategy like you build

8/14/2019 Should You Build Strategy Like You Build

1/7

STRATEGY

Should You Build StrategyLike You Build Software?n many companies, the corporate planning department has gone the wayof the dot-matrix printer. But if the analyst-driven, top-down, formalprocess of the previous era is dead, what have companies replaced it with?

Not much, according to recent surveys of executives and managers of globalcompanies. Many companies still hew to the decades-old annual strategicplanning process, only now the responsibility for creating plans falls on groupmanagers and department heads, with little central support. Other companieshave jettisoned formal strategic planning altogether - centralizing strategydiscussions to a few top executives or replacing planning with a hodgepodgeof informal or semiformal retreats, executive leadership councils and boardcommittees. Some observers have gone so far as to argue that strategy makingis simply too uncertain and complex to be handled by a defined process. Un-fortunately, senior managers may be throwing the proverbial baby out withthe bath water. The fact that the process is failing doesn't mean that strategymaking is resistant to a process approach.

Consider software development. Around the same time that managerswere losing confidence in strategic planning, software development wentthrough its own crisis, as the demand for faster design and integration of in-creasingly robust systems began to make the traditional "waterfall" approachto software development obsolete.' (See "The 'Waterfall' for Software Devel-opment/Strategic Planning," p. 70.) The crisis in software prompted a fewvision-aries to rethink how software gets built. They didn't abandon a processapproach to the problem; rather, they invented new development processes,such as rapid application development, extreme programming and agile soft-ware development, to confront the new realities.

The pressure to accelerate and at the same time cope with increased uncer-tainty and complexity led to the decline of traditional software developmentin some settings - and similar pressures seem to have played a pivotal role inthe demise of formal strategic planning. What's interesting, though, is howdifferently people in these distinct worlds responded. Software developerswent to work rethinking how software should be created. But strategy makingdidn't really change.

The insights upon which new software development approaches are basedmay point the wa y for the development of newer, faster and more effectivestrategy-making processes. Over the past six years, my colleagues and I have

Keith R. McFarland, former technology CEO, is founder of McFarlandStrategyPart-ners n Sandy,Utah,andauthorof The Breakthrough Company: How EverydayCompanies Become Extraordinary Performers. Comment on this articleor contact theauthoratsmrfeedback @mit. edu.

The strategic planningmodel is due for a"new release," onethat enables companiesto keep pace withchanging environments,quickly create and adaptstrategy and empowerpeople throughoutthe organization tomake effective choices.

Keith R. McFarland

SPRING 2008 MIT SLOAN MANAGEMENT REVIEW 69


2/7

STRATEGY

experimented with some of these principles to streamline thestrategy-making processes at around 60 companies, in industriesincluding hospitality, heavy manufacturing, financial services,education, healthcare and distribution. What we've learned haschanged our view of strategy and how it ca n be created.

Many software development and strategic planning processesfall on a continuum from predictive to adaptive. Traditional ap -proaches tend to be highly predictive in nature, in that theyattempt to plan for the future in detail. With predictive ap-proaches, planning is optimized for the original targets; it isgenerally difficult to change direction once implementation isunderway without throwing out the work that has been done andstarting over. (See "Shortcomings of Traditional Software Devel-opment and Strategic Planning Models.")

Management guru Peter Drucker pointed out the problemswith predictive approaches years ago: "Most discussions of theknowledge worker's task start with the advice to plan one's work,"he noted. "This sounds eminently plausible. The only thingwrong with it is that it rarely works. The plans always remain onpaper, always remain good intentions. They seldom turn intoachievement."2

Adaptive approaches, on the other hand, focus on adjustingquickly to changing realities. In software development, theyprovide teams with the flexibility to revise software programs asdesigners learn how people actually use them or according tochanging environmental forces. They do more than just tweaktraditional models - they allow for a fundamental rethinkingof the process from the ground up. A similar change for strategyis long overdue: Among 72 companies we surveyed with reve-nues exceeding $250 million, 80% still use a traditional strategicplanning model, despite its recognized shortcomings.3

"The I* fo r ISfD PILike the conventional "waterfall" techniques for software developmstrategic planning model makes it difficult to jump back to a previounew learning.

Software Concept SOFTWARE Corporate VisionDEVELOPMENTRequirements Analysis WATERFALL Strategic Plans

Architectural Design Forecasts and Budgets (CorpiDetailed Design Strategic Busines

Coding and Debugging l Forecasts aSystem Testing4O

Adapted from S.McConnelI, "Rapid Development" (Redmond. Washington: Microsoft Press, 199

An Adaptive Approach to Strategy DevelopmentIn 2001, a group of software industry veterans convened in Utah toexamine the commonalities of the new approaches that werespringing up to address the challenges in the software industry. Thgroup produced what became known as the Manifesto for AgilSoftware Development,4 which highlighted, among other thingsthe importance of emphasizing individuals and interactions oveprocesses and tools, of working software over comprehensive documentation and of responding to change over following the planThese themes are equally applicable to strategy.

Strategy is a mechanism through which a company makesense of the world around it. It is a collection of ideas about howthe company intends to win, the source code upon which every-thing else depends. Because strategy ca n only capture a company'sbest thinking at a given point in time, strategy (like a softwareprogram) needs to be refined and improved as people gain anddistribute new experience and knowledge.

Given this reality, sound strategy development processes shouldenable a company to create and adapt strategy quickly and iterativelyso that people can triage issues effectively and allocate resources inchanging environments. Strategy development should be optimizedfor two important outcomes: surfacing the best ideas for improvingthe company's current and future market positions; and ensuringthat individuals throughout the organization have access to the latesversion, so that what people do every day is aligned with the mosimportant strategic insights. The "customer" in this process is theentire organization-- everyone who is involved with making strate-gic choices and allocating resources. Since people at many levels oan organization make daily tradeoffs that impact the company'strategic success, the process needs to be designed to tap into ideafrom all corners of the organization - more than just the top ex-

ecutives. As my colleagues and I havhelped companies work on strategicissues, three major themes havemerged: the importance of having

ent, the traditional an iterative, or "spiral,' approach aus stage an d integrate opposed to a linear approach; th

importance of organizing the stratSTRATEGY egy-making process around peopl

0 PLANNING rather than processes; and the recogWATERFALL nition that in strategy there is nosuch thing as a "silver bullet?'Adopting aSpiral Rather Than aLineaUnit Plans Planning Model New software devel

nd Budgets (SBU1 opment techniques recognize thaActio Pla7N many software programs can be deAction Plans t veloped more quickly and

successfully when they are devel5).139. oped iteratively. Software developer

70 MIT SLOAN MANAGEMENT REVIEW SPRING 2008 SLOANREVIEW.MIT.EDU


3/7

I Shoto ig of Trdtoa Sotwr DevelopmenIt an Staei Plnnn Model I

The need to more rapidly integrate new learning and adapt to changing environments demonstrates the limitations of conventionalsoftware development and strategic planning processes.

SpeedProblems in software developmentSymptomSystem delivered isalready significantly obsolete at systemrelease date.Process root cause"Cascading" sequential process (requirements definition,design, implementation, integration, testing, deploy-ment) is too slow.SymptomSystems fail to deliver on their potential, often due to a dis-connect between those building the system and those whowill use it.Process root causeCumbersome process and extensive documentation rigidlyenforced by information technology heads despite systemusers being increasingly under the gun to deploy bettersystems faster. Users blame system failure on bad develop-ers. Developers blame system failure on poorly defined or"moving target" requirements.Symptom"Kitchen-sink" approach to software development, in whichusers try to spec and developers try to build the perfectsystem with every feature users can imagine ever wanting.Process root causeMethodology fails to recognize and incorporate how theworld really works. Users won't know up front exactly whatkinds of functionality they will need, nor exactly how theywill need it to work.Speed and substance problems above create the "de-velopment doom loop." Users know that if something isnot in the spec, they are not going to get it. As specs bal-loon, development cycles lengthen, the wait in the cue fordevelopment resources grows long as well - prompting auser to throw everything he can imagine into the specdocument when his number finally comes up.

realize they can't have all the knowledge they need to produce theperfect system up front, so they get working models into the handsof users as early as possible. By interacting with users in a "spiral"process, developers can integrate their feedback in future versions. 5(See "The Spiral Model of Software Development." p. 72.)

A spiral approach to strategy making recognizes that strategycan be created most effectively when formulation and implemen-tation go hand in hand. It recognizes that if people only focus oncea year on what drives the business, strategic ideas may evolve tooslowly and fail to keep pace with the changing realities of the mar-ketplace. As Jack Welch, former CEO of General Electric Co., wrotein 2000, "We've long believed that when the rate of change insidean institution becomes slower than the rate of change outside, theend is in sight."6 The spiral approach rejects the traditional, hard-

line separation between strategy and implementation, just asmodern approaches to software development reject the separationbetween design and coding or implementation.7 Strategy formula-tors and implementers need to develop mechanisms to get togetherin person and wrestle with the issues. Otherwise, strategy won'tkeep pace. In fact, the most valuable learning takes place when thepeople on the front lines attempt to execute the strategy.

Shamrock Foods Co., a leading food distributor based in Phoe-nix, Arizona, has used a spiral planning model to acceleratestrategic change since 2004. From its beginning in 1922 as a smalldairy, Shamrock has grown to become one of the ten largest foodservice distribution companies in the United States. The company'sstrategic planning activities evolved from a process introduced bythe Boston Consulting Group in the early 1980s: Senior executives


Problems in strategic planningSymptomRealities in the marketplace call for a change in strategybefore the ink dries on the strategic plan.Process root cause"Cascading" sequential process (environmental analysis,corporate goals, corporate strategies, business unit goals,business unit strategies, budgets, action plans) is too slow.SymptomStrategies fail to deliver on their potential, often due to adisconnect between those creating the strategy andthose tasked with executing it.Process root causeLimiting strategy discussions to top managers reducesleaders' access to "rich" and "soft" data, and that failsto give implementers (who have that rich data) the op-portunity to rapidly adapt strategy. Implementersblame failure on bad strategy. Strategists blame failureon bad execution.Symptom"Kitchen-sink" approach to strategy development inwhich each year the company begins an analytical searchforthe "correct and complete" strategy, which is ex-pected by senior management to be reflected evenly andmeaningfully across the organization.Process root causeMethodology fails to recognize and incorporate ho w theworld really works. Strategies built on an annual basisand treated as "complete" reduce the organization'sability to make changes on the fly. There is little opportu-nity to collect, codify and expand on incrementalopportunities to shape the strategy as learning occurs inthe organization.

Substance

Scope

SLOANREVIEW.MIT.EDU


4/7

STRATEGY

Th Spia Moe of Sofwr Development

The principles of a spiral approach to software developmentcan be effectively applied to strategy making, where strategiesare built iteratively over time as strategic assertions are testedand refined in implementation "scrums."

Adapted from B. Boehm, "ASpiral Model of Software Development and Enhance-ment," ACM SIGSOFT Engineering Notes 11, no. 4 (1986):25.

spent several months each year updating a five-year plan. Bu t in2004, CEO Kent McClelland began to have misgivings about therigidity of this approach. Shamrock's regional organizational struc-ture made it difficult to adapt to the quickly changing needs of itscustomers. He worried that Shamrock's strategy was being opti-mized at the regional level - with too little attention paid to thedevelopment of critical capabilities that transcended divisionalboundaries. He also noticed that managers always juggled toomany priorities, making it difficult to focus on the few strategicleverage points that were most important for growth.

In February 2004, we helped McClelland organize a three-day,48-hour planning session for 45 Shamrock managers and employ-ees. With participants from all divisions, functional areas and thefront lines, the group represented a cross-section of the company.By the end of the second day, the group had zeroed in on five stra-tegic themes: the need to revamp the way Shamrock serviced largecustomers like Burger King and Taco Bell; the need to improveorder accuracy and service reliability; the need to increase revenueon Shamrock's own house brands; the need to reduce logistics coststhrough selected regional acquisitions; and the need to improve thecompany's ability to recruit, retain and develop good talent.

Selecting themes was just the beginning. Shamrock also needed

to update its strategic thinking in each of the five strategic areas.While organizations can learn a lot from analyzing and discussingstrategic issues, the most important learning occurs when peoplejump back and forth quickly between strategy formulation and im-plementation - testing their strategic assumptions with what theyare learning. Indeed, even ideas that seem to be right on target needto be shaped and reshaped to reflect the realities of the marketplace.Fo r this reason, at the end of each planning session, participants cre-ated and prioritized a handful of specific and measurable strategicinitiatives that would advance each strategic theme. Then they builtdetailed action plans and set measurable outcomes they thoughtcould be achieved within the next 90 days.

Shamrock's process revolved around quarterly strategy"scrums": Team members met at an offsite location for a day toevaluate the company's performance against the action plansfrom the previous quarter. We asked them to identify the mostimportant things they had learned about the company's strategysince the previous meeting and to suggest how those insightsshould be integrated in the strategy going forward. The groupcreated new action plans for the upcoming period. In addition tothe quarterly scrums, the participants met every year for threedays, during which time people were asked to step further backand revisit the company's strategic assumptions.

Since adopting the new approach to strategy, Shamrock hasflourished. It has consolidated the part of the business that serveslarge national chains in one of its warehouses, which has led tonew insights into how best to serve the needs of those customers.Shamrock's strategic initiative to drive world-class customer ser-vice has led it to segment its customer base and to rationalize itslogistics systems in a wa y that will not only increase customersatisfaction but also eliminate unnecessary costs. It hasn't been acompletely smooth ride - the company ran into difficulty inte-grating one of its recent acquisitions. But this generated someimportant lessons and forced management to reexamine its ac-quisition and integration strategies.

Organizing the Strategy Process Around People, Not Tools An effec-tive strategy needs to magnify the efforts of people throughoutthe organization. It's much easier to do this if lots of people un-derstand the strategy and are able to apply it to the variousdecisions they face each day. The best way to ge t people in themiddle and on the front lines of an organization to understandand embrace the strategy is to involve them in creating it.

In 2003, my colleagues and I were hired by Metrowerks, then aleading software development tools company owned by MotorolaInc., to develop a new strategy. Metrowerks (best known for itsCodeWarrior programming product line) was plagued with prob-lems - products were behind schedule, there were intensedisagreements over where resources should be directed, employeemorale was low - and there was chronic red ink. Matt Harris, the

72 MIT SLOAN MANAGEMENT REVIEW SPRING 2008 SLOANREVIEW.MIT.EDU


5/7

new CEO, was feeling the pressure. "Days are months in Metrowerkstime," he explained. "We simply don't have time to spend monthsfiguring out where we have to go." Harris wanted to develop a ne wstrategy that identified the markets, customers and products thecompany should focus on. He also wanted to get people throughoutthe organization to move beyond their differences and work together.Since Harris wanted a plan in 45 days, there wa s no time to waste.

Get more people involved. Motorola, the parent company, hada highly refined, top-down strategic planning process, which wa sconducted by senior executives and some staff analysts. We turnedthat model on its head. Without time for study groups and taskforces, we needed to gather the information quickly and organize itfor immediate action. In order to capture the hearts and minds ofdifferent factions, a cross-section of the organization had to ge t in-volved in developing the strategy.With Harris' help, we identified 60people - senior executives, middle managers, sales and marketingpeople, research engineers and top programmers - to take part ina three-day session. We also included a handful of people fromMotorola to facilitate strategic coordination with the parent.

Each person was invited to participate actively in setting the direc-tion of the company.As a first step, each person identified the most

important issues facing the company and what he or she thought thecompany should do . There wa s virtually no consensus, and conflictand mistrust were widespread. But much of the conflict could betraced to a lack of agreement on strategic direction. Almost everyonehad a strong commitment to turning the business around.

Work the issues face-to-face. The new software developmenttechniques stress the importance of face-to-face collaboration withusers as a way to mitigate conflicts between developers and usersand to get beyond the organizational silos that often interfere withmultidisciplinary problem solving. The same principle applies tostrategy: In a strategy process that values individuals and interac-tions over processes and tools, there is no substitute for face-to-facedialogue. During the three-day meeting, we discussed the big issuesfacing the company, including the future of Metrowerks' strategicrelationships with other parts of Motorola, their markets' custom-ers, product sets, competitive threats and internal and externalchallenges. To avoid getting bogged down by the size of the group,we divided it into groups of ten; each subgroup had people fromdifferent areas, and participants moved in and out of groups on aregular basis. By design, the process wa s iterative, moving fromdetailed observation to idea to strategic assertion, then back to

I he Fuur of Statg Main

The key principles that underlie new approaches to software development point the way to further innovation in strategy making.New approaches in software developmentReplace "big design up front" with a model that breaksthe development cycle down into shorter iterations thatallow users and changes in the environment to influencelater iterations.Rather than trying to build all possible needed func-tionality into the product at release, build quickprototypes and focus first on getting important corefunctionality in the hands of users; add additional functionality in later iterations.

Short-circuit developer-user adversarial relationshipsby making users and developers jointly responsible forthe success of the project.

Whether through implementing a true "open source"approach or by generally emphasizing transparencyand openness, encourage everyone in the communityto help make the program as good as it can be.

Bugs are identified and corrected continually througha process called "code and smoke" instead of waitinguntil the end of the project. People who find bugs arecelebrated and appreciated.

Implications for strategy makingReplace "big design up front" - highly structured andanalytical annual planning processes - with more fre-quent and compressed "strategy scrums" that enablethe iterative development of strategy.Toss the three-ring binders filled with hundreds ofpages of strategic planning data. Develop a funda-mental set of strategic assumptions and get them intothe hands of the users (managers) to enhance itera-tivelyin web communities and face-to-face strategy"reset" meetings.Broaden the group responsible for strategyto includeboth "thinkers" and "doers." Use streamlined strategicprocesses to involve people from several levels in the or-ganization and from all the major functional areas.

Give people throughout the organization constant andup-to-date access to the "strategy source code" - themost recent set of assertions and assumptions about howthe organization intends to win - and encourage themto make improvements.View strategy as simply a collection of ideas that islikely to be shaped, refined and corrected as the orga-nization learns. Don't let people who question thestrategy be labeled as "not team players" - celebratepeople who ask great questions that cause the com-pany to question and refine its assumptions.


Start smalland iteratefrom thereGet somethinginto people'shands quickly

Get people towork together

Integratethe best ideasavailable

Debug asyou go

SLOAN REVIE1W.M IT.EDU


6/7

STRATEGY

observation. It was also disciplined, with a set of clear ground rulesto keep people engaged and on topic.

Get senior executives off center stage. Someone other than asenior company executive should facilitate important strategicdiscussions. There are several reasons for this. First, even when theydon't realize it, people naturally defer to those in authority. Also,people are less eager to promote ideas that haven't been endorsedby others (sociologists call this "social proof"). Vhen senior execu-tives attempt to facilitate the discussions, they are often identifiedwith positions - even when they try hard to be even-handed. Agood outside facilitator can use conflict to distill issues and buildconsensus; participants don't have to worry that they have sup-ported the "wrong" side of an issue.

Trust the process. We expected that including the people respon-sible for implementing the strategy (the strategy "customers") in theprocess would make them more effective in their jobs. Among otherthings, we thought that an understanding of some of the nuances ofthe strategy would improve their decision making. Bu t would in -volving these people in strategy making actually result in betterstrategy? Initially, the senior people at Metrowerks were skeptical.Bu t over time, as they learned how to integrate people into the pro-cess, they saw that opening up the strategy formulation process toinclude people from several levels of the company significantly im-proved the quality of strategic thinking in the company. They shouldnot have been surprised. As computer programmer Eric S. Raymondwrote in his now-famous essay, "The Cathedral and the Bazaar"about the open-source community: "I believed that most importantsoftware ... needed to be built like cathedrals, carefully crafted byindividual wizards or small bands of mages working in splendidisolation.... Linus Torvald's style of development - release earlyand often, delegate everything you can, be open to the point of pro-miscuity-- came as a surprise. No quiet, reverent cathedral -buildinghere - rather, the Linux community seemed to resemble a greatbabbling bazaar of differing agendas and approaches ... out ofwhich a coherent and stable system could seemingly emerge only bya succession of miracles. The fact that this bazaar style seemed towork, and work well, came as a distinct shock."'8

Senior executives at most companies have ye t to consider thistype of approach. A 2006 survey of 796 executives by consultingcompany McKinsey & Co . found that in 73% of companies, strate-gic decisions are made by the CEO, business unit leaders or a smallgroup of senior executives.9 But executives who have been througha more participative and adaptive process immediately see themerits of that process. Matt Harris of Metrowerks admits he wa sskeptical when the new process got underway. Among other things,he worried about the size of the group. He says: "I had already de-cided beforehand that if the process wasn't working by the end of thefirst day, I'd shut it down - it would be a great way to signal to theteam that I would demand performance in everything we did -even from consultants that I myself hired. Not only was this

unnecessary - the process itself wa s an unmitigated success. Peo-ple who participated still call and write me years later, telling mehow invigorating the whole experience was, and asking how toreplicate it. No one wa s more surprised than me."Forgetting About Finding the "Silver Bullet" In a 1986 essay entitled"No Silver Bullet," software industry visionary Frederick Brookswrote, "However frustrated we may be in the writing of computerprograms, we will never find a magic, transformational break-through--we should expect only modest, incremental advances."' 0Had someone said something similar about the field of strategyearly in its development, strategic planning might never have de-clined so precipitously. Today, few people still believe that strategyis merely a matter of understanding industry structure or writinga long-term plan. Most managers operate in settings that are toodynamic and complex for simple success recipes. Instead of seekinglong-term sustainable advantage, good managers need to createsustainingadvantages on an ongoing basis.

Like software developers, strategists must learn to add new stra-tegic features (capabilities, insights and tools) incrementally. Takentogether, these features can add up to an advantage, but managersneed to be realistic and recognize that any competitive edge mayonly last for a short time. An ongoing challenge will be to continueto develop and refine new approaches to strategy making that helpcompanies bridge the gap between strategy formulation and im-plementation. (See "The Future of Strategy Making," p. 73.)

REFERENCES

1.S. McConnell, "Rapid Development" (Redmond, Washington: MicrosoftPress, 1996), 139.2. P Drucker, "The Essential Drucker" (New York: Harper Business,2001), 225.3. Unpublished survey, McFarland Strategy Partners, 2006.4. Manifesto for Agile Software Development, 2001, http://agilemanifesto.org.5. B.Boehm, "ASpiral Model of Software Development and Enhance-ment," ACM SIGSOFT Engineering Notes 11, no. 4 (1986): 21-42.6. J. Welch, J. Immelt, D.D. Dammerman and R.C. Wright, "Letter toShare Owners," in"GE Annual Report: 2000" (Fairfield, Connecticut:General Electric, 2001), 1-7.7. Mitchell Kapor, founder of Lotus Development Corp., said, "If esign andimplementation are inwatertight compartments, itcan be a recipe for disas-ter because the natural process of refinement and change isprevented."Se e S. Rosenberg, "Dreaming inCode" (NewYork: Crown, 2007), 150.8. Ibid., 23.9. "Improving Strategic Planning: AMcKinsey Survey," McKinsey Quar-terly (September 2006).10. FR Brooks, "No Silver Bullet: Essence and Accidents of Software En-gineering," Computer 20, no. 4 (April 1987): 10-19.

Reprint 49315. For orderinginfortnation,seepage 1.CopyrightMassachusettsInstitute of Technology, 2008. All rights reserved.

74 MIT SLOAN MANAGEMENT REVIEW SPRING 2008 SLOAN REVI EW.M IT. DU


7/7

COPYRIGHT INFORMATION

TITLE: Should You Build Strategy Like You Build Software?

SOURCE: MIT Sloan Manage Rev 49 no3 Spr 2008

The magazine publisher is the copyright holder of this article and it

is reproduced with permission. Further reproduction of this article in

violation of the copyright is prohibited. To contact the publisher:

http://mitsloan.mit.edu/smr/

should you build strategy like you build

Documents