cases

M2-28 ANALYTICS . TEAMS . ORGANIZATIONS . SKILLS

Strategic Design at Dynacorp

oWetre too slotv, t00 unresplnsipe to the rnarhet, and too undisciplined abowt costs. And the wain reason is how we're organized.4t jwst isn't working anl rnore. Wetye lutgrlwn the old. d.esigru in s0 ?vta.n! nrays, but wetre still n'ying to wanage n,ith the satne stracture we hnd nrhen nre stnrted the cowpany."

-Dvnacorp prodwct wa.nager

"There's a lot of tnlk nont abowt cbanging owr otganization d.esign. Bwt nre showld be carefwl thnt we're not throwing the baby out tpith the bnth- 'tttr.ter. We)re Jixating on n,bat's wrlng nrith the orylanizntion instead of thinhing about hont to mnhe it nrork better.t'

- Dyn a c orp eng in e ering rm a.n &g e r

The Dyr-ra Corporation, known in the industry as l)ynacorp, is a n-rajor global infbrn-ration systcms aud communications company. Originating in an office equipment corrpau.ly that uroved into high- tcchnology applications in the 1960s ancl 1970s, Dynacorp l-rad, by the 1980s, establisl-red ;r positior-r as an industry leadcr, known fclr its technological innovatior.r. Dynacorp was first to the rnarket with innovative and high-quality products that were sig- nificant zrdvanccs on anythirlg its con.rpetitors were offcring. (lustomers would gladly wait lnonths, arncl even a year or so, to take delivery of prr<>ducts bear- ing thc l)ynacorp logo. The cLrstomcrs were typi- cally sophisticirtcd users who were willing to do somc of thcir or,r,n applications work and to figure oLlt how to intcgratc Dynacorp's neur products with the rcst of their opcrations. During this pcriod, the comprrny grcw at a very fhst ratc, and expandcd its markct to Europe , Asia, ar-rd Latir-r An.rerictr.

Thc 1990s wcre a rr.ruch more diflicult ocriod fbr thc company. It contintred to gro\\', bit at a slower rate, ancl expericnced periods of significantly reduced carnings. Critics both insidc ancl outside thc company attributed Dynacorp's clilficulties to a loss of leadership in getting nerv products to ntar- ket, costs that were too high, and changes in the markctplacc tl.rat Dynacorp was slow to recognize. Compctitors werc closing the technologl'gap, and were often fhster gettir-rg products to market. In a growing numbcr of product areas, Dynacorp hild been surprisecl b1, competitors who, ;rlthor.rgh the1, l"rad started working ol'l a ncw product much later than Dynacorp, were fhster at getting the product to the custonrcr at an attractive price. In addition, a growing number of information tecl-rnology and

communications (ITC) cor-rsulting firms were cap- tr,rring the relationship with the large customer by offbring "value-added services and solutions." The consulting firms acted as interr-nediaries, supervis- ing the purchase of ITC hardware and software, provicling intcgration services, and capnrring n"rucl-r ol' tlrc high-nrargirr busirress.

A growing number of cxecutives in Dynacorp were coming t<> belicve that the problems could not be addressed effectively with Dynn6erp's current organizational structurc. Like most cornpanies, Dynacorp had been established with a functional organization. As the courpany expandcd its activities rcross five contillents and greatly increased its prod- uct rallge, top managernent began to ask whcther Dynacorp needed a rnajor redesign.

As a first stcp, the CE,O appointed a small internal tt'rsk force to mtrke preliminary recomr.nendations on clrganization design to tl-re top management team. The CEO personally laur-rched the task force with a comparry-widc communication listir-rg dre challenges Dyr.racorp was fhcing: l-righ costs, being too slow tc'r gct new products to thc r.narkct, and a necd tcr incretrse the value creatcc'l firr customers. Thc CEO emphasized that the current task fbrce's mandatc was driven by l)ynt'rco4r's cor-nmitment to maintain- ir-rg its leac'lership ;-losition in its industry, ar-rd asked all Dynacorp cmployecs to cooperate fully and fimkly rvitl-r thc task firrce's inquiries.

A nurnber of the task fbrcc members traveled around tl're companl, conclucting ir.rterviews with mal'ragers at n-rulti;lle lcvels across the three nrajor divisi<>ns of the cornpany, while others embarkcd on r'ur "cxternal bcnchmarking" exercise to look at how other c()mpanies in the indr-rstry were organized. As they consolidated their findings, they fbund thernseh'es in agreement tl'rat Dynacorp's current organization had serious shortcomings, but they disagreed vehemently on r.vhat the best design solu- tion n-right bc. The CEO hacl made it clear, lrow- ever, that he did not want a simplc listing of trlternativcs; he wi'rntecl thc task force, at the very least, to provicle tr ranking of the altemative choiccs.

Dynacorp's Design: The Functional Organization

Since its earliest days of operation, D),nacorp had been orsanized fi.rnctior-rallv. The three line divi-

Source: From Competing by Design: The Power of Organizational Architecture by David A. Nadler, Michael L. Tushman. and Mark B. Nadler, copyright @ 1997 Oxford University Press. Used by permission of Oxford University Press, Inc.

Module 2 . Three Lenses on Or.qanizational Analysis and Action M2-29

sions, each led by an executive vice president) \r.ere engineerir-rg, manufhcturing, and rnarketing. These \'vere supported by a number of corporate staff divi- sions, including finance, human resource rnanage- ment, and corporilte aflairs.

The engineering division u'as organized b1, 1s.h- nical area into l0 groups. The advanced technol- ogy grorlp worked on developing the ltext generation of technologl., often in close collabora- tions u,ith unir.ersity labs and independent research institlltes. In addition, eight groups n.ere based on technical specialty (devices, storagc systents, soft- ware , etc.). Finally, a process tecl-urology group had been set up relatively recentl,v to u'ork on the devel- opment of advanced computer-aided manufactur- ir-rg systems.

Originirlll', the personnel of the engineering division \\,cre cllrstcrcd in one location. Over time, hon'ever, both constririr-rts on space and the recog- nition of the need to har.e a presence in ke1' lsch- nolog.v "hot spots" lreant th:rt the enginecring division eventually spread across sites ir-r three regiolls of the Uniteci States, including the two coasts. Rccentlt, discussions u,erc held about set- ting up cngineering centers in Europe ancl Asia, rvhere srrles rvere grorvir-rg rapidly, bnt no decision had been rnade, irlthough the n-rarkcting organizir- tions in those regions strongly advocated the intcr- nationirlization of engilreerirrg.

Dynacorp was onc of the fbrv American compa- nies that had successfully irnplernented a dual career ladder in its engineering division, whereby out- standing engineers and scientists could continlle to advance ir-r salar,v and status lvithout leaving the "bench" or the lab. Instead of moving into cngi- neering management positions, they could rise on a ladder of positions that paralleled the management career steps. Manl' 6s111panies tried to implernent such a svstem, but as a rule '.he technical ladder does not begin to match the prestige and pay of the man- agelnent ladder. In Dynacorp, however, a significar-rt number of outstanding technict'rl people rose to top positions on the technical ladder. Most engineers, hou'cver, advanced into nranagement positions in the engineering division. Over tin-re the number of rLrngs on both the technical and tl-re management ladders had increased, allou,ing fbr advancement opportunities u,ithin the division.

The rnanufacnrring division r,vas organized by location; the basic Lutit was the factory, which was a cost center, measured on overall costs, productivity (rvhich was strongl,v shaped b), capacig, utilization), and qualiq'. Some ofthe factories had begun b1, man- ufacnrring one palticular prodr-rct line, but over tinre, as some products were phased ollt and new product Iines introdr"rced, most fhctories had come to produce a range of products. This production varietv made it somer'vhat difficult fbr fhctory mallagers to calculate

clear productior-r costs for any one product line, because ofthe shared factorl, overheads.

Dynacolp had a higher percentage of its prod- ucts manufactured ir-r its own factories than most of its U.S. competitors. The company had been an early American adopter of total quality control and was a lead user of new process technology The career ladder in manufacturing had originally been quite short (Dyr-racorp had prided itself on its flat orgirnization), but as the size of the plants grerv, the number ofmanagement grades increased over time. The most coveted position in the function was that of general manager of a manufactr.rring plant.

Like engineering, manufacturing had begun in one location, but it expanded earlier into diffbrent locations within the Ur-rited States, in search of lower costs and proximity to growing rnarkets. The com- pany had plants in 12 states ir-r the United States and tr,vo countries in Europe. Production had r-rot yet becn set up in the Asia-Pacific region, but the com- pany \vas increasingly drawing on sr,rppliers located in thtlt region, and some managers advocated setting upl mirnufacturing operations there in order to take advantage of its local manuficturirlg expertise and lou,er costs. Some of the markcting people u4ro had been pushing hard fbr this option blamed the com- pany's slowness to move on the rellrctance of manu- fhcturing to lose domestic production. They also criticized manr.rfhctr.rring for being unwilling to out- source more of its production, because this rvould affcct the promotion possibilities rvithin the ftinc- tior-r. Manufacturing execlrtives denied this accustr- tion vehemently, and asserted that their ovcrriding collcenl was to etlsur€ the continuation of Dyna- corp's replltation fbr production qualir,v.

The marketirrg division was also organized by geography, but the geographic divisions diffbred f-r'om those of the manufacturing division. Six regions in the Ur-rited States r,vere in turn divided into sales territories or branches. Marketing and sales outside the United States were handled b), an international group that rvas organized by country. Marketing people cultivated close ties witl-r their custofflers, so close that junior recruits often com- plained that their managers "owned" custorners and were reluctant to involve them deeply in serving existing clients. The marketing division prided itself on the continuiq' of its cttstomer relationships and on the depth of its people's knorvledge of and fierce loyalty to Dynacorp products.

The marketing division's tasks had changed con- siderably over time. In its early years and well into the company's high-growth era, Dynacorp had tar- geted the sale ofits products to technical specialists in customer companies. These customers recog- nized the technical superiority of Dynacorp's prod- ucts and rvere happy to rvork r,vith its technical support people to rvork out any problems. How-

M2-3O ANATYTICS . TEAMS . ORGANIZATION$ . SKILLS

ever, more and more companies were taking system purchasing out of the hands of their technical enthusiasts, and Dynacorp's marketing people increasingly found themselves selling to business unit purchasing groups-often backed up by, or even represented by, consulting frrms with special practices in ITC. This shift meant a high demand for technical support, lo\ter clrstomer tolerance of earll, 51xt. problems with new product lines, dre need to integrate Dynacorp products with other systems to provide solutior-rs, and more polished marketing pitches. It also meant that it became more difficult to cultivate the close oersonal rela- tionships with buyers that had been a long-stand- ing element of Dynacorp's marketing strateglr.

Dynacorp's top managemeltt was justifiably proud of what their functional organization had accomplished. Its engineering division had an excellent tcchnical reputatiolt and was able to attract top technical talent and to clevelop the close working rclationships with leading scientists and engineers et major rescarch universitics that enabled it to stay on the cutting eclge of nerv tech- nology. Its manufhcturing division had built strong capabilities in quality manufhcturing at a time when U.S. industry overall was widely criti- cized fbr fhlling behind in production systems. It cor"rld also boast of a dedicated and knowledgeable marketing ftinction whose people haci br"rilt close working relationships with customers who were fiercely loyal to Dynacorp products.

Top management recogrlized that the competi- tive environment was changing, and that Dyna- corp's high costs, slou/ pace of getting ltew products to market, and the need to generate value-added services fbr custorners made the com- panv vulnerable. )y'nxgorp was not fhcing arl in-rmediate crisis-it was still profitable, grorving, and highly respected in its indr-rstry. But the indus- try was changing, and Dynacorp's top manage- ment believed that they had a windou, of a year or two to address the problems befbre the current trends led to reallv serious oroblems.

Internal Investigation: Diagnosing the Organizational Problems

The task force found widespread agreement on the importance of the problems that Dynacorp was facing, bllt considerable variation of opinion on what was causing them or on how to fix them. The engineering division's marlagers r,r.ere convinced that the primary responsibility fbr high costs and delays in getting prodLlcts to market lay with man- ufacturing. One senior engineering manager said that even though his people tried to hand the manufacturing site a detailed and complete set of

specifications, the manufhcturing manager kept corning back vvith questions or objections; he wanted to substitute stalldard parts that would mean reconfiguring the design or lowering the product lunctionaliry or to change elements of the design to make it easier to produce, or to clarify things simply because his people didn't understand something about the design. Each iteration took time, because the engineers on the project had by that time gon€ on to other activities and had to be hauled back "kicking and screaming" to work on something tl-rey thought they'd finished.

Manufacturing managers recognized the prob- lem, but they believed it was due to the fact that engineering cared more about the design than the product. The1, qepplained that epgineerilg liked to design "fi'om the ground up," instead of r-rsing standard components that could considerably lower the product cost. On€ complained that "if a fbu, engineers get really good at designing widgets, then they want to design a better, cooler widget fbr every product they r'vork on, instead of accept- ing that some other engineer's widget would do the job just fine." But one manufhcturing manager put the problem in the context of Dynacorp's growth:

When I first joined the clwpany, we were all still bnsically in the sarne tonn, nnd. we gzt to hnow each other. So if I bnd a problern with the d.esign, I'd just nralk oyer to engineering nnd grab one of the gwys to give me a hand.. Of course he'd grwrnble ablut ur rnanwfacturing guys being n bwnch of idiots, but he'd. either sit d.owru with ws or f,.nd one of the engineering tenw n,ho would. help owt-and half the tirne tbey'd. realize that we had a point, and. we'd worh things out. Now I'rn in a. pla.nt that's in a d.ffirent state and I don't really hnow any of the bench engineers the sarne way They hnnd ws a spec [specificationsJ f.le tbnls hund.red.s of pnges long, nnd they think their job's finished. By the tirne ure'pe id.entified. a problern, they're nll off on nen, projects, and. ftnd.ing tbe ones responsi- ble for tbat pnrt of the d.esign, gening their ntten- tion, con'rincing them that we nrentt nwrnbsh.ulls and that there really is a problern---uell, that all tohes a lot of tiwe. And cbanging the specs is a big d.enl, becawse et)erylne has signed off on thern, so e'pen a. ruinor change becornes a rnajor tirne sinh.

The only thing on which engineering and man- ufacturing people unanimously agreed was that one reason fbr falling profits was marketing's eagerness to cut prices to make a sale. The view that "those fmarketing] guys are measured on sales volume, and to make their targets they'll let the customer squeeze the margins" was wide- spread. Several engineers also expressed the view that marketing also bore some of the rcsponsibiliry

Module 2 . Three Lenses on Orxanizational Analysis and. Action M2-3L

for products coming late to market, because they kept trying to add features or functions during the development process, even in the late stages, when changes were increasingly difficult and costly. According to one engineer, "They come in when we're nearly at the last gate to say, 'FIey, lr,e've talked about this product to one of our best cus- tomers and they think it would be cool if we had a model that would yodel the Star-Spangled Banner on the Fourth of July' or some other crazy idea. We can usually talk them out of it, but it can sure slow things down while we sort them out. The problem is that those marketing guys have no idea of how much engineering is needed to change the functionality even by a small increment."

To meet the challenge of linking the ftinctions in product development, Dynacorp relied heavily on two linking mechanisms: cross-fwnctional prod.- wct deteloprnent tea.ws, with representatives from manufacturing and marketing as well as engineer- ing, and the position of prod.wct lnona.ger, who had the formal responsibility of overseeing the entire process from initial conception to product launch.

In their interviews inside the company, the design task force members heard widespread com- plaints about the cross-ftinctional teams, especially from manufacturing and marketing. One manu- facturing manager put his misgivings this u,a)':

We d.on't really have cross-fwnctional prod.uct d.eveloprnent teq.rns; we hnpe cross-fwnctional prod.act d.evelopruent rneetings. But eueryone is sit- ting there wenring their fwnctional bats, and. no one really listens to a.nylne else. They close rnnks against tbe other fwnctiorus. Euen if the d.esign engineers d.isngree ernlng thernselues, they'd. neter bring the isswe to the tearn to discwss and. resolvel they'd worh it out nwong thenoselpes and tberu ptres- ent e wnited front in swpp7rt of "the engirueering solution" at the next ,,neeting.

Another manager pointed out that the com- pany did not have an easy time identifliing and developing good product managers:

Id.eally, a prodwct wa.na,ger would. hnye sowe evpe- rience in ench fwnction. Bwt here nt Dynncorp we nll woye ap within n fanction, and. ven, few peo-

ltle rnove across the functional bowndaries in their ca.reers. So the prodwct rua.nflgers clr'ne t0 tbe job with a fairly n&rrlw experience base. Theybe all been on cross-fwnctionnl tea.ws, bwt in a func- tional role. They have to learn on the job-fnst The few really good prod.uct ma.na.gers that we baue nre in bigb dernand., and. are too bwsy rnan- nging prod.wct teaTns to train anybod1 else. We burn owt a lot of good. fwnctional pelple whl get prowtoted. to be prod.wct lnt.na.geys and then find they aren't up to the job.

As one product manager (who was widely regarded as very successful) pointed out, the job itself 'w'as a very dlfficult one.

I don't haye ynwch in the ww of carrots and. stichs to inflwence the behavior of the team rnewbers. I have an inpwt into their evalwations, but the epal- wntions theynselves are d.one by theit, fwnctional vna.nager5 nnd it's the qwnlity of their fwnctional expertise that d.eteyrnines bow they are l)a.lwed., rult their abilitl, to subordinate fwnctional criteria to prodwct success. And. becawse ench engineer worhs on anywhere frow six t0 twenty prljects in a yeary rny inpwt is only a srnnll piece of the overall epa.lw- ntion of the engineering project wewbers. It gets epen rn07,e clvnplicated: if I gite slr'nelne who is seen q.s a renlly gootl engineer e p00r evelaati7n becawse be\ n bad teatn plnyer, word. gets arownd, nnd then I'm going to hnpe trowble gening good people onto wy tea.wrs. Fwrtherrnore, tbe bwdget allocationsfor the prlject are controlled. by the sen- iorfwnctionnl rnanagers, nlt by vne or by tbe proj- ect lead.ersfrorn each fwnction. If tbe project needs mlre resl!4rces) I hape to coax thern owt of the fwnc- tions. So I bave a lot of responsibility ntithowt the

forwal awthority to bnck it up.

The internal investigation also collected a num- ber of more general criticisms of Dynacorp's organization design. Younger members of the organization complained that the organization had too many layers of managers, and even some of the older rnanagers who had benefited by the prolifbrati<xr of steps in the career laclder looked back nostalgically 6n the flat, speedy organization of their early 1'ears in the company. Information seemed to take a long time to travel up and down the compan,v. Several managers complained of a growing tendency to push problems Llp the hierar- chy fbr resolution, instead of taclding problems when and where they arose. One relatively junior marketing manager said that he had grown dis- couraged by the amount of time it took to get a response to his suggestions for ways of improving the customer databasc:

I went to w! rua.na.ger with n proposal, nnd. she sent it on to her rnanager, and. I d.ontt hnow where it went frorn there. Id.eas nnd swggestions go ap, nnd nzthing ever seerns t0 cllne down, or if aruything cornes d.own it\ n request for fwrther stwdlt. I've given wp trying to chnnge a.nything arownd here.

On the other hand, many of the employees expressed some apprehension about the mandate of the task force, fearing that a major redesign of Dynacorp would damage rather than improve the company. As one engineer said,

We nll cornplain abowt the orgnnization. Bwt I hnve friend.s who work in other high-tech clrmpa.-

M 2 - 3 2 A N A L Y T I C S ' T E A M S . O R G A N I Z A T I O N S . S K I L L S

nies, and. their problerus ,tre eten worse thnn ours. One friend's clTnpeny bas n product dipision stlnucture, nnd they can't seetn to transfer engi- neers 0r ruanufactwring capacity across prod.uct lines withowt huge ftghts nbowt wbo's going to pay

for it and how wwch. If wetre running lnte on one prodwct d.eveloprnent project, for exnrnple, senior engineering wa.negernent cnn sbwnt ingood engi- neers frorn other less implrtant prljects &t a. wlruenls notice. Or if one prod.wct is in wnexpect- ed.ly high dernand., a plant ,m&nager can n'itch lines qwichly t0 get the prod.wct volwwes lut the door. I would.n't wa.nt t0 see us lose that.

A marketing executive who had been hired from the outside made a similar point:

My old colnpa.n! was organized by prodwct divi- sions, nnd nie n)ere stepping nll oper ench other sell' ing d.ffirent prod.uct lines to the same custlrners. Sornetinces we n7eyle owr own fiercest corwpetitor,: the small systems custlruer representtltives were push- ing one solwtion, the large systems people a dffir- ent one. Thnt vas lne reosln I left, and I dontt nla.nt t0 see Dynacorp get into the snrue d.ilemwn.

External I nvesti gation: ldentifying Alternative Designs

Some melnbers of the task fbrce fbcused on the search fbr alternative organization designs. They surveyed the leading com;lanies in inforn'ration technology and communications (ITC) indr-rstry, talked with executives in some of those companies, and interviewed a set of consultants who specialized in organization design (and who r'vere eager to sell Dynacorp their services). They also ttrlkecl lvidr some of Dynacorp's own managers who had been hired out of other companies in the industry, to gct their views on the strengths and weaknesses of the organization design at their previous ernployer compared to Dl,nacorp's. Front these investiga- tions, task force members identified five models that thcy fblt were viable alternatives fbr the company. t Prorl.wct d.ivision stractul,e: Tl-ris design would

divide Dynacorp's functional divisions into product-based mr"rltifr"rnctional product divi- sions based on the product tcchnologl,, r,vith a general manager in charge of the entire value chain (from technology development through after-salcs service) fbr each rnajor product line. Related product divisions u,ould be grouped into business divisions (e.g., large systems) headed by an executive vice president. Shared corporate services would include not only the clrrrent set ofactivities but also advanced tech- nology development, which r'vould be in ellbct a corporate R&D center providir-rg cutting- edge technology 16r future product lines.

. Castlmer dip'ision stt/ucture: Dynacorp's func- tional divisions u'ould be divided into multi- hlnctional divisions, but the criteria for creating the divisions lvonld be primarily the cLrstonler segment tl-re}, 5s1t.4 (e.g., sn-rall business clivision), rather than the technology ofthe product.

o Functionnl/product tnatrix: This strucnrre u'ould matrix the existing functional structure with a product division structure, so that indi- viduals u'ould rcport both to a fr.rnctional rnan- ager and a product rnanager. Only or-rc company ir-r the industr)r was using this strlrctlrre slrccess- ftill1', and several companies that hacl adopted this structure had abandoned it after a fbu'years. Howo'er, rather than regarding it as a failure, nvo or three had employed it as a transitional strlrctllre to develop the people and the systems to move more easily fiom a flinctional to a prod- Llct-based organization design.

. Front/bnck'This nervest strllctllrc, adopted by a very small number of companies, was strongly advocerted b1, 5s1,cpn1 of the consnltants as the "cr"rtting edge" of organizatior.t design in the ITC field. The structure woulcl keep tl"re mar- keting division as a separate front-eud struc- tllre) responsible fbr selling the entire range of l)yr-racorp products and systems. Hor'vever, it u,oulci enhance its tcchnical support capabilities to cnable it to delivcr svstems and soh,rtions morc eflbctively to cnstomers. Engineering and manufbcturing would be divided into prodr,rct- based business units that wor-rlc1 be responsible fbr dcsign and production.

o Functional structure with stronger linhing mechanisrns: This most conservative altcma- tive would keep Dynacorp's functional struc- ture, blrt u,ould set up a much stronger array of cross-functional linking rnechanisms and alignrnent s)/stcr.ns to support cross-fiurctior-ral Iinkage.

None of the designs seemed to have been suc- cessful in ever1, colnparly that tried it; each had success ar"rd fhilure cases rvithin the indr-rstrlr The front/back stlLlcture was so neu, that no obvious "failure" cases had yet ariscn, but even sorne of its strongest adr,ocates in the industry admitted that it had proved more challenging to implement than they had initialll, expectecl.

After tl-re tersk force had digested bodr the infbr- mation on the probler-ns of Dynacorp's cLlrrent structure and tl"re mapping of tl-re five design alter- natives, the task fbrce began the discussion with a quick show of l-rands on each member's first and second choice fbr a new design. The result showed a wide distribution across the five alternatives, and it was clear that no quick consensus r'vould point to the best r'vay foru'ard for the companlr.