a dynamic theory for sustaining process improvement

A DYNAMIC THEORY FOR SUSTAINING PROCESS IMPROVEMENT TEAMS IN PRODUCT DEV ELOPMEN T

Elizabeth K. Keating and Rogelio Oliva

ABSTRACT Process improvement has beco~ on imperative for businesse:i seekinl! competitive advantage. However, finns often expcriem;e difficullies in both gaining acceptance for improvement programs ond sustairung lhem, partictilarly in product developmenl. We describe a theory lo explain the 1mliation and sustamab11ity of process improvcnlf enhancmg future productivity. Without sufficient management support and organii.ational comrnitmenl, employees will con tm ue their normal day-to-day work and underill\lest in improvement activities. causing process improvement iruttncives lo fail. On the other hand, if the employees become engaged in improvement efforts, their normal product dev lopment projecL~ may become delayed. As a result, active process improvement programs can 'orr: belier dynamic (Forrester. 1961) that can activate team and interdcpart ' ures that undermine the improvement effort_ In addition. clue to the 1 I ys between the allocation of improvement effort and the generation pcclatie>ns created to en list participation ;ind motivate product dcvel • u trip tual progress. leading to disappointed teams and program Failure. To o~rcome lht problem. process improvement efforts can adopt a more undcrst.nted implementation approach and also reduce the improvement delays by decomposing the development process into less complex subprocesse$, each addressed by separate teams. while still maintaining oordination of the overall effort.

Process Improvement Team5 in Product Development

247

We present a dynamic theory of process improvement teams which augments the existing theories with: (l) feedbacks between improvement teams and the firm's physical, operational, and organi zational structures : and (2) the endogenous dynamics of the improvemen t process. The paper integrates theories from product development, teams, and process improvemen t literatures and relies on grounded, intensive ca.se study research o extend the literature. In particular, we draw from a study of a prod uct development cycle-time reduction initiati ve, hich utilized techniques-largely overlooked by existing research-to produce substantial process improvements.

RESEARCH METHODOLOGY Overview of the Research Project and Site The study presented in this paper was conducted as part of the Designing Sustainable lmprovemenl Programs research by the System Dynamics Group currently being done at the Sloan School of Management at MlT (Jones. Krah mer, Oliva Repenning, Rockart, & Sterman. 1996; Sterman, Repenning. Oliva, Krahmer. Reckart, & Jones, 1996). The purpose of this project is to provide the basis for a dynamic framework through which to understand the key detemunants of Lhe success or failure of quality improvcm nt programs. Our research involves detailed field study with four partner organizations to ground the formal models m an intensive longitudinal study of important improvement programs. The models will be synthesized into a management flight simulator-a simulation environment m which manage~ and students will be able to explore lhe long-term dynamics of improvement programs and design more effecLive programs. We employ an inter disciplinary research approach, that includes detailed longitudinal assessments of lhe organizations' experience with improvement programs, followed by development of hypotheses and theory grounded 1n dllta. The hypothes s are drawn from extensive fi Id research as well as existing theory in e>perations management (Chase & Aq uilano, 1989), quality (Deming, t 986; Juran 1969; Shiba, Graham, & Walden, 1993). and organizauonal theory (Cyert & March, 1963; Forrester, 1961; Lyneis. 1980) This paper describes and refiecLs on an innovative and successfu product development ycle-time reductio n initiative (referred to as the A hieving Proc ss EXcellence Teams or PEX initiati e) undertake n by Bell Laboratories engineers assigned to the Transm ission ystem Busi ness Unit (TSBU) of AT&T, subsequentl y absorbed into Luc nt Technologies after the 1996 trivesti tu re. TSBU' quality program involved all areas including manufacturing, product development, and supplier qua lity (McPherson . 1995). PEX was an important part of the overall initiative which culminated in the Ma lcolm Baldrige National Quality Award for TSB U and was thus selected for study Other studies conducted by ur research

248

ELIZABETH K. KEATING and ROGELIO OLIVA

group at AT&T/Luc nt focus on the history of quality (McPherson, 1995) and supplier quality (Krah mer, f997a, l 997bJ. Research Method The primary tool u d to develop the theory articulated in this paper was intensive h 1Bonoma, 1985; Eisenhardt. 1989: Yin. 198L. 1984). The case study r~ research w reim ~t1ve: the primary initiative studied had be n subsumed b\ another cros -p OJCCI team program at the time the research was undertaken Data we collectal through semi -structured interviews and re iews of rc hi vat data The researchers mtervi wed leaders and participants in the APEX ini1iat1ve. as wel I as other product developers, whose work was affected by the program Most interviews were conducted at the Merrimack Val ley Works (MVW) in North Andover, MA, a major product development location which develops new transmission products. although ome APEX leaders based in New Jersey w re interviewed by telephone Interviews were conducted according to a basic interview protocol and lasted 45 to 90 minutes Given t11e nature of the research, the interviewees were not required to stay with the standard questions. If the mterv1ewers felt that they were pursuinr a profitable avenue, the interviewee was allowed to continue in that direction. Quotes were taken from the trnnscription of notes taken nt each interview. Several participant~ were (;Ontacted subsequently to elaborate on issues raised or clarify comments . The intervrcws were supplemented with extensive review of archrval data including training books, quality improvement stones, h1stoncal perform::mce data. and company publications. The dala are summarized in the form of a detailed case study (Krahmer & Oliva l 995 >describing the history of 1he imtiutive. The case documents were provided tr· participants for their feedback; participants were asked to review their quotation. for accuracy but were not allowctl to change the content . Participants were al~ n asked lo review the entire case for accuracy. The research was supported and enhanced by a team of corporate advisors formed specifically for the study. The theory is articu lated with the aid of causal loop diagrams (Forrester, I 961 . Mausch. l 985; Rich;mlsnn , 1991; Richardson & Pugh, 198 l: Weick, J 979). Causal loop diagnm idea compact and precise representatio11 of the interdependencies in a re useful rn portraying the feedback slru ture of systems nnd feedback relationships are substantiated through Particular ing academic theory, and discussions with the research advisory grou The caus I loop Jt gTllm~ should not be interpreted as precise mathematical specifications or the relationship$, which may be linear or nonlinear. In subsequent work, the lhcury will be converted into formal mathematical models that will permit identification of crillcal interactions between quality programs and other organiz:itional structures, allow the theory to he tested against data for a wide range of quality

Process Improvement Teams in Product Developmen t

249

programs, and aid in policy design (for exam ples of formal feedback models of quality improvement programs, see Repen ning, 1996; Sterman, Repen ning, & Kofman, l 997). The fol lowing sectio n offers background regarding the APEX ini tiative . The case highlights the rationale for the cy le-time reduction initiative, the fai lure of earlier yc!e-time programs, the factors underly ing the initial and medi um-tenn success of the APEX teams. and the dynamics associated with its eventual termination . The APE. case along with exis ing theories are then employed to dev lop an integrated dynamic theory for sustaining product development process improvement teams.

THE ACHIEVING PROCESS EXCELLENCE TEAMS (APEX) IN ITIATIVE Recognizing the Importance of Quality and Cyc le Time

Responding to the Divestiture and Increased Competition After the 1984 divestiture of the Regional Bell Operating Companies (RBOCs). AT&T faced new and fierce competition for the telephony equipment market. To become more responsive to the new competition, AT&T reorganized its business operations in 1989, combining the transmission-related manufacturing facilities of Western Electric and the associated product development expertise of Bell Laboratories into a new division, the Tra115mission System Business Unit (TSBU) which immediately commiued to addressing several key business issues through the use of quality and process improvement programs. The focus on quality was not new to the employees in the TSBU. The programs prior to reorgan1z.ation often emphasized quick fixes, provided little training, lacked careful implementauon strategics, and generated more frus1.ration than change (sec McPherson. 1995 for case history of MVW's total quality management ffQM] initiatives) Given the previous quality improvement attempts, deve loping employee support fo r new quality efforlS became a management concern. In late I989 and early J990, two events helped galvanize suppon for quality improvements. The first was the re lease of the first customer report ard. To th surprise of many, rice was not the top customer concern; rather quality/reliability and features/functionality were paramount. One interviewee describ d the latter oncern as, "The customers wanted what rhey wanted, and they wanted it when they wanted it." dev loper noted· We nec:cU!d to compete with the Japanese. Bell Labs had the late.• l Jechnology, bu1 ii was no good did not see ll. AT&T was nol pecce1ved as havi ng lhe fQles t stuff. To recapture the RBOC cus tomcn; , such as Bel!A1lanlic, BellSouth, and Pacific Telesis. TSBU 's quality netdcd to be e11hanccd and lime to rn arlo:el l'l!duced i( 1he customer

250

ELIZABETH K KEATING and ROGELIO O LI VA

A mock Malcolm Baldrige National Quality Award (MBNQA) application process ge nerated a second source of momentum beh ind quality improvement efforts. The new President, Pete Fenner, brought in certified national MBNQ examiners to train TSBlJ managers on the award cri teria and evaluation process. On lhe last day. a MBNQA examiner scored the mock application and provided k. One attendee at this session remarked : "Fenner sent his extensive feed executives the mes age thnt he wanted TSBU to obtain the award in fou r years." Early Attempts to Reduce Cycle Time

Following th ,se events, Fenner held his Leadership Team to thei r ommitment to pursue clrnnge. One effort, led by Bell Labs Vice President Bill McCurdy, focused on time-to-market as a critical com petitive advantage for TSBU. Mccurdy vowed to reduce the Product Realization Process (PRP) interval-the time fro m specifications of conceptual designs unti l re lease to manufacwring- by 50% in three years (1989- 1992). Although the lack of precise PRP in terval data was a significant problem, they esumated thnt the development of a full system was averaging )Q months. Red ucing the interval from 39 lo 19 months by 1992 (three years) becam~ the goal of the interval reduction initiative_ As one interval reduction champion recalled: The interval data wus imp~c1se and not complete. The 39-monlh esti mate may not have been a pure or certified nurnbe1, but I thought it was a pretty good number, and we needed to put a stake in the grollnd

Mccurdy named Al Hofmonn, a New Jersey-based Bell Labs director with di rel development respon.~ i bilit1es, as project uirector for the reduction of PRP interval Hofmann immcdia.tely launched two initiatives onented to tapping into the exper . l1se of successful product deve lopment teams and disseminaung their lesson< through the rest nfthe organizotion. These early defined a1 !he time fmm GAiElll IProject Dei inilion)to Gr\TEVIDep loymenll Sin
Failure Modes of Improvement Programs ~

There are many environmental reasons why improvement programs fail or wane, for example. reorganization. lechnological obsolescence, and market changes (Oliva Rock.arl, & Sterman, forthcoming). Jn contrast, we identified several structural factors ' ndogenously mak sustaining improvement programs difficult The self-111'! dilemma highlights a fai lure mode; 1f employees do not have adequat.c I tion to engage m improvement effort~. the program does not live bry In some cases, a process improvement program is successfully l ugh managerial pus h but does not yield improvemen t results due Lo the long lime delays associated with introducing operational cha nges and translating them into productivity gains. Ocher programs , such as APEX, succeed in generating substantial productivity gains yet later falter. In this section . we will use the feedback structure lo examine two late-emerging fai lure modes that

~~~=:a:~

(

Org1nlzatlonal 111d Techntul ComplHlty

~ ~

\ + Tim• Required " lat lmprov1m1nl ,..,...- Op1r1tlonal .,._....,, fi\ + Changu ' Tim• ., Alloea l•d lo lmprovimenl &95'1

•mp~:•m•nt

@

R•:r

>::J ,

Commitment to Perce ived TinglbllltJ lmprovem1fl + Jmprvomant i nd Proi!m ly ~of R.aulta

Figure 9. Comp lexity Slows the Improvement Proces.-.

276

ELIZABETH K. KEATING and ROCE.LIO OLI VA

Time Time ~ Allocated to Allocated Improvement

example, several product developers complained that the simpl istic QI methodology could not tackle difficult product development problems. As the QI techniq ues were inflexible, the QI teams were unable to address effoctively the relatively comple:it problems. Tilis led to a substantial drop rn tJ1e participation in QI teams within the product development area after two years. Finally, due 10 the srowing complexity, it became increasingly difficu lt fo r product developc'II to relate the benefits- oth potential and realized- of thei r improvement ffon to their deve!opmenl work. With low Tangibility and Proxim ity of Benefits (e pcclancy and instrumentality), each member began to behave as passenger or free rider, receiving recognition for the improvemenl results of the whole team witho ut mak ing the personal com.rrUtment to undertake the improvement effort. As a result, the process improvement Learns were further weakened by the B5 self-correcting feedback process. Better Product Development Prod ucts and Processes Boost Customer Demand As previously described, the rewards from a success fu l improvement in itiative include greater productivity, higher product quality, and foster time-to -market. These wckome improvements permit an organi7..ation to book additional customer orders. increasing profitability The supplemental customer demllnd, however, generates throughpul pressure that can undermine the continuation of the improvement initiative. In the case of AT&T/Lucent. three factors resulted in a substantial rise in throughput pressure First. the TSBU managemenl modified their marketing plans once the APEXgencrated product i.levelopment productivity gains became apparent. For example, two senior managers reported that AT&T responded to the enhanced produc1i11ity by increasmg its willingness to accept customer orders and increase throughput rather than ensuring additional productivity gains by explicit managerial measures designed Lo sustain the improvement effort While the managerial response to great.er product development efficiency benefited TSBU financially through higher orders, it also produ ed a side effect-less Time Allocated to Improvement-that undcnruned the APEX initiative and lessened future productivity gains (Loop 86 in Figure 10). he success of the APEX teams and improvement efforts in Second, ri;on, 1995) and supplier quality (Krahmer, J 997a. 1997b), more altrnctive with hig her quality and shorter dehvery time The a chc M kolrn Baldrige Nati onal Quality Award by TSBU ti I new orders from quality-conscious customers, thereby helped attr c u achieving the ongm11l 01m of the TSBU amprovemenl program. It, however, created unanticipated side ffccts that undermined the ability to suslain the improvement effort . Specifically, the higher product quality and lower time-to-market, while producing the desired increase in c1ntomer orders. placed additional thrcmghput

277

Process Improvement Teams in Product Development

~

{+ ®J · \ to Work

@ BB

Throughput Throughput Pressure ......._ •_ + ,_. j

~ 1

~·

CUBIOmEtr •

r +

·· ~ +Operetlonal

Time to~ LJ Marke

Order& ...,..., +

"--N....

Changes ;::J; ) ~

Cycle Time j

87'

~ +

~Product

!A~

+ Improvemen t Quality~ Results

~ Productivity~

Figure 10. Better Product Deve lopment Products and Processes Boost Customer Demand

prr:.~sure

on the Bell Labs engineers' processes. creating two balancing forces (Loops B7 an thal have assis1ed in selling up interviews and collecting data. cspe.cially Len Wi nn. Bot> Hough, and Al Hofmann. In addition. we would like Lo thank Lhc other members of the

Designing Sustainable Improvement Programs project at MIT (Nelson Repenning, John SLennan, Andrew Jones, and Scott Rockart) for their valuable ommcnts and insight~. Pem11ss1on for use of personal quotes wa.~ granted by the individuals cited and by AT&T

NOTES QUEST Consulting 1s an AT&T inrcrnal comulling firm that offen cxtcnus.t improvemenr at llarley-Davidsan . Case Srud; .1vail•ble from QU1J1on, IT Sloan School of Mnnagcmcnc, Cambridge, MA. Jard:ui. PC. ( 19861. EJTects of rui Ulr1ns1c reword on inums1c mohvalmn A field exrcnmcnl. Arodtmy a/ Mwaag•m«nr Journal. Z9. 405-41 l Juran, J.M.\ 19691 . Mana11rnal bm1ktlirou;;h . ·\ ""'' •l)llceptriftlie mrlllaJitr'SJh. New York: McGu"' llill Knhmer. E.l\1 (I 9