om article 9

8/17/2019 OM Article 9

1/26

OrganizationScienceVol. 20, No. 2, March–April 2009, pp. 384–409

issn1047-7039 eissn1526-5455 09 2002 0384

informs ®

doi 10.1287/orsc.1080.0408

© 2009 INFORMS

Capabilities Unveiled: The Role of OrdinaryActivities in the Evolution of Product

Development Processes

Carlo SalvatoManagement Department, Bocconi University, 20136 Milan, Italy, [email protected]

In contrast to the prevailing interpretation of capabilities as collectives, this inductive study of product developmentin a leading design firm highlights the centrality of the myriad ordinary activities that may shape the evolution of capabilities. A detailed comparison of 90 diverse product development processes over a 15-year period shows, first, thatmindful microactivities carried out by individuals in and around the organization and at all levels of the organizationalhierarchy are central in shaping the content of the product development capability and its dynamic adaptation. Understandingorganizational renewal and competitive advantage may hence require a partial shift in focus from capabilities as aggregateentities, to the practical realities of core organizational processes. Second, this more fine-grained perspective leads to

a set of insights on how organizational renewal may be partially shaped by timely managerial interventions aimed atencoding successful experiments into higher-level organizational capabilities. Third, higher-level capabilities resulting fromthe conversion of heterogeneous experiences display higher process homogeneity and a permanent increase in performance,because of stabilization of managerial attention. My findings contribute to unveiling the concept of capabilities, extendingprior research on dynamic capabilities and organizational renewal and providing a lens for research on the microfoundationsof capability evolution and organizational advantage.

Key words: capability evolution; capability microfoundations; mindfulness; dynamic managerial capabilities;event-sequence analysis; optimal matching analysis

History : Published online in Articles in Advance March 6, 2009.

IntroductionCapabilities and their constant refreshment have a sub-

stantial effect on the long term prospects of many orga-nizations (Agarwal and Helfat 2009). Interpreted aslearned and stable patterns of collective activity, capa-bilities incorporate most of the idiosyncratic knowledgethat determines differential success rates across organi-zations. This source of observed heterogeneity in capa-bilities is traced back to the evolutionary history of theorganization as it interacts with the environment, and toresulting path-dependent complexity (Kogut and Zander1992, Nelson and Winter 1982). In this picture, orga-nizational agents play the role of homogeneous, mal-leable beings, whose knowledge and behavior are largelydriven by organizational routines and capabilities (Felin

and Foss 2005, p. 443; Levitt and March 1988, p. 320).Despite their path dependency, capabilities are alsoregarded as change factors, because firms can systemat-ically remodel their capabilities by creating “dynamic”capabilities (Eisenhardt and Martin 2000, Helfat 2003).Although much of this research sees a role for manage-rial deliberation in shaping dynamic capabilities, here,too, individual action is relegated to a secondary role.Dynamic capabilities are conceived as routines or proce-dures designed to overcome the local nature of learningimplied in individuals’ ordinary operations (Teece et al.1997, Winter 2003, Zollo and Winter 2002).

Received wisdom is therefore converging on an inter-pretation of capabilities as collective entities that directly

drive organizational heterogeneity and competitive out-comes over time (Felin and Foss 2005, Gavetti 2005,Winter 2003). Individual agents and their ordinary activ-ities are placed in the background, and their role ineffecting organizational advantage is largely disregarded.Resulting descriptions of how capabilities develop, andof how this evolution affects performance over time,may be perceived as not being “final,” because theyraise additional questions about the set of causes respon-sible for capabilities emergence and change. Althoughinsightful, these descriptions bear little relationship topeople’s day-to-day work. The purpose of this study isthus to unveil the capabilities concept by exploring how

mindful acts (Levinthal and Rerup 2006) of individu-als in and around the organization may explain theirdynamic renewal, thereby extending thinking beyond thecurrently prevailing view of capabilities evolving as col-lective entities.

Unconditional focus on capabilities as collectives isproblematic. Probably some of the mystery and confu-sion surrounding the concept of capability arises fromlinking this organizational-level construct directly toorganizational advantage, thus excising the role of themyriad intentional microactivities performed daily by

384


2/26

Salvato: Role of Ordinary Activities in Evolution of Product Development ProcessesOrganization Science 20(2), pp. 384–409, ©2009 INFORMS 385

organizational agents (Felin and Foss 2005; Felin andHesterly 2007; Gavetti 2005; Winter 2003, p. 994). Thisapproach raises at least two controversial interpretations.

First, although organizational capabilities differ amongfirms in form and detail, specific capabilities exhibit com-mon features, equally associated with effective processes.

These commonalities are sometimes labeled “best prac-tice” (Eisenhardt and Martin 2000, p. 1106). Capabilitiesper se may hence be characterized by greater equifinal-ity, substitutability, and fungibility across firms than cur-rent theory proposes. This suggests that the idiosyncraticfirm-level effects embodied in capabilities are proba-bly overstated relative to the idiosyncratic attributes of individuals and individual actions performed within theorganization (Winter 2003).

Second, the fact that organizational capabilities areinterpreted as being strongly path-dependent entitiessharply restricts the role of strategic choice (Dosi et al.2000, Helfat 2003). Strategic management is premised

on some role of managerial intentionality in driving thefate of an organization. Yet, one important manifesta-tion of organizational capability is the firm’s ability toadapt without having to rely on top management fore-sight (Burgelman 1996), or daily maintenance by agentsat different levels of the organizational hierarchy (Nelsonand Winter 1982, Murmann et al. 2003).

These controversial corollaries are apparent in main-stream literature on capabilities. Lacking focus on the“sophisticated workings” of ordinary individual actionwithin the organization, the capability lens regards out-comes of managerial problem solving as stochastic(Nelson and Winter 1982, p. 136). As long as the relevant

collective features of a capability are present, differencesin form and detail are deemed irrelevant (Eisenhardt andMartin 2000, pp. 1109–1110). However, by restrictingthe scope of analysis to organizational-level consistenciesand development paths, the primary sources of hetero-geneity in knowledge and knowledge transfer may be lost(Argote and Ingram 2000, p. 156; Nahapiet and Ghoshal1998, p. 247; Salvato and Leif 2008), along with convinc-ing attempts to closely understand how capabilities aregenerated and evolve (Zollo and Winter 2002, p. 341).

These problems are evident in recent empirical studiesthat tried to closely follow the emergence and devel-opment of organizational capabilities and of their com-

ponent routines (e.g., Edmondson et al. 2001, Feldman2000, Howard-Grenville 2005). These works reveal howseemingly mundane details may have surprising sig-nificance in driving organizational outcomes: collegestudents are effectively enlisted by free muffins andfriendly greetings (Allen 2002); complex computer net-works are successfully implemented as users deliber-ately ignore some software functionalities (Orlikowski2000); focus groups and customer visits drive successfulproduct innovation (Dougherty 1992). In these practi-cal examples, actors are not seen as simple automata

executing preordained behavioral programs, but as mind-ful interpreters of organizational capabilities. Accord-ingly, recent definitions of dynamic capabilities hint ata state of active awareness, openness to new informa-tion, and a willingness to view contexts from multipleperspectives (Schreyögg and Kliesch-Eberl 2007).

Understanding how routine-driven behavior impliedby a collective approach to capabilities may yield or-ganizational adaptation will hence require the com-plementary discussion of how elements of individualmindfulness underlie the emergence of adaptive capabili-ties (Langer 1989, Levinthal and Rerup 2006, Weick andSutcliffe 2006). The concept of mindfulness builds onthe psychology literature and is conceived as involvingattentiveness of individuals and their ability to respondflexibly to contextual cues. At the organizational level,mindfulness has two components: high levels of atten-tion to errors, unexpected events, and the numerous cuessurfacing from the organization’s environment and its

own processes; and the ability to effectively respondto such cues in a flexible manner. Mindful organiza-tional behavior is hence nonautomatic behavior char-acterized by an active awareness, in contrast with theroutine-driven, automatic nature of less-mindful behav-ior (Levinthal and Rerup 2006, Weick et al. 1999).

The foregoing discussion suggests that truly explain-ing organizational capabilities, their evolution, and, ulti-mately, variations in firm performance may requirestarting with individuals’ everyday actions as the unitof analysis. More specifically, I argue that understand-ing a firm’s ability to systematically renew its strate-gies and underlying capabilities requires an in-depth

understanding of the microprocesses that make up anorganizational capability and its component routines,of the day-to-day events that, at some moments intime, induce mindful alterations in such sequences, andof the role managerial intentionality has in leveragingsuch alterations with the aim of achieving systematicimprovement in capabilities. Therefore, in this paperattention is directed at incremental renewal in productdevelopment at the individual level and on resultingdiscontinuous transformations in the organizational-levelcapability (Agarwal and Helfat 2009).

To better understand the dynamic process throughwhich capabilities are adaptively renewed by means of

everyday individual activities, I engaged in an induc-tive study of the new product development (NPD) pro-cesses of Alessi, a world-class Italian firm in designerhome furnishings. Over the period of interest to thisstudy (1988–2002), Alessi evolved from a niche makerof precious stainless-steel objects developed by a lim-ited number of top-notch architects into a world-classmanufacturer of countless design products developed byover 200 designers in different materials, styles, colors,and technologies. How was this reorientation possible?What forces shaped Alessi’s core capabilities in design


3/26

Salvato: Role of Ordinary Activities in Evolution of Product Development Processes386 Organization Science 20(2), pp. 384–409, © 2009 INFORMS

and product development? What role did the ordinaryactivities carried out by individuals within and aroundthe organization play in these processes?

I addressed these questions by means of sequen-tial analysis methods (Abbott 1990). Over the 15-yearperiod, I tracked the event sequences through which

90 NPD processes unfolded, that is, 90 performancesof Alessi’s underlying NPD capability. I then comparedthese sequences with optimal matching and clusteringtechniques (MacIndoe and Abbott 2004, Pentland 2003,Sabherwal and Robey 1993). Investigating capabilitiesand their component routines according to their tradi-tional definition as semiautomatic sequences of individ-ual actions (Nelson and Winter 1982) offers a detailedand grounded vantage point for understanding their evo-lution, in that the origin of mutations can be tracedback to specific agents, actions, and junctures. I rec-ognized different clusters of NPD processes, and therelated different roles played by internal and external

agents and their day-to-day occupations in shaping capa-bilities’ innovation and performance.

Alessi shows how mundane details of product devel-opment processes may have remarkable significance.At Alessi, the minutiae of workday design practicesbecame a source of advantage in systematically shap-ing and responding to strategic change. Emerging resultssuggest that the coarse rendition of capabilities currentlyprevailing in organizational and industry-level literaturemay be significantly improved by reducing “grain size”to include everyday actions performed by individualsin and around the organization. Close engagement withthe actual performance of organizational capabilities can

unveil the real sources of their renewal and of organiza-tional advantage.

Following the typical format of inductive research,I begin by discussing theory building through my struc-tured event-sequence analysis of 90 NPD processes overtime. I then describe the data and the insights drawn fromthem, and conclude by tying these insights to the broaderagenda of exploring the evolution of organizational capa-bilities and its role in strategic renewal processes. Over-all, this work extends knowledge on capability renewalby offering a perspective on how adaptive capabilitiesdevelop, founded on a detailed and systematic under-standing of mindful individual acts. It also suggests

insights into the role managerial intentionality plays inthe development of capabilities, and into the impact theserenewal patterns have on performance.

Methods

Empirical SettingNew product introductions and novel designs have animportant role in strategic renewal (Kim and Pennings2009, Knott and Posen 2009). My study addressesthe microfoundations of strategic renewal through a

longitudinal investigation of NPD capabilities withinAlessi, one of the world leaders in designer home furnish-ings. Alessi represents an instructive case study of strate-gic renewal. Founded in 1921 by Giovanni Alessi, Alessiwas initially a metal- and lathe-works factory that manu-factured small, metal kitchen and tableware objects com-

missioned by external clients. In the 1930s and 1940s,Carlo and Ettore, the founder’s sons, joined the firmand gradually started to collaborate with external design-ers. These initial collaboration projects with outsidershelped Alessi understand the company “could do excel-lent things by hiring architects” (Ettore Alessi, Vice Pres-ident). When Alberto Alessi, Carlo’s eldest son, joinedthe firm in the early 1970s, design gradually became thedriving company logic. Over the 1980s and 1990s, sys-tematic efforts were made to collaborate with externaldesigners and architects. In those years—the focal periodof this study—Alessi gradually transformed itself froma small firm that sporadically attracted the attention of

design experts with unique, though isolated objects, intoa trendsetter, which since then has been systematicallyshaping the features of its target environment (Collins1999, Bhaskaran 2005).

This “cultural” lead has resulted in widespread marketrecognition, in Italy and worldwide, which entailed supe-rior financial performance. A comparison of Alessi’smain financial indicators with those of selected Italiancompetitors over the focal period reveals the positiveimpact these choices had on the bottom line. Alessihas consistently outperformed its direct rivals in theperiod of interest, as revealed by the significantly higherprofitability and cash generated by operating activities

(Table 1).Data available for Alessi consistently show a history

of superior adaptive capabilities and an enduring com-petitive advantage. It is therefore interesting to explorehow Alessi’s product innovation capability developed

Table 1 Comparison of Selected Financial Indicators of Italian

Producers of Home Furnishings: 1995–2002

Sales increase Average Average Average cashFirm 1995–2002 (%) ROI (%) ROE (%) flow /sales

Alessi 328 1817 397 978Bialetti 3594a 693 885 454Lagostina 475 228 −870 496

F.lli Guzzini n.a. 351 485 1180Metaltex Italia −67 583 753 630Abert 1758 386 254 951Pinti Inox −275 513 244 548Lumenflon 653 664 668 413Inoxpran −412 102 −962 147Frabosk 167 087 −187 3795Mepra −345 752 665 690Barazzoni −58 272 081 871Calderoni −174 037 −497 253

Notes . ROI, return on investment; ROE, return on equity.

Source . Databank “Competitors plus” reports, 2002 and 1996.aResulting from the merger of Bialetti and Rondine, a large pro-

ducer of metal pots and pans.


4/26


over the period that has been crucial to the firm’s strate-gic transformation.

Research DesignI drew insights on the microfoundations of strategicrenewal from a single-case design. I chose an induc-tive case-study methodology as I was interested indeveloping theory on the capability evolution processesthat underpin strategic renewal within the context of theorganization (Eisenhardt 1989).

The study format I adopted, however, departs fromcommon case-study methodology. I addressed the ques-tion of how capabilities evolve by tracking 90 prod-uct innovation processes that took place at Alessi overthe 15-year period (1988–2002) that was crucial to thefirm. Availability of a significant number of instancesof the focal unit of analysis—the product innovationprocess—allowed a structured analysis of collected qual-itative data. More specifically, I carefully reconstructedthe action sequences of the 90 NPD processes, cluster-analyzed them, and drew insights about how Alessireshaped its key capability to address the challengesposed by its task environment. This structured analy-sis of qualitative evidence improved the reliability andvalidity of emerging insights.

Sampling. At Alessi, I analyzed 90 NPD processesthat the firm initiated between 1988 and 2002. Thisperiod is suitable for my analysis. First, it includes aphase of major strategic reorientation and some yearsdirectly prior to it. In the early 1990s, Alessi resolutelyengaged in the development of plastic objects with thehelp of a whole new group of young designers who have

since dramatically reshaped Alessi’s catalogue in termsof product types, materials, philosophies, and develop-ment processes. This intense renewal stage resulted insignificant alterations of Alessi’s NPD capability.

Second, internal documents going back to 1987 andearlier were much less systematic and complete, whichwould have reduced comparability with subsequentyears. Key informants corroborated this impression dur-ing interviews by recalling that around 1988 the NPDfunction became more structured and began placingemphasis on process articulation and data management.

During the data collection process, I figured that timeand resource constraints would not allow me to ana-

lyze more than 90 projects (Van de Ven and Huber1990). Previous works investigating sequences of com-plex events focused on a smaller number of instances(e.g., five critical events in Isabella 1990, 25 strategicdecision processes in Mintzberg et al. 1976, 23 processdevelopment processes in Pisano 1994, and 53 infor-mation system implementation processes in Sabherwaland Robey 1993). Hence, 90 NPD processes constitutea relatively large database, given the nature of the datacollection process.

Because my aim was to investigate the evolution of NPD routines over time, the choice of the number of

sampled projects in each year was critical. I opted toselect the number of projects sampled each year (e.g.,nine projects in 1998, representing 10% of the wholesample of 90 projects) based on the proportion of allprojects developed by the firm that year (i.e., 206) tothe total number of projects (i.e., 2,142) developed in

the 15 years of interest (i.e., 206/2,142, that is, approx-imately 10% of all the products developed over the15 years of interest). This way, action sequences thattook place in years in which a higher number of projectswere carried out would have a relatively higher chanceof determining enduring effects on the evolution of the firm’s NPD capabilities. The underlying behavioralassumption is that routines and capabilities develop asa result of learning through repetition (Cyert and March1963, Dosi et al. 2000, Nelson and Winter 1982).

Projects actually sampled in each of the 15 focalyears were randomly selected in terms of the mainaspect of interest, i.e., the product development sequence.

In choosing projects, I was blind to the structure andcomposition of the product development process, whichresulted from later coding and analysis. However, inchoosing project dossiers to be coded and analyzed, I alsotried to ensure that enough variability was observed in theproject dimensions key informants indicated as relevant:strategic business unit (SBU; Alessi formally codifiesits products in eight separate SBUs: plastic, stainlesssteel, wood, porcelain, glass, electric appliances, histori-cal reproductions, miscellaneous); product type (e.g., cut-lery set, coffee maker, kettle, kitchen scale); productdesigner (over the focal period, Alessi collaborated withapproximately 200 external designers); NPD process out-

come (i.e., the product actually entered the catalogue or itwas withdrawn); product performance (i.e., actual salescompared to sales aspirations). This sampling strategy isnot uncommon in qualitative inquiry, where it is some-times labelled as “random purposeful” sampling (Milesand Huberman 1994); it aims to add credibility to thesample when, as in my case, the potential purposefulsample is very large.

As a result, the 90 selected projects were developedover 15 years (1988–2002) by 40 different designers,and they were quite heterogeneous in terms of producttype and sales performance. Finally, 10 projects werenot developed beyond the prototype stage.

Data Sources. Data gathering spanned the period2000–03, during which over 40 company visits werecarried out. The data collection tactic implied a mix of secondary and primary data and direct observation. How-ever, given the need to carefully trace sequences of prod-uct development processes over 15 years, the main sourceof evidence came from access to company archives.

Archival Data. Several hundred pages of documentswere available for each of the 90 sampled products. Of particular value to the study was access to the “NPD


5/26


dossiers” in which the firm files all documents relatedto each product development process. For each prod-uct Alessi developed until at least the prototype stage,there was a NPD dossier file containing, in chronolog-ical order, formal documents, letters, faxes, handwrit-ten notes, designs and drafts, pictures, small artifacts(e.g., color samples, removable color chips, small prod-uct components), minutes of meetings, e-mail printouts,newspaper clippings, brochures of related productsor materials, packaging items (e.g., labels, flattenedboxes, instructions, logos, recipe books), transcripts of speeches, technical data about products and produc-tion processes, photocopies from relevant books or cat-alogues, and results of focused technical tests. Accessto Alessi dossiers provided extremely detailed and com-plete descriptions of selected NPD processes, as con-firmed by three key informants.

The documents included in each dossier were selec-tively transcribed, reporting document type (e.g., letter,fax, manuscript note, consultant’s report, supplier’sproposal, etc.), date, the person who originated thedocument, the addressee, and, most important, abridgedcontents of the document with extensive quotes.Although selective transcription may have resulted inmissing details, raw data was available to resolve uncer-tainties. Despite the slight chance of missing data, directaccess to company documents warranted a reliabilityin reconstructing NPD processes that far exceeded thatachieved with alternative tactics (e.g., Mintzberg et al.1976, Sabherwal and Robey 1993). This allowed me tocarefully generate the event sequences of investigatedNPD processes.

Primary Data. Thirty-one interviews with twenty-twokey informants enriched my understanding of secondarydata and allowed me to investigate the rationale behindchoices, decisions, and actions traced through archivalevidence. Access to both internal and external infor-mants allowed me to investigate how routines and capa-bilities result and evolve from interactions within thefocal organization, as well as among internal and exter-nal actors. Informants’ collaboration with Alessi rangedfrom 7 to 59 years, with an average tenure of 23 yearsfor internal informants, and 18 years for external ones,hence providing an extensive first-hand coverage of thefocal 15-year period (1988–2002). Other primary data

included company visits, participant observation, andhundreds of informal talks ranging from a few minutesin front of the coffee machine to over one hour in thefirm’s cafeteria.

Data AnalysisData analysis consisted of four sequential steps. In the

first step, I reconstructed the event sequences of the90 product development processes, I coded all eventsaccording to a classification scheme, and I checked thereliability of event classification. Analysis of NPD pro-cess documents allowed me to carefully reconstruct the

event sequences of each of the 90 NPD processes. Forinstance, the development process for Branzi’s 1988wooden box (Product 1 in the appendix) consistedof 32 separate events, whereas 42 events were nec-essary to develop Venturini’s plastic clothes rack in1996 (Product 40 in the appendix). Altogether, the event

sequences of the 90 NPD processes yielded a total of 2,897 events (an average of 32.2 events each). This calledfor a scheme that could classify such events into mutuallyexclusive and cumulatively exhaustive categories. Theclassification scheme (reported in Table 2) was devel-oped through a number of successively refined versions,as suggested by Bakeman and Gottman (1997). A simi-lar approach was adopted by Isabella (1990) to classifyevents in one organization, and by Sabherwal and Robey(1993) to classify events in 53 different organizations.

The 18 coding categories were developed using acombination of deductive and inductive reasoning, as istypical in sequential analysis (Abbott 1990, Bakeman

and Gottman 1997, Blair-Loy 1999). Through a prepara-tory deductive phase, I defined the preliminary orga-nizing categories (conceptualization or “colligation”)(Abbott 1984) that guided the subsequent inductivedefinition of the final coding categories (measure-ment). I started by developing an initial categoriza-tion scheme of 25 types of events that took place atAlessi using a subset of the 90 sequences of events.Two research assistants—blind to the purpose of theresearch—independently classified a random set of 100events into one of the 25 types. They agreed on theevent type in only 76 cases, which was not consid-ered satisfactory (Isabella 1990, Sabherwal and Robey

1993). I then repeated this procedure twice, until I devel-oped a leaner and clearer classification of 18 typesby creating new categories, merging categories the twoindependent raters perceived as similar, and eliminatingcategories they seldom used. Adopting this new clas-sification scheme, the two research assistants indepen-dently classified another random set of 100 events intothe 18 categories. This time they agreed on 94 of the100 events. The 94% interrater agreement was comparedwith the chance classification accuracy using Cohen’s(1960) kappa. The kappa measure showed that the agree-ment between the two raters was significantly greater(p


6/26


Table 2 Classification of Events Within Alessi NPD Processes

Event type/

No. coding category Event description Examples

1 Desiderata Document that formally starts product evaluation

and development

81 Desiderata documents selectively transcribed

and coded

2 Internal development:

design

Activities carried out by Alessi employees aimed

at developing product design

The drawings of a new product received from the

designer are transformed into a detailed “rendering”


color definition


at developing new-product color(s)

The “color development form” is filled in by Alessi

new-product development staff for Phase 3 (out of six

sequential phases)


administrative issues


at dealing with financial and administrative

issues related to development of new products

A cost estimate for an internally produced item is

developed


production phases


at physically making a new product

The sequence and timing of the production process for a

new, internally produced item is defined


packaging


at developing the packaging of a new product

Packaging of a newly developed product is selected


NPD process check

up


at checking new-product progress along NPD

phases

The “new-product development” form (“Scheda SNP

Prezzi”) is filled in by Alessi new-product development

staff for the “Capitolato” phase

8 Internal development:critical issues raised

Activities carried out by Alessi employees aimedat assessing and solving difficulties and

unexpected events in new-product

development

Minutes of an internal meeting reporting problems anddifficulties on a new product


ad hoc approval

requested


at obtaining approval for nonroutine activities

The desiderata for an unusual new project is sent to the

CEO for approval

10 Everyday contacts with

designers

Contacts between Alessi employees and external

designers aimed at developing a new product

Minutes of a meeting between Alessi new-product

development staff and designers, in which the final

colors for a newly developed product are discussed

and decided


habitual suppliers:

product definition

Contacts between Alessi employees and

recurring suppliers,a aimed at defining product

characteristics before production is started

A drawing of the product is sent to the supplier for

examining production-related technical aspects


habitual suppliers:

administrative issues


recurring suppliers,a aimed at solving

administrative and financial issues

A cost estimate for realizing a product component is

sent from the supplier to Alessi


habitual suppliers:

production phases


recurring suppliersa aimed at producing new

items (typically early phases, e.g., first trial

production batch, because the study is

focused on product development activities)

A supplier reports that the first batch of a new product

has been completed and delivered timely to Alessi

A color supplier reports on the match between the

developed color “master” and the samples approved

by the designer

Alessi reports serious faults in the last delivered batch of

products to the supplier

14 Ad hoc modifications

by external actors

Activities carried out by external actors (i.e.,

excluding Alessi employees) aimed at altering

established procedures

Designer SG requires a preliminary study of traditional

pots and pans before starting own project

15 Ad hoc modifications

by Alessi personnel


at altering established procedures

Alessi engineer suggests gathering plastic samples of all

new colors developed for new products

16 Contacts with unusualcollaborators

Contacts between Alessi employees andnonrecurring collaboratorsa

A research lab is required to perform a study of the forcerequired to extract a cork

A research lab is required a study of the effects of garlic

on colored plastic

17 Information gathering

on unusual topics


at gathering information about topics, issues or

problems which are nonrecurring in analyzed

documents

Data on existing product colors used by a competitor

are collected and analyzed

Data on performance and price of spectrophotometers

are collected by NPD staff member

18 Others Events wit hin Alessi or in exter nal environment

that cannot be classified in the above 17 types

A minority of activities not falling under any of the

previous types

aA list of recurring/usual suppliers was provided by the company. Coded transcripts of company documents already reported whether a

supplier could be considered habitual. Critical assignments were solved with the help of key informants.


7/26


of individual event sequences, which work well when thenumber of sequences is small, and when they comprisefew events (e.g., Mintzberg et al. 1976). In my case,however, the relatively high number of sequences (90)and event types (18) examined required the adoptionof an automatic procedure to compute intersequence

distances. I relied on optimal matching techniques,which directly measure sequence resemblance. Optimalmatching methods (Abbott 1990, 1995) are a subset of sequence analysis techniques (Bakeman and Gottman1997) and operate by dynamic programming, which is aclass of iterative maximization techniques operating onstepwise processes. They allow distance measures to becomputed among any set of sequences represented bya string of well-defined elements, drawn from a rela-tively small total set. The most common sequence met-ric establishes a “distance” between sequences basedon how difficult it is to transform sequences into oneanother. The standard algorithm for alignment is theNeedleman-Wunsch algorithm, which calculates align-ments based on costs associated with substitution andinsertion (Blair-Loy 1999, Sabherwal and Robey 1993).I analyzed the sequences of NPD codes with AndrewAbbott’s software program OPTIMIZE© (release 2.18).Once computed, these distances can be clustered, scaled,or averaged to show patterns and to reveal any commonfeatures or trajectories (Blair-Loy 1999, Sabherwal andRobey 1993, Pentland 2003).

The third step in data analysis was a cluster analysis of the 90 NPD sequences. Optimal matching algorithms donot directly arrange observations into sequence patterns.However, they generate interval-level measures of resem-blance/distance between sequences (called Levenshteindistances) (Sankoff and Kruskal 1983, pp. 18–23) overa sequence data set. These measures can then be usedas inputs to clustering algorithms, which in turn make itpossible to develop patterns of sequences (Abbott 1995,Blair-Loy 1999, Sabherwal and Robey 1993). I classifiedintersequence distances by the cluster procedure avail-able in SAS. I explored several distance measures andclustering methods (Aldenderfer and Blashfield 1984).The (no-square) Euclidean distance and Ward methodyielded the most easily interpretable clusters.

The adoption of the sudden jump in agglomerationcoefficients as a stopping rule suggested a five-clustersolution. This was validated by an independent samplest-test. Each of the five clusters represents a differenttype of NPD routine, i.e., a variant of Alessi product-development capability. Many Alessi NPD routines seemto fall within a category that can be interpreted as the“standard” way of developing a new design object atAlessi, because they closely mirror sequences dictatedby internally formalized NPD procedures. All other clus-ters reflect some kind of “mutation” of the standardNPD capability (Nelson and Winter 1982, pp. 115–116).These modifications are typically the result of everydayexperiments by external and internal agents. They repre-

sent the different responses of the organization to bothenvironmental and organizational stimuli.

The fourth and final step of the analysis was an inter-pretation of cluster results. This was aimed first at gradu-ally developing an in-depth understanding of the specificreasons behind each alteration of Alessi NPD capability.

Tracing each alteration to the underlying organizationalevents—by going back to the raw data or directly askinginformants in focused interviews—allowed me to under-stand which specific actions and actors determined themodifications within the NPD sequence and accordingto what logic.

The other aim of cluster interpretation was to under-stand the performance impact of differences withinand across clusters.1 Differences—i.e., heterogeneity—within clusters can be assessed by considering themedoid sequence for each cluster (Kauffman andRousseeuw 1990). The medoid sequence is the individualsequence that is least distant from all the other individ-ual sequences in the cluster. The medoid NPD processis thus a real process that can be used as an idealtypicalprocess to describe the cluster. More important, describ-ing a cluster by its medoid allows to easily compute thedispersion of the other cluster sequences around it. Thus,by using the notion of medoid we can compute the min-imum, the maximum, and the average distance within acluster. This is an important feature because the disper-sion informs us whether any given cluster is highly het-erogeneous or, alternatively, highly homogenous (Aassveet al. 2007). Obviously, a homogenous cluster containssequences that tend to be very similar (i.e., similar to theideal-typical medoid), whereas a high dispersion suggeststhat sequences within the cluster are heterogeneous. Thisinformation is crucial when interpreting the meaning of clusters. Medoids and heterogeneity/homogeneity mea-sures for my five-cluster solution are reported in Table 3.Clusters vary in terms of their level of heterogeneity. Twoare relatively homogeneous (Clusters 1 and 5), i.e., themean distance and the maximum distance are low, andthe proportion of the total dissimilarity within the clusteraccounted by the medoid (“Percentage within”) is high,whereas for the others (Clusters 2–4) dispersion aroundthe medoid is high.

In addition to considering heterogeneity within clus-ters (“Percentage within cluster” in Table 3), I reporta measure of heterogeneity across clusters (“Percent-age among clusters”). This is computed as the propor-tion of the total dissimilarity within the whole sampleof 90 sequences accounted by sequences within a clus-ter. Results discriminate even more markedly Clusters 1and 5—closer to the “average” Alessi NPD process—from Clusters 2–4, which are highly heterogeneous whencompared to the whole sample.

Finally, I computed two product performance mea-sures (Table 3): the percentage number of bestsellerproducts (illustrated in the appendix) in each cluster,and the average sales performance of products in each


8/26


Table 3 Cluster Heterogeneity and Product Performance

Heterogeneity/homogeneity measures Performance measures

Percentage Percentage

within among Percentage of Average salesCluster Size Medoid Mean Max cluster clusters best sellers performance (%)

(1) (2) (3) (4) (5) (6) (7) (8) (9)

Cluster 1 36 (49—“Titan”) 0.299 0.563 234 302 11.1% +68

“Recipe book” (4/36)

Cluster 2 12 (11—“Helmut”) 0.452 0.527 92 111 8.3% +6“In-house adapted” (1/12)

Cluster 3 9 (7—“Mix Italia”) 0.416 0.623 193 81 11.1% −28

“Externally adapted” (1/9)

Cluster 4 13 (41—“Placentarius”) 0.586 0.758 77 46 23.1% −35

“Externally driven” (3/13)

Cluster 5 20 (16—“Gino Zucchino”) 0.176 0.594 323 536 25.0% +127“Recombinant” (5/20)

Higher value= higher Higher value= lower

heterogeneity het erogeneit y

Notes . (1) The five rows report data concerning each of the five clusters of NPD sequences emerging from analysis.

(2) “Size” is the number of sequences in the cluster.

(3) The column “Medoid” reports the cluster’s medoid sequence, i.e., the individual sequence that is least distant from all the otherindividual sequences in each cluster.

(4)–(5) “Mean” and “Max” are the mean and maximum distances between sequences in the cluster and the medoid. The higher these

values, the higher heterogeneity within a cluster.

(6) “Percentage within cluster” is the proportion of the total dissimilarity within the cluster accounted by the medoid. The higher this

value, the lower the heterogeneity within a cluster. Hence, for instance, sequences in Cluster 5 are relatively more homogeneous amongeach other than sequences in Cluster 4, which are relatively more heterogeneous among each other.

(7) “Percentage among clusters” is the proportion of the total dissimilarity within the whole sample of 90 sequences, accounted by

sequences within a cluster. The higher this value, the lower the heterogeneity of cluster sequences compared to the whole sample. Hence,

for instance, sequences in Cluster 5 display more similarities to sequences in the other four clusters than sequences in Cluster 4, whichare highly dissimilar from sequences in the other four clusters.

(8) “Percentage of best sellers” is the percentage number of best-seller products (illustrated in the appendix) in each cluster. A best

seller is defined by Alessi as a product whose actual sales (volume) exceeded aspirations by at least 200%.

(9) “Average sales performance” is the average difference (for all products in each cluster) between (volume) actual sales and salesaspirations set by Alessi.

cluster, computed as the average difference between(volume) actual sales and sales aspirations. Thesedescriptive data were traced to explore possible relationsbetween degrees of heterogeneity/homogeneity in NPDprocesses and product performance.

Analysis and FindingsThe preponderance of literature tends to describe thespecific capabilities of an organization as homogeneousentities (Felin and Foss 2005). In contrast to this com-mon view, my investigation of the microprocesses thatembody a capability reveals that actual enactmentsof Alessi’s NPD capability cluster into distinct types.In other words, Alessi’s ability to develop new designhome furnishings takes different forms when expressedin different projects and over time. These differenceshave performance consequences. My analysis allows afine-grained interpretation of how Alessi’s NPD capabil-ity evolved over the period of the firm’s most intenserenewal efforts. In this section, I will first describethe five clusters and forces that shaped their distinctivetraits. Second, I will offer an interpretation of the logicdriving capability evolution, and of its impact on productperformance.

Heterogeneity of NPD Capability Patternsas Revealed in ClustersThe appendix shows the 90 NPD sequences groupedaccording to the 5-cluster solution resulting from myanalysis. I interpreted clusters by carefully analyzingcluster medoids, by going back to the raw data, whichprovide a rich description of each sequenced action, andby referring to secondary data or to interviews in whichthe focal project had been mentioned by informants.

Cluster 1 “recipe-book” project sequences includes36 NPD projects (40% of the total sample). Theseprojects are a closer—though never perfect—match forthe “ideal-type” NPD sequence than those grouped in

the other four clusters. The “ideal-type” NPD sequencewas reconstructed from a company document developedin November 1989 by Alessi’s most experienced engi-neer. The document illustrates a detailed sequence of allthe activities typically involved in the development of aproject. It was developed by systematizing Alessi’s bestpractices in new product development in previous years,and it was meant to provide formal guidance for futureprojects.

As my data suggest, 40% of the 90 NPD processesthat I observed over the period 1988–2002 followed


9/26


relatively closely what Alessi considers a conventionalNPD sequence, as key informants confirmed. An Alessi“recipe-book” product development sequence is typicallyprimed, with very few exceptions, by the “desiderata”(Type 1; see Table 2), which is the routine throughwhich Alessi sets the basic guidelines for product devel-

opment, in line with designer’s intentions and com-pany targets. Right after the desiderata, “recipe-book”sequences normally feature contacts with the designerwith the purpose of defining details of the preliminarydrawings (Type 10). Such contacts will recur severaltimes throughout the sequence. Contacts with designersare followed by internal activities (Type 2)—and relatedexternal ones (Type 11) involving suppliers—aimed atdeveloping product design in line with designers’ direc-tions and Alessi’s targets. These activities allow Alessi todetermine the technical details of the design through aninternally developed rendering and in close contact withsuppliers of raw materials and components. After design

details are defined, internal/external activities aimed atdefining production details (Types 5 and 13) and admin-istrative issues, such as the cost of moulds or compo-nents (Types 4 and 12), follow. These preliminary phasesare typically followed by a checkup of performed activ-ities in the form of a revision of the desiderata (Type 1)or an R.U.D.E. process-checkup meeting (Type 7), orboth. These checkup stages usually lead to further prod-uct development phases, both internal (Types 2, 4, and5, and 3 if the product is colored) and involving exter-nal suppliers (Types 11, 12, and 13). Toward the endof the process, product packaging is usually decided

(Type 6) and final checks are carried out (Types 1 and7). Critical issues (Type 8), ad hoc approvals (Type 9),ad hoc interventions by internal (Type 15) and external(Type 14) agents, or other “unusual” activities (Types 16and 17) are only seldom significantly present, reinforc-ing the interpretation of these processes as occurring ina rather “smooth” and uneventful fashion.

Cluster 2 in-house-adapted project sequences in-cludes 12 NPD projects (13% of the total sample). Theseprojects show clear deviations from what can be consid-ered a “standard” NPD process at Alessi (i.e., the aver-age “recipe-book” project sequence in Cluster 1) in thenature of activities involved or in the sequence, or both.

Such variations are, in most of the instances, the resultof extemporaneous day-to-day activities of agents pri-marily from within Alessi.

Compared to “recipe-book” projects in Cluster 1,“in-house-adapted” project sequences exhibit a highernumber of activities that signal alterations in the smoothfunctioning of the product development routine: criticalissues raised by Alessi employees (Type 8 in Table 2),requests for ad hoc approvals (Type 9), ad hoc modifica-tions of “standard,” “recipe-book” procedures introducedby Alessi personnel (Type 15) and, although less often,

contacts with unusual collaborators (Type 16), and infor-mation gathering on unusual topics (Type 17). A closeranalysis of sequences—by going back to raw data, fur-ther secondary data, or interviews—suggests that theseactivities carried out along the project, though ordinaryand apparently uneventful, altered the NPD sequence.

At least four Types of these internal adaptationsemerged: first, suggestions by Alessi personnel toinvolve external consultants following project complica-tions (e.g., Thun’s “Campari” cocktail shaker, Product 2and Rossi’s “La Cubica” cooking box, Product 8); sec-ond, adaptations improvised during early experimentsaimed at discovering new design talents (e.g., Starck’s“Hot Bertaa” kettle, Product 6); third, the suggestion toadopt a sort of “target costing” method to reduce prod-uct cost of Castiglioni’s AC04 fruit bowl (Product 27);finally, internal mutations because of the peculiarities of the designers involved as well as of the objects devel-oped (e.g., “Girotondo” Products 34, 48, 58, 66, and 79).

Cluster 3 externally adapted project sequences con-sists of nine NPD projects (10% of the total sample).These projects have in common variations from themore standard sequences in Cluster 1, which tend toresult from everyday adaptations introduced by agentsexternal to Alessi. These product development processesare characterized by the presence of ad hoc modifica-tions by external actors (Type 14), and by some relatedcontacts with unusual collaborators (Type 16) and infor-mation gathering on unusual topics (Type 17). Some-times (e.g., Sequences 7, 59, and 60) these externalalterations raise critical issues inside Alessi (Type 8).Consequences of these alterations on the product devel-

opment process require specific internal ad hoc approval(Type 9), e.g., in Sequences 35, 67, and 78. Threetypes of these external adaptations emerged. A firstinstance are those daily experimentations and sugges-tions by external actors aimed at improving color devel-opment routines (e.g., NPD Sequences 15, “Firebird;”59, “Dr. Kleen;” and 60, “Rondo, Sden, Otto”). A secondtype of external mutations are those driven by specialdesigner traits, as in the case of the “Mix Italia” espressocoffee maker (1991, Product 7), designed by Giovannoniand Venturini. Giovannoni’s set of “Mami” pots andsaucepans (Product 62, 1999) is a third example of anNPD sequence altered by designers intervening in the

process. In developing what was, since its inception, avery demanding project, designer Giovannoni decided tofollow an unusual procedure that resulted in a signifi-cantly altered NPD sequence, in which several internalproduct development activities are carried out early inthe process, whereas most of the designer’s activities,which are usually at the inception, follow. A final type of external adaptation relates to the development of a newedition of a historical Alessi project (e.g., AlessandroMendini’s “How much white” table set in white porce-lain, Product 35).


10/26


Cluster 4 externally driven project sequences in-cludes 13 NPD projects (15% of the total sample). Theseprojects are reproductions of historical objects, such asthe metal tray from the Pompeii excavations, or a teainfuser designed by Bauhaus architects in 1924 (Prod-ucts 5, 13, 25, 33, and 41), or they are the result of

collaborations with unusual industrial partners such asPhilips, 3M, or Häagen Dazs (Products 32, 36, 51, 53,56, 76, 80, and 87). Hence, just as in Cluster 3, Cluster 4project sequences are shaped by external forces. Themain difference is that whereas “mutations” in Cluster 3were determined by specific individual actions withinthe framework of ordinary Alessi products, here, prod-uct development sequences differ from “recipe-book”sequences because the underlying product idea itself arises from an interaction between Alessi and someexternal actor. Therefore, Cluster 4 projects can alsobe interpreted as adaptations of Alessi’s NPD capabilitymade necessary by externally driven choices.

Finally, Cluster 5 “recombinant” project sequencescomprises 20 NPD projects (22% of the total sam-ple). In contrast with the four previous clusters, thesesequences encompass patterns of activities that resultedneither from following rules codified before the sampledperiod (1988–2002), nor from experiments performed inthe course of the project. Rather, activities and patternsthat seem to pool these sequences were shaped by man-agerial actions geared to reproduce select improvisationsobserved by managers in previous years as internal orexternal mutations. In other words, sequences in Clus-ter 5 incorporate internal (Cluster 2) or external (Clus-ters 3 and 4) experiments that occurred in previous years,

and that Alessi managers recognized as potentially valu-able improvements to the NPD process, and eventuallyselected and retained.

These higher-level managerial activities regard threemain areas. First, some projects result from execu-tives’ intervention aimed at structuring the workshop,a procedure aimed at discovering and attracting newdesign talents worldwide. Projects 14, 16, 18, and 19resulted from the Family Follows Fiction FFF work-shop, a more structured and refined version of theMemory Containers: Creole Project workshop organizedsome time before. Projects 26 (Memory Containers: Bio-logical Project) and 70–72 (subsequent versions of the

FFF workshop) also resulted from further refinements of the original workshop process and structure.

Second, Projects 28, 42, 43, 44, 52, 73, and 82, whichgenerated colored plastic products, share a new proce-dure for developing product color. This procedure wasdeveloped by Alessi’s management from the embryoniccolor-filing system that emerged from both the internaland external mutations classified in Clusters 2 and 3. Thedevelopment sequence of Giovannoni’s “Mary Biscuit”plastic box (Project 28, 1995), for example, displays thefirst signs of the development of a structured color-filing

system that would eventually allow the firm to easilycommunicate with both designers and suppliers. Theseinitial experiences later yielded a full-blown formalizedfiling system (“The Color Box”), adopted in the devel-opment of nearly all new colored products.

Third, new editions of successful past Alessi prod-

ucts (e.g., “Colorbavero” table set, Product 74, 2001,the decorated version of Castiglioni’s “Bavero” tableset developed in the 1990s; “B 9093, 9093 GD” ket-tles, Product 84, colored versions of the famous kettledesigned by Graves in 1985) were the result of a system-atic search for opportunities to revitalize old productsthrough colored decorations or minor technical improve-ments, which followed early experiences of new edi-tions of old products (falling in Clusters 3 and 4). Thisresulted in a project known as “New edition of histori-cal projects,” launched by Alessi’s CEO in 1996, whichconsisted of meetings with designers, presentations toemployees, and formal documents reporting basic direc-

tions for developing new versions of old products.In summary, the 90 sampled NPD sequences fall intoa few general patterns characterized by minor within-group variation. These five patterns represent as manyenactments of Alessi’s NPD capability. Forty percentof the expressions of this capability closely mirrorthe NPD process sequence formalized by top man-agement. Over one-third, the “in-house adapted,” the“externally adapted,” and the “externally driven,” showalterations determined by everyday experimental activ-ities performed by internal or external agents. Nearlyone-fourth, the “recombinant,” ensue from intentionalattempts to sift out, refine, formalize, and reproduce

apparently promising, though minor, mutations.Capabilities Evolving: Interplay of Improvisationand IntentionalityEmpirical evidence provided so far portrays an orga-nization whose core routines and capabilities developmore as a result of everyday, mundane activities thanof managerial cognition. This would shed a rather dimlight on managerial prospects of intentionally improv-ing capabilities and facilitating strategic renewal. In spiteof that, my data show that the emergence of whatI have termed “recombinant” processes (Cluster 5) showsclear signs of intentional managerial intervention. AtAlessi, I traced—through additional secondary and pri-mary data—seemingly intentional managerial activitiesthrough which managers selected, developed, and repro-duced interesting “experiments” that had occurred inClusters 2–4 sequences. These activities were alwayssubsequent to the “experiments” they aimed to refineand reproduce, whereas they preceded the “recombinant”sequences (Cluster 5), in which refined practices are laterinserted. Table 4 summarizes the supporting data.

The eight rows in Table 4 report distinct instancesof evolutionary patterns followed by product develop-ment activities at Alessi. The first column describes


11/26


Table 4 Reproduction Patterns of Improvisational Acts: Summary of Evidence

(IV) Observed “recombinant”

(II ) Internal/exter nal (III ) Intent ional selection sequences (C luster 5)

(I) Description of activit ies adaptations (Clusters 2–4) and development of (reproduction of developed

involved (instances of improvisation)a adapted traits improvisations)a

1. Development of a set of

routines tosystematically

encourage projects of

young designers not

previously known to

Alessi (the workshop )

In-house adapted (Cl.2)

6, Hot Bertaa, 1990 (F)11, Helmut, 1992 (F)

17, Tin Man, 1994 (T)

64, Gsâa, 1999 (BS)

Externally driven (Cl.4)

56, Hi Fi set, 1998 (n.a.)

• Appointment of Metaproject/

Workshop coordinator (1991)• Development of workshop

routines (over 10 workshops

per year) and documents

(1993 onward)

• Workshop schedule posted on

company website (2000)

14, Merdolino, 1993 (BS)

16, Gino Z., 1993 (BS)18, Molly, 1994 (BS)

19, Escar-gogò, 1994 (F)

26, Black J., 1995 (T)

68, Tralcio muto, 2000 (T)

69, Strawbowls, 2000 (F)

70, Te ò, 2000 (F)

71, Canaglia, 2000 (T)

72, Mr. Cold, 2000 (T)

2. Development of a

formalized color-filing

system, and related

routines for faster and

less costly development

of new colors

Externally adapted (Cl.3)

15, Firebird, 1993 (T)

59, Dr. Kleen, 1998∗ (F)

60, Rondo 1998∗ (T)∗Color-related problems

emerged early in these

projects,i.e., Nov.-Dec. 1996

•“Research on color” (1996)

• Development and adoption of

a formal procedure for color

development (“color

development form”; early

1997)

• Development of a formalcolor-filing system (“The

Color Box”; early 1997)

28, Mary Biscuit, 1995 (BS)

42, Fred Worm, 1997 (T)

43, EM01, 1997 (F)

44, Happy spices, 1997 (T)

52, Alibaba, 1998 (BS)

73, Okkio!, 2000 (T)

82, Bunny & Co., 2002 (T)

3. Development of new

versions of earlier,

successful products


35, How much white, 1996 (T)

Externally driven (Cl.4)

5, Caccia, 1990 (BS)

13, “Christy,” 1993 (T)

25, “90043,” 1995 (T)

41, “Placentarius,” 1997 (T)

Project: “New edition of

historical projects” (1998).

74, Colorbavero, 2001 (F)

84, B 9093 2002 (T)

89, Techno t. 1, 2002 (T)

4. Development of ad hoc

work practices to

increase involvement of

a small subset of

promising designers


7, Mix Italia, 1991 (F)

67, Girotondo, 2000 (F)


34, Girotondo, 1996 (T)48, Girotondo, 1997 (T)

58, Girotondo, 1998 (T)

66, G. soaps, 2000 (F)

79, Girotondo, 2001 (F)

• Focused briefing documents

to a small subset of promising

designers.

• Regular meetings with a

small subset of promisingdesigners.

No evidence of systematic

reproduction : reproduction

activities devised by Alessi

management (Column III) are

too flexible and too creative tobe routine. Structure and

regularity are soon dropped and

systematic reproduction is thus

prevented. Rather, projects

resulting from reproduction fall

into the internally adapted

project cluster.

5. Involvement of external

consultants within the

NPD process


2, Shaker Co., 1988 (T)

8, La Cubica, 1991 (T)

No evidence of intentional

selection and retention .

Involvement of external

consultants always resulting

from ad hoc decisions.

Several instances not included in

the sample (e.g., JM13, 2000;

JM20/56, 2003; “Cosmo,” 2003;

“Chiringuito” shaker, 2004).

6. Ad hoc decision to

adopt a target costingprocedure


27, AC04, 1995 (T)


selection and retention


reproduction

7. Decision to follow a

“deductive” product

development process


62, Mami, 1999 (BS)




reproduction

8. Various activities Externally adapted (Cl.3)

78, Kalura, 2001 (T)

86, Babyboop, 2002 (T)




reproduction

aFor each listed product I report the following: product number, e.g., 86 (within the n = 90 sample; see the appendix); product name (e.g.,

“Babyboop vase”); year the product entered Alessi’s catalogue, or its development was discontinued (e.g., 2002); product performance,

according to the appendix footnote (i.e., BS = best seller; T = troops; F= failure; n.a. = discontinued project; Cl= cluster). For the concept

and definition of improvisation, see Miner et al. (2001).


12/26


the product development routine or activity that experi-enced the change-inducing events I observed at Alessi( Description of activities involved ). The second col-umn ( Internal/external adaptations) reports instances of ordinary activities that induced alterations in Alessi’sNPD routine. Such “experiments” were carried out

by Alessi employees at all levels of the organiza-tional hierarchy (Cluster 2, in-house-adapted project sequences), by external agents (Cluster 3, externallyadapted project sequences), or by carrying out externallydirected projects (Cluster 4, externally driven project sequences). Each of these activities implied some devia-tion from standard NPD procedures (Cluster 1, “recom-binant ” project sequences), and thus induced variationin Alessi’s design capability.

The third column ( Intentional selection and develop-ment of adapted traits) reports instances of managerialactivities aimed at improving the emerging alterations,and at creating an overall organizational awareness of

the underlying problems and of ways to solve them.A first instance of intentional managerial interventionsin the evolution of NPD routines was the development of a formal workshop procedure in the early 1990s, aimedat systematically discovering new young talents world-wide. The first observed instance of a workshop (Clus-ter 2, “Hot Bertaa” kettle, Product 6, 1990) resultedin a failure, performing below 5% of expected sales.Project documents show that Alessi had failed to prop-erly channel, within the boundaries of technical feasibil-ity, the imaginative product idea resulting from the Parisworkshop. To strengthen workshop activities, the CEO,in 1991, appointed a workshop coordinator charged with

the explicit task of structuring work with young design-ers by formalizing and reproducing an activity that hadbeen more the outcome of intuition and experimenta-tion. Since then, the workshop coordinator has orga-nized an average of nearly 10 international workshopsper year. The workshop structure—a five-day residentialmeeting—has been gradually formalized and posted onAlessi’s Internet site, together with a schedule of forth-coming workshops, an application form, and notes onvenues, organization, and fees.

A second example of intentional managerial activi-ties is the research on color suggested by Alessi’s topmanagers as a result of numerous stimuli originating

from within and outside the organization. Problems inthe definition of colors for new plastic products eventu-ally prompted some organizational actors to introduce anew method of selecting colors:

We used to work with Pantone a color-filing systembased on removable color chips on paper, but Pantoneis paper based, and it’s a mix of more than one color.So, there’s never perfect conformity between Pantoneand plastic. We often started from a Pantone and theywould go crazy, because they could never get the rightcolor. And we were never satisfied, because we wouldalways run into a bunch of problems In the end,

they developed this Color Box, because they were sodesperate that whatever instrument would have helped.Designer, Interview, March 2003

These suggestions gradually convinced Alessi to developa structured color-filing system (The Color Box) meantto help designers find new colors from an array of colors

already developed by Alessi, thus dramatically reducingthe time and costs involved in developing colors, andincreasing aesthetic coherence among the objects devel-oped by different designers.

Other instances of intentional managerial interventionsreported in Column III of Table 4 relate to initiatives thataim to systematically reproduce early experiments withnew editions of existing Alessi products, or to increasethe involvement of external consultants in NPD pro-cesses successfully tested in earlier projects. Some suchinterventions are later retained by the organization. Oth-ers are abandoned, typically because they imply activi-ties that are too flexible and too creative to be subject to

effective intentional manipulation.In the last column of Table 4, I report observed in-stances of NPD processes resulting from the retention of previous adaptive experiments. Some of these sequencesare the result of the reproduction of alterations (secondcolumn in Table 4), selected and developed by Alessiexecutives (third column) as potentially adaptive mod-ifications of “recipe-book” NPD sequences. Hence, inthe first row (Activity 1, encouraging projects of youngdesigners), the 10 reported projects in the last columnwere developed through versions of the “workshop” rou-tine developed and refined through the managerial activ-ities described in the third column: products developed

in the 1991 “FFF” workshop (Products 14, 16, 18, and19) and the 1993 “Memory Containers” workshop (Prod-ucts 26, 68, 69, 70, 71, 72). Similarly, the seven projectsin the second row feature the use of the color develop-ment procedures that Alessi executives had refined in pre-vious years, as a result of early experiments in the devel-opment of a structured color-filing system. The threeprojects in Row 3 result from following the procedure,refined by managers and built on previous extemporane-ous experiences in developing new versions of old prod-ucts and “historical reproductions.”

A vivid example of managerial intervention aimed atencoding previous mindful experiments, and hence sta-bilizing attention on emerging issues and opportunities,is provided by the development of a formalized color-filing system and related routines for faster and lesscostly development of new colors (Row 2 in Table 4). Theupper part of Figure 1 reports four instances of everyday,ordinary experiments carried out in 1993 by both exter-nal and internal actors (two events of Type 14 in Table 2,“ad hoc modification by external actors,” and two eventsof Type 17, “information gathering on unusual topics”).

The middle part of Figure 1 reports top managementactivities carried out in 1995 (development of a struc-tured color-filing system mainly consisting of guidelines


13/26


Figure 1 Enhancing Product Development Capabilities by Leveraging Mindful Acts: A Simplified Example from Available

Product-Sequence Data

10 1 23 3 11 11 13 13 18 4 14 14 14 18 1 17 14 8 14 11 11 4 12 18 6 17 11 6 6 5 14 14 7 16

Cluster 3. “Externally adapted” Sequence 36: “Firebird” electronic gaslighter (1993)

Cluster 5. “Recombinant” project sequences

Top management activities (1995–1999):

• Research on color

• Training internal staff on color

• Development and adoption of a formal procedure for color development

(“color development form”)

• Developing a tailor-made filing system for colors

• Organizing systematic production of plastic samples with suppliers

Event Type 14: “Ad hoc modifications by external actor”:

• “ Dear C . [ Alessi engineer ], I sent you ‘ fluo’ colors for a new firebird proposal … is it possible to see a plastic sample of them? If we pursued these apparently insignificant

experiments, when we then need a new color it is already formulated and all, isn’t it ?"

[ December 1993, Designer to Alessi].

• Experimental development of new-color plastic samples by R.Ltd., Alessi color supplier.

Consistent use of systematic color-filing system (after 1995) and of “The Color Box” (after 1999) in projects requiring new color definition:

• 42, Fred Worm, 1997

• 43, EM01, 1997

• 44, Happy spices, 1997

• 52, Alibaba, 1998

• 73, Okkio!, 2000

• 82, Bunny & C., 2002

Development of systematic color-filing system (1995)

Development of “The Color Box” (1999)

Event Type 17: “ Information gathering on unusual topics”:

• Data on existing product colors used by a competitor are collected and analyzed to experimentcolor replicability

• Data on performance and price of spectrophotometers are collected by NPD staff member to explorethe benefits of more “scientific” color development process (in particular, to improve consistency

between prototype and end-product colors)

12

Everyday, ordinary

experiments

(1993)

Subsequent higher-level

managerial refinement and

codification

(1995–1999)

Subsequent

reproduction of

adapted routine

(1997 onward)

to NPD team members and the gradual collection of colored plastic samples from suppliers), which culmi-nated in the development of The Color Box in 1999.These activities were aimed at refining and codifyingprevious experiments resulting from everyday practice.

Finally, the lower part reports examples of new prod-ucts developed by enacting the adapted routine arisingfrom previous experiments and subsequent manage-rial interventions. As Rows 1 to 3 in Table 4 show,observed adapted routines (Column 4) were alwaysimplemented after managerial refinement and codifica-tion (Column 3), which in turn have always followed in time experiments carried out by means of everydayactivities (Column 2).

Strategic renewal encompasses an investigation of out-comes and performance, besides process and content(Agarwal and Helfat 2009). Data in Table 3 reveal how

the evolution of capabilities described in this sectiongenerated different patterns of heterogeneity and per-formance within and across clusters. Cluster 1 “recipe-book” project sequences are characterized by relativehomogeneity, as the medoid sequence (Product 49, the

soap dish “Titan,” designed by Marc Newson) accountsfor 23.4% of sequence dissimilarity within the cluster(Column 6, “Percentage within cluster”). These fairly“standard” sequences are also comparatively homoge-neous when contrasted with the whole sample of 90 NPDprocesses, as they account for 30.2% of the total variance(Column 7, “Percentage among clusters”). “Mutations”introduced by internal and external agents amplify pro-cess heterogeneity. Experiments carried out by externalagents (Clusters 3 and 4), in particular, result in fairlylower levels of both within-cluster homogeneity (19.3%and 7.7% in Clusters 3 and 4, respectively) and, even


14/26


more markedly, among-cluster homogeneity (8.1% and4.6%, respectively). Mutations introduced by internalagents (Cluster 2) are also a source of significant hetero-geneity (9.2% within and 11.1% among clusters). On theopposite, Cluster 5 “recombinant” sequences display thehighest levels of both internal (32.3% within cluster)

and across-cluster consistency (53.6% among clusters).Apparently, mindful managerial interventions aimed atencoding and reproducing Clusters 2–4 experiments gen-erated relatively automatic, uneventful sequences.

Equally striking is the impact different heterogeneitylevels seem to have on product performance. As resultingfrom Table 3 data, the stable “recipe-book” routines inCluster 1 yielded positive product outcomes. Four of 36products exceeded volume sales aspirations by more than200%, and were hence classified as best sellers by Alessi(Column 8, “Percentage of best sellers”). On average,actual volume sales of these 36 products exceeded aspira-tions by 68% (Column 9, “Average sales performance”).

In contrast, variations by internal and external agents inClusters 2–4 processes apparently had a negative impacton product performance. Although the proportion of bestsellers resulting from these mutated NPD processes is inline with “recipe-book” routines, if not higher, rangingbetween 8.3% in Cluster 2 and 23.1% in Cluster 4, aver-age product performance is significantly lower. Muta-tions by Alessi actors (Cluster 2) yield products whoseaverage performance relative to aspirations (+6%) is sig-nificantly lower than in Cluster 1, though still positive.Mutations by external actors (Clusters 3 and 4) result ineven more marked decline, as average product perfor-mance is negative, hence signaling a decreased capacity

of these altered routines to yield expected results. In con-trast, Cluster 5 “recombinant” products are the best per-formers within the sample. One of four was classified as abest seller—the highest percentage of best sellers amongthe five clusters. More important, average product per-formance (+127%) is almost double the performance of “recipe-book” sequences. Although product performancein the design industry is driven by several, often elu-sive factors (Bhaskaran 2005, Collins 1999), my datasuggest that a significant relationship may exist betweenroutine/homogeneity and variation/heterogeneity on oneside, and product performance on the other. An inter-pretation of this relationship, and of its drivers and

outcomes, is discussed in the next section. Equallyinteresting is the observation that routine/homogeneousprocesses (Cluster 5) resulting from the encoding of variance-enhancing mutations (Clusters 2–4) performsignificantly better than “baseline” routines (Cluster 1).

Discussion and ImplicationsThis paper explores evolutionary patterns of organi-zational capabilities within the context of organiza-tional renewal. Focus is on the role day-to-day activitiesplay in shaping event sequences representing different

enactments of a capability over time. The aim is tounderstand the extent to which the evolution of capabil-ities is shaped by ordinary interactions, and the extentto which such evolutionary patterns can be intentionallydirected to increase the adaptive chances of change pro-cesses. As noted at the outset, this issue is central in

recent conceptualizations of capabilities-based strategicrenewal (Gavetti 2005, Helfat and Peteraf 2003, Winter2003, Zollo and Winter 2002), but it has rarely benefitedfrom accurate empirical treatment.

There are three key findings, synthetically illustratedin Figure 2. First, established capabilities, which func-tion as semiautomatic, less-mindful entities (left ver-tical box in Figure 2, “level n” capability), are sys-tematically reshaped by mindful ordinary acts carriedout by individuals within and around the organization(two horizontal boxes on the right of Figure 2). Theseintentional experiments recombine the existing capabili-ties repertoire on a day-to-day basis, hence resulting in

increased process heterogeneity and temporary perfor-mance decline. Second, timely managerial interventionsencode successful experiments into higher-level orga-nizational capabilities (right vertical box in Figure 2,“level n+1” capability). The resulting adapted capa-bility incorporates previous mindful experiments intoa homogeneous, semiautomatic entity. Third, whereas“mutated” processes resulting from mindful experimentstemporarily underperform the original “level n” capabil-ity, “n+ 1 level” capabilities resulting from the encod-ing of such heterogeneous experiments display higherprocess homogeneity and a permanent increase in per-formance. Apparently, insights from previous experi-

ments encoded in higher-level capabilities increase thequality of organizational attention. Hence, adapted capa-bilities integrate mindful improvisation by internal andexternal agents and thus thrive on heterogeneity. Theseinsights extend prior work on capabilities and organiza-tional renewal. They suggest future research directionson how the concept of mindfulness can be used to unveilthe microfoundations of capabilities evolution and orga-nizational advantage.

Mindfulness in Action: Ordinary Activities andthe Evolution of CapabilitiesMy findings suggest that micro, ordinary activities car-

ried out by individuals within and around the organiza-tion and at all levels in the organizational hierarchy arecentral to determining the idiosyncratic content of capa-bilities and their dynamic adaptation over time. Gettingcloser to the subtleties of Alessi’s practices, as I triedto do in my study of 90 NPD processes, one is eventempted to conclude that not only are individuals andordinary activities important in conferring capabilitiestheir uniqueness, but that they are in some ways the onlyrelevant things. My data reveal that the rich repertoireof existing product development routines at Alessi has


15/26


Figure 2 Interplay of Mindful and Routine Behavior in the Evolution of Capabilities

Agents

External

Internal

Behavior

Less mindful

Semiautomatic

NPD capability - level n

(Cluster 1)

Mindfulness in action

Improvisation by INTERNAL agents

Heterogeneity: higher than capability n

Performance: lower than capability n

(Cluster 2)

Mindfulness in actionImprovisation by EXTERNAL agents

Heterogeneity: higher than capability n

Performance: lower than capability n

(Cluster 3– 4)

NPD capability - level n+1

(Cluster 5)

Heterogeneity: lower than n

Performance: higher than n

a

a

b

b

Mindful

intentional

Notes. (a) Improvisation: Availability of a stable repertoire of action (level n) forms the basis for mindful enactment of routines, which results

in experimentation (“mindful acts”). (b) Encoding: Mindful retention and reproduction of improvisational acts (level n + 1).

been constantly adapted through mindful improvisations

by internal and external agents, resulting in increasinglyrich and heterogeneous processes.

New product development processes in Clusters 2 (in-house adapted), 3 (externally adapted), and 4 (externallydriven) exhibit clear mutations of the “baseline” NPDroutine (Cluster 1 “recipe-book” NPD processes). Thesemutations result from highly local attempts at respond-ing to the demands and contingencies of the present. Thecurrent sophisticated workshop format—which allowsAlessi to systematically bring in young, promisingdesigners—stems from an early test carried out by CEOAlberto Alessi and Alessandro Mendini and organizedby Alessi in Paris in November 1986. The ingenious

“Color Box”—currently allowing Alessi to quickly andefficiently develop aesthetically homogeneous coloredobjects—was first suggested by designer Venturini in afax sent to Alessi’s NPD staff on December 6, 2003. Thedecision to reverse the traditional product developmentsequence by starting with conceptual research on form—which yielded the blockbuster “Mami” series in 1999—was made by designer Giovannoni when the CEO askedhim to develop “yet another series of pots and pans.”These examples, and others reported in Table 4, areall instances of experiments aimed at improving the

smoothly functioning NPD routine by mindfully adapt-

ing it to the conditions at hand.Although subsequent activities—discussed in the next

section—through which top management encoded someof these early experiments can be interpreted as higher-level, conscious strategic initiatives, these initial trig-gers of capability change have the nature of ordinary,though mindful, activities carried out by various indi-viduals at all levels within and outside the firm. Theseactivities depart from traditional descriptions of con-scious strategic initiatives because they are improvisa-tional in nature, highly local, and they involve thinkingin real time—a state of active awareness simultaneouswith the execution of action (Miner et al. 2001), defined

by Levinthal and Rerup (2006, p. 505) as “mindfulnessin action.” Each time Alessi product development capa-bility is enacted into the design of an actual product,individuals make apparently mundane adjustments thatcan have durable effects.

Mindful experimentation by internal and, moreintensely, external agents results in high levels of within-and between-cluster heterogeneity. As Figure 2 suggests,the semiautomatic or less-mindful repertoires of actionunderlying Cluster 1 NPD processes form the basis formindful experimentation in Clusters 2–4 projects. Estab-


16/26


lished product development routines provide a repertoireof initiatives allowing internal and external actors torespond rapidly to stimuli and to engage in a wide setof possible experiments. As Levinthal and Rerup (2006,p. 505) suggest, “mindfulness in action is local and sit-uated and involves spontaneous recombination of wis-

dom accumulated from prior experimental learning Recombination of well-rehearsed routines can thereforebe seen as an important activity that makes mindfulnessin action possible.”

These efforts of internal and external agents are mind-ful, because their aim is to obtain local improvements inthe existing repertoire of action dictated by the conditionsat hand. Alessi employees and external collaborators sys-tematically search for process improvements that mayyield more promising or apt product development pro-cesses. My study reports several instances of this mindfulconversion of ambiguous stimuli into creative responsesto difficulties or opportunities. All adaptations listed in

Table 4, Column II are as many instances of special casesthat had to be fitted by internal and external agents toAlessi’s given repertoire of NPD routines through mind-ful conversions. These creative adaptations testify boththe richness and flexibility of Alessi’s baseline routines,and the effortful, nonautomatic enactment of such rou-tines by agents within and around the organization. Thisway, routines are constantly modified and adjusted toaccommodate unexpected contingencies and to captureunanticipated opportunities (Chirico and Salvato 2008).

Insights emerging from Alessi ground and advanceconceptual literature on the interplay between mind-ful and less-mindful behavior, and its role in shaping

organizational capabilities. Levinthal and Rerup (2006)identified four key interrelationships between mindfuland less-mindful processes: (1) the richness of exist-ing “less-mindful” routines enhances the richness of organizational mindfulness by providing a reliable basefor recombination and improvisation; (2) “less-mindful”processes of monitoring sustain mindfulness across timeby providing reliable routines to systematically scan theenvironment; (3) mindfulness allows the flexible andadaptive execution of routinized “less-mindful” behaviorby interpreting the context before selecting the appropri-ate routine to be performed; and (4) mindfulness enablesprocesses of reinforcement learning, which allow the

encoding of ambiguous outcomes into increasingly adap-tive “less-mindful” routines.

Evidence of “mindfulness in action” discussed in thissection points to the existence of a relentless process of mindful adaptation of organizational capabilities result-ing from the recursive enactment of interrelationships (1)and (3). Although the repertoire of existing routinesprovides the building blocks for flexible recombination,flexible responses to a changing environment resultin adaptations of the baseline routines. Mindful andless-mindful processes coexist in shaping capabilities

evolution. Predominantly mindful processes of

om article 9

Documents