LEARNING SEQUENCES: THEIR EXISTENCE, EFFECT, AND …jasond/learningsequences.pdfscribed for the unit. Experimental learning is another direct learning process. Experimental learning

LEARNING SEQUENCES: THEIR EXISTENCE,EFFECT, AND EVOLUTION

CHRISTOPHER B. BINGHAMUniversity of North Carolina at Chapel Hill

JASON P. DAVISMassachusetts Institute of Technology

Much is known about the importance of learning and some of the distinct learningprocesses that organizations use (e.g., trial-and-error learning, vicarious learning,experimental learning, and improvisational learning). Yet surprisingly little is knownabout whether these processes combine over time in ordered ways, because mostresearch on learning explores one particular process. Using theory elaboration andtheory-building methods and data on the accumulated country entries of entrepreneur-ial firms, we address this gap. Our core contribution is an emergent theoreticalframework that develops the concept of learning sequences. We find that learningsequences exist and are influenced by initial conditions. We also find that learningsequences evolve in fundamentally distinct ways over time and with repeated use.Finally, data show how different learning sequences differentially affect both shorter-and longer-term performance, suggesting that it matters which learning processes areused and when. Overall, our findings on learning sequences have important implica-tions for learning theory, international entrepreneurship, and the growing literatureon process management.

Organizational learning is of fundamental inter-est in organizational theory and strategy. Somestudies have shown that firms learn to diversifyinto new countries (Barkema, Bell, & Pennings,1996; Zahra, Ireland, & Hitt, 2000) and product-markets (Miner, Bassoff, & Moorman, 2001) to cap-ture scale and scope economies. Other empiricalwork has shown that firms learn to expand theiroperations through acquisitions (Haleblian & Fin-kelstein, 1999; Hayward, 2002) and alliances(Anand & Khanna, 2000; Hoang & Rothaermel,2005) to create corporate value. Still other researchhas revealed that firms learn to disseminate knowl-edge (Darr, Argote, & Epple, 1995), augmentthroughput (Lieberman, 1984), and reduce defects(Levin, 2000) to improve pricing and productivity.Indeed, research suggests that organizational learn-

ing is a central means by which firms generateinnovations, adapt to environments, take advantageof emergent market opportunities, and create com-petitive advantage (Argote, 1999).

However, despite the importance of organiza-tional learning, empirical research generally ex-plores one particular learning process (e.g., “trial-and-error learning,” “vicarious learning,”“experimental learning,” and “improvisationallearning”) while underexploring the question ofwhether different learning processes get used to-gether in sequence. For example, much work exam-ines direct learning—that is, a firm’s learning fromits own experience (Schwab, 2007)—in particulartrial-and-error learning (Baum & Dahlin, 2007;Tsang, 2002; Van de Ven & Polley, 1992). Studies inthis stream suggest that learning occurs as organi-zations change their subsequent behavior in re-sponse to prior performance outcomes. To illus-trate, in his study on the internationalization ofJapanese electronics firms, Chang (1995) suggestedthat firms learn how to enter new countries bydrawing on performance outcomes. Executivesstarted with an initial investment in a foreign coun-try such as the United States. If results from theinitial investment proved positive, they expandedinvestment in the same country; but if results werenot positive, they did not. Other work on directlearning examines experimental learning or impro-

We appreciate the generous support of the NationalScience Foundation (IOC Award #0323176) and the Ke-nan-Flagler Business School at UNC. We also thank mul-tiple individuals for their very helpful comments, in-cluding Kathy Eisenhardt, Harry Sapienza, and MikeRoach; seminar participants at Brown University, theUniversity of North Carolina, and the University of Mary-land; our anonymous reviewers; and our associate editor,Jim Combs.

Editor’s note: The manuscript for this article was ac-cepted for publication during the term of AMJ’s previouseditor-in-chief, R. Duane Ireland.

� Academy of Management Journal2012, Vol. 55, No. 3, 611–641.http://dx.doi.org/10.5465/amj.2009.0331

611

Copyright of the Academy of Management, all rights reserved. Contents may not be copied, emailed, posted to a listserv, or otherwise transmitted without the copyright holder’s expresswritten permission. Users may print, download, or email articles for individual use only.

visational learning. Scholars argue that through ex-perimental learning organizations gain knowledgeand insight through deliberate, small-scale tests,generally conducted “off-line” (i.e., in controlled,nonmarket settings) that are explicitly designed tohelp managers better prepare for the future (Pisano,1994; Thomke, 2003). By contrast, through impro-visational learning organizations learn in real timeas design and action converge to solve emergentproblems and take advantage of surprising oppor-tunities (Miner et al., 2001). Finally, some scholarsfasten their attention on indirect learning—that is,learning from others’ experience (Ingram, 2002).Work in this stream generally focuses on vicariouslearning, which occurs as firms observe actions byother firms and then change their own behavior orbeliefs as a result (Haunschild & Miner, 1997).Through vicarious learning firms thus gain the ben-efits of accumulated knowledge while avoiding theexpense of accumulated experience (Kim & Miner,2007; Srinivasan, Haunschild, & Grewal, 2007).Empirical studies have shown the relevance of vi-carious learning in a variety of settings, such asmarket entry (Greve, 1998), investment bankerchoices (Haunschild & Miner, 1997), hotel chainlocation decisions (Baum & Haveman, 1997), nurs-ing home acquisitions (Baum, Li, & Usher, 2000),and product introductions (Srinivasan et al., 2007).

Although it is well known how each of these oflearning processes is used alone, what is not knownis whether firms use them together over time inordered ways. Some research has attempted to ex-plore interactions. Scholars have explored directlearning and indirect learning (Schwab, 2007),showing that both occur concurrently in organiza-tions (Baum & Dahlin, 2007) or partially consider-ing interactions (Chuang & Baum, 2003; Shaver,Mitchell, & Yeung, 1997). Missing from this body ofresearch, however, is empirically grounded under-standing about if a temporal order exists in thelearning processes firms use and if this matters. Weexplore this gap.

This article is organized around three researchquestions: (1) Do learning sequences exist? (2) Dolearning sequences matter? and (3) Do learning se-quences evolve over time? In keeping with other pro-cess-based organizational research (Abbott, 1990,1995; Langley, 1999; Van de Ven & Poole, 1995), wedefine a learning sequence as an ordered use of learn-ing processes. Given the state of extant theory, we usetheory-building (Eisenhardt, 1989) and theory elabo-ration methods (Lee, 1999). The setting is the inter-nationalization of nine entrepreneurial firms withheadquarters in Singapore, the U.S., and Finland.

THEORETICAL BACKGROUND

Research from the trial-and-error learning, exper-imental learning, improvisational learning, and vi-carious learning streams is pertinent to our re-search questions. We focus on these specific directand indirect learning processes since the literaturesuggests their particular importance and preva-lence (Argyris & Schön, 1978; Baum & Dahlin,2007; Huber, 1991; Leavitt & March, 1988; Miner etal., 2001; Srinivasan et al., 2007). Following previ-ous research, we define learning as a regular shift inbehavior or knowledge informed by prior action(Argote, 1999; Cyert & March, 1963; Levitt & March,1988; Miner et al., 2001). This definition incorpo-rates both behavioral learning models, which stresschange in action, and cognitive learning models,which stress change in ideas.

One common direct learning process discussedin the literature, trial-and-error learning, is definedas the process by which firm executives undertakea course of action, and the consequences of thatcompleted action lead to change in the firm’s actionor knowledge base (Argyris & Schön, 1978; Baum &Dahlin, 2007; Greve, 2003). An important charac-teristic therefore is that trial-and-error learning oc-curs after a firm experiences the consequences ofan action and changes its behavior or bases infer-ences on that completed action. Managers repeatseemingly successful organizational actions, reflecton the outcomes, and then revise understandingsand/or actions as needed (Haunschild & Sullivan,2002). As one illustration, Van de Ven and Polley(1992) described trial-and-error learning in onefirm during the development of a biomedical inno-vation. The authors found that when the prior ac-tions of entrepreneurs were deemed successful,more resources were devoted to the innovationunit’s pursuit of that same course of action. Also inkeeping with trial-and-error learning, the authorsobserved that when the actions of entrepreneurswere deemed unsuccessful, resource controllers in-tervened, and new courses of actions were pre-scribed for the unit.

Experimental learning is another direct learningprocess. Experimental learning takes place in con-trolled situations that organizations use to testcausal propositions and create new knowledge(Cook & Campbell, 1979). Because the central pur-pose of learning through experimentation is to ac-quire new knowledge of relationships, post hocreflection on outcomes is high (Miner et al., 2001).Scholars have argued that organizations deliber-ately vary inputs off-line in comparative contexts(e.g., assessing the functionality of a product withdifferent technical features integrated) and then

612 JuneAcademy of Management Journal

closely monitor outcomes to correctly attribute out-comes to inputs (Thomke, 2003). Scholars havefurther argued that this off-line nature of experi-mental learning sets it apart from other direct learn-ing processes such as trial and error (Miner et al.,2001). Thus, in experimental learning, variation inconditions is planned and intentionally introducedto produce insights about input-output relations.The literature also suggests that experimentallearning often relies on two key characteristics.First, it involves relatively low-cost initiatives (e.g.,use of economical, easy-to-modify prototypes orinexpensive focus groups to test different productsizes, colors, or packaging materials) that help yieldmore robust designs and solutions and direct atten-tion to potential downstream risks (Brown & Eisen-hardt, 1997; Thomke, 2003). Second, the newknowledge derived from experimental learning, ifuseful, can quickly be incorporated into firm activ-ities (Pisano, 1994). Hence, because experimentallearning often involves low-cost initiatives, firmscan use a variety of them to learn without the fearof suffering crippling mistakes or financialovercommitment.

A third direct learning process is improvisationallearning, defined as a real-time learning process inwhich firms learn to solve unexpected problems orcapture surprising opportunities in the moment(Miner et al., 2001). Real-time learning influencesnovel action at the same time that the action istaking place (Miner & Moorman, 1998; Weick,1998). This emphasis on learning in real time asdesign and action converge sets improvisationallearning apart from experimental learning and trial-and-error learning. Hence, whereas deliberate for-mation of contrasting situations during experimen-tal learning results in the creation of new,generalizable knowledge, solving a surprising prob-lem during improvisation results in knowledge id-iosyncratic to a particular time or place. Improvi-sational learning is also distinct from trial-and-error learning, where prior experience plays a keyrole in changes to action or cognition. With trialand error, learning occurs only after consequencesof past actions occur. Actions and their outcomesinform subsequent action or cognition (Miner et al.,2001). In contrast, with improvisational learning,managers do not wait for the consequences of pastactions. Changes in action or cognition are made“on the fly,” as planning and doing occur simulta-neously. Yet, as firms often retain and repeat suc-cessful activities discovered after an improvisedoutcome, improvisational learning may representthe first step in longer-term trial-and-error learning(Miner et al., 2001).

A common indirect learning process—that is, aprocess of learning from others’ experience ratherthan firsthand—is vicarious learning (Huber, 1991;Kalnins, Swaminathan, & Mitchell, 2006; Levitt &March, 1988; Srinivasan et al., 2007), which gener-ally occurs when firms alter their behaviors or cog-nition in response to the actions of competitors(Kim & Miner, 2007). Through observation, deci-sion makers gather information about the charac-teristics and outcomes of competitors. The frequentresult is imitation of seemingly successful practices(Denrell, 2003). Some research suggests that vicar-ious learning may be an important initial learningprocess. Faced with insufficient information forlearning from their own experience, organizationscan rely on others’ experiences to cover their deficitin understanding (Baum et al., 2000; Henisz & De-lios, 2001; Kraatz, 1998). Research shows that vi-carious learning is particularly valuable in newindustries and when uncertainty is high. For exam-ple, to explore how vicarious learning takes placeas firms introduce new products in nascent mar-kets, Srinivasan et al. (2007) examined product in-troductions of 67 firms in the U.S. digital cameramarket and found that changes in a focal firm’s rateof new product introductions were influenced bychanges in new product introductions of other sim-ilarly sized and successful firms. Yet other researchsuggests that vicarious learning may not be a goodinitial learning process because new and/or inex-perienced firms lack the “absorptive capacity” tolearn from others, so that even if they are able togain knowledge, they may not be able to internalizeand leverage it fully (Henisz & Delios, 2001; Zahra& George, 2002).

In summary, the extant research on organiza-tional leaning has generally focused on the rele-vance of discrete direct learning processes (e.g.,trial-and-error learning, experimental learning, orimprovisational learning) and indirect learningprocesses (e.g., vicarious learning). Like other or-ganizational process research, this research hasgenerally addressed how learning using a discreteprocess occurs over time.1 But although research onorganizational learning provides much understand-ing about how firms use each of these discretelearning processes alone, it has provided little un-derstanding about whether firms use multiple

1 Process research generally focuses on understandingthe temporal dynamics of organizational phenomenasuch as learning (Van de Ven, 1992; Langley, 2007). Forexample, as noted earlier, research on trial-and-errorlearning describes how firms engage in an action andthen the consequences of that action influence subse-quent action (Van de Ven & Polley, 1992).

2012 613Bingham and Davis

learning processes together in ordered waysover time.

Some empirical researchers have attempted toaddress this gap by exploring whether direct andindirect learning get used together (Baum & Ingram,1998; Henisz & Delios, 2001; Schwab, 2007). Baumet al. (2000) examined acquisitions made by chainnursing homes in Ontario from 1971 through 1996and found that a chain was more likely to acquire atarget the more similar the target was to all the focalchain’s current components and the nearer the tar-get was to other, similar, chains’ current compo-nents. Likewise, Schwab (2007) indicated that re-lying on both direct and indirect learning leads to asubstitutional interaction in which knowledge thatis consistent from both sources exhibits a weakereffect than the linear addition of their independenteffects. Yet, although these studies are important inthat they suggest that firms appear to use bothdirect and indirect learning and that each may in-fluence the other, future research opportunities re-main, since it is unclear whether direct or indirectlearning occurs first and whether sequence matters.Moreover, prior empirical research on learning sug-gests that direct learning can be unpacked into a setof distinct processes (i.e., experimental, trial-and-error, and improvisational learning) and that eachmay distinctly influence how learning takes placeover time (Miner et al., 2001). As a push in thisdirection, Miner and colleagues explored improvi-sational learning and how this process contrastswith experimental and trial-and-error learning. Butbecause the focus of this study was improvisationallearning, the authors did not directly explore howit might be used in connection with experimentallearning or trial-and-error learning, except to con-jecture that improvisational learning may drive outexperimental learning or serve as an episode inlonger-term trial-and-error learning. Further, theauthors did not discuss whether vicarious learn-ing temporally links with experimental learning,trial-and-error learning, and/or improvisationallearning.

Overall, much is known about how firms usespecific learning processes. However, little isknown about whether firms use multiple learningprocesses in temporally ordered ways. In otherwords, understanding of learning sequences re-mains extremely limited. This gap is critical. Froma practical perspective, if a particular order oflearning processes leads to better performance out-comes than another order, there are immediate ap-plications for managers. From a theoretical per-spective, not understanding about learningsequences in process research on learning is prob-lematic, since the concept of sequences is central in

much organizational process research (Adair &Brett, 2005; Amis, Slack, & Hinings, 2004; Burgel-man, 1994, 1996; Langley, 1989) and is explicitlyhighlighted when scholars build theory from pro-cess data (Langley, 1999, 2007; Rindova, Ferrier, &Wiltbank, 2010; Van De Ven, 1992).

Our core contribution of the present study ishelping to establish sequences as a meaningful con-cept and focus in process research on learning.First, we uncover the existence of distinct learningsequences. Second, we reveal how they evolve indistinct ways with repeated use. Finally, we showhow different learning sequences differentially af-fect performance, both in the shorter and longerterm, thereby implying that it matters which learn-ing processes organizations use and when they usethem. Beyond contributing to the field of organiza-tional learning, these findings have implicationsfor international entrepreneurship and the growingliterature on process management.

METHODS

The research setting was nine entrepreneurialcorporations in the global information technology(IT) industry. The IT industry was attractive for thisstudy because its high rate of change suggests theneed for learning (Davis & Eisenhardt, 2011). Wechose entrepreneurial firms because their smallsize simplifies the observation of learning pro-cesses. In addition, studying entrepreneurial firmsminimizes “left censoring” of data because thefirms can be tracked from inception.

We focus on how entrepreneurial firms learnduring internationalization (i.e., new country entry[Root, 1994]). Internationalization is a useful con-text in which to explore learning. First, countryentries are discrete and easily detected events thatcan be analyzed in isolation or as part of a larger setof experiences. Second, during internationaliza-tion, firms enter countries that may or may not besimilar to previously entered countries. This sug-gests variance in the effects of prior experience andthe degree to which a focal firm may or may notrely on learning from other firms. Third, data foreach country entry can be isolated, enabling bothsingle and multiple country analysis.

We studied nine firms with headquarters in eachof three culturally distinct (Hofstede, 1980) regions:Finland, the U.S., and Singapore. Studying multi-ple regions enhances the relevance and generaliz-ability of results.2 All sample firms were eight years

2 We focused on these regions for several reasons.First, the mutual cultural distinctiveness of Singapore,


old or younger at the time of data collection andhad made all their country entries within the fouryears prior to data collection. This relative recencyshould enhance accurate recall of events. More-over, each sample firm had entered at least fourcountries. This number of entries ensures sufficientexperience from which to examine if and howlearning processes get used together over time. Ta-ble 1 summarizes information on our sample.

We relied on four data sources: (1) quantitativeand qualitative data from semi-structured inter-views with company leaders, (2) e-mails, observa-tions, and phone calls made to follow up on inter-views and to track internationalization over time,(3) quantitative data on companies’ international-ization performance, by country, from companyand public sources, and (4) archival data, includingcompany websites, business publications, andother materials produced inside the firms.

The primary data source was the semi-structuredinterviews. We conducted two types of interviewscorresponding to two types of informants. The firsttype, labeled “HQ-level,” were interviews with in-dividuals such as a firm’s CEO, founder, COO(chief operating officer), and others responsible forfirmwide activities; the second type, labeled “coun-try-level,” were interviews with “country manag-ers” and “country team members” who were in-volved with entry into a particular country. Thedata comprised over 50 interviews on three differ-ent continents with both the multicountry viewfrom the corporate perspective and specific detailfrom individual countries (see Table 2 for moreinformation about informants).

Each interview consisted of three main parts: (1)firm background information, (2) event chronologyfor a specific country entry (country-level inter-view) or for several entries (HQ-level interview),and (3) direct questions related to learning pro-cesses. For the event chronology, we asked open-

ended questions that focused on the stream ofcountry entry events (e.g., How did your companygain its first sale? How did you move to your sec-ond sale?), and avoided broad speculation that wasnot grounded in specific events. We then reviewedthe chronology, and asked if we had covered all keyevents.

We concluded the interview with direct ques-tions related to learning, such as “What, if any,were the lessons gained during this country entry?”and “What, if any, lessons from other country en-tries were used in this country entry?” The tech-nique of asking different questions (i.e., nondirec-tive and directive) provides a stronger grounding oftheoretical insights, mitigates bias (Eisenhardt,1989; Yin, 1994), and is consistent with theoryelaboration (Lee, 1999) and theory building (Eisen-hardt, 1989). We also sent follow-up e-mails, addedextra interviews as needed, and triangulated inter-view data with observations and archival data toimprove accuracy and completeness (Jick, 1979).

Informant bias was an important consideration.We addressed this issue in several ways. First, wecombined real-time and retrospective data. Such acombination is valuable, since the retrospectivedata enable efficient collection of many observa-tions (for good grounding), and real-time datadeepen understanding about the order of events(Leonard-Barton, 1990). Second, previous researchsuggests that our interview techniques (e.g., “court-room” questioning, event tracking, nondirectivequestioning, establishing a “back in time” cognitiveframe) typically yield accurate information conver-gent among informants and with archival data(Brown & Eisenhardt, 1997). The few differencesthat may arise primarily result from informants’relating different parts of the story, not from inter-informant conflicts. Third, reliance on informantsat multiple levels of hierarchy helps yield a morecomplete and thus, more accurate, picture of eventsthrough complementary perspectives and granular-ity. Combining qualitative stories with quantitativemeasures has similar effects. We also relied on in-formants who were particularly knowledgeableabout the relevant events surrounding internation-alization and for whom internationalization wasquite important, thus improving memory accuracy.Fourth, ensuring anonymity for both companiesand informants encourages candor.3 Finally, we

the U.S., and Finland (Hofstede, 2001) allowed us to havemore generalizable findings that apply beyond a partic-ular set of firms, such as those coming from a largemarket such as the United States. Second, Singapore, theU.S., and Finland have clusters of technology-basedfirms, which allowed us to find and compare regionalpatterns for similar firms. Thus, we have Singaporean,Finnish, and U.S.-based entrepreneurial firms focused onsoftware (Table 1). Third, the choice of Singapore, U.S.,and Finnish entrepreneurial firms is important from amethodological point of view, as in each country man-agers speak English fluently. This language uniformityimproves candor, depth of comments, and mitigates lossof data that may occur through translation and back-translation of interview data.

3 Firms are disguised with pseudonyms drawn fromearly national presidents of the respective home countries:“Jackson,” “Polk,” “Tyler,” and “Adams” (U.S.); “Kallio,”“Stahlberg,” and “Ryti” (Finland); “Wee” and “Nair” (Sin-gapore). See Table 1 for information on our sample firms.


supplemented interview data with archival infor-mation from each time period in question.

We began data analysis by writing individualcase histories (Eisenhardt, 1989; Eisenhardt &Graebner, 2007), synthesizing the interview and

archival data from each focal firm into a chronolog-ical story of internationalization and how the firmlearned in each country entered. We used thesehistories for two types of analysis: within-case andcross-case. Within-case analysis concentrated on

TABLE 1Description of Firmsa

Firmand HQLocation Product

Sales andEmployeesb

YearFounded

First FourCountryEntries

(in Order)

CulturalDistance

to HQEntryMode

Importance of CountryEntryc Additional Data

WeeSingapore

IT securitymonitoring

$3.2M 100 2000 Hong KongMalaysiaJapanChina

0.280.855.150.47

2223

“International expansionis key in ourexpansion plan.”

42 archival documentsOne day of on-site meetings/

observations at corporateHQ in Singapore

Follow-up interview withCEO

JacksonU.S.

Wirelesschips formobiledevices

$1M 100 1999 ChinaTaiwanKoreaJapan

3.012.803.392.63

5511

“We were founded fromthe beginning to be amulti-siteinternationalcompany.”

6 archival documentsTwo days of on-site

meetings/observations inU.S. HQ

Follow-up interview withdirector of marketing

RytiFinland

Clinical datacapturesolutions

$9.3M 75 2000 SwedenU.S.CzechRepublicGermany

0.741.371.131.21

5555

“It’s vital, it’s thecornerstone. It’s theonly way to goforward.”

One day on site in FinlandHQ

Discussions with Finnishprofessor advising TMT

11 archival documentsFollow-up interview with

cofounderKallioFinland

Wirelesssolutions

$1.5M 35 1999 ItalySwitzerlandIrelandU.K.

1.761.441.581.70

3334

“There’s no choice tonot be international.”

One day on site in FinlandHQ

Discussions with localadvisor

Business case on firmFollow-up interview with

cofounderAdamsU.S.

Real timeanalytics(supplychain,CRMs)

$8.5M 65 1996 AustraliaU.K.FranceGermany

0.020.081.540.41

1511

“Ultimately you need tobecome global.”

10 archival documentsOne day of on-site with

meetings/observations inU.S. HQ

Follow-up interview withformer chairman

StahlbergFinland

Securitysoftwaresolutions

$13.9M 104 1996 SwedenGermanyU.S.Japan

0.741.211.374.34

1555

“From the beginning, ahigh degree of visionand concept arounddoing what was rightfor the global market.”

Discussions with boardmember

Press releases and industryreports

One day (each) of on-sitemeetings in U.S. andFinland

Follow-up interview withU.S. CEO

PolkU.S.

Securitysoftwaresolutions

$11M 100 1996 ChinaGermanySwitzerlandU.K.

3.010.410.340.08

5115

“We have a globalcustomer so just defacto, we have to beinternational.”

20 archival documentsOne day of on-site meetings/

observations at corporateHQ in U.S.Follow-up interview with

CEOTylerU.S.

Clinical datacapturesolutions

$70M 192 1994 SwedenNetherlandsGermanyJapan

2.731.770.412.63

1111

“It is a global problemthat we are trying tosolve. It’s not limitedto the US.”

Follow-up discussions withCEO

Six years of press reportsand industrydocumentation on firm

Three days of on-sitemeetings in U.S. HQ.

NairSingapore

Medicalsoftwaresolutions

$1.8M 10 2000 IndiaJapanAustraliaMalaysia

0.805.153.660.85

3333

“We have to gointernational—it isunavoidable.”

One day of meetings in HQDiscussions with advisor to

TMT13 archival documentsFollow-up interview with

CEO

a HQ � “headquarters.” Calculation of the cultural distance of each focal country from each HQ nation is based on Hofstede’s (1980) rankscores. In the entry mode column, 1 � “distributor,” 2 � “joint venture,” 3 � “alliance,” 4 � “acquisition,” 5 � “greenfield.”

b Assessed at the end of data collection.c Representative quotes.


emergent themes and theoretical relationships link-ing experience and learning based on the insightsfrom each firm.

Since one of our research questions asked iflearning sequences influence performance, an im-portant element of within-case analysis was deter-mining each firm’s performance in each countryentered. We focus on country-level performancerather than overall corporate performance since thelearning sequences described in this study are spe-cifically related to learning during internationaliza-tion. In keeping with prior studies on internation-alization, we assessed country performance in

multiple ways (Brush & Vanderwerf, 1992; Delios &Beamish, 2001; Dunning, 1980; Geringer & Hebert,1990). First, we assessed country performance asthe time until the first sale in a new country, cal-culated in months. Second, we assessed countryperformance as the time until a firm broke even(covered costs) in the new country entered, calcu-lated in months. We chose time until first sale andtime to break even because these metrics providereliable, objective measures of performance avail-able across the sample. Third, we assessed countryperformance through a question at the end of eachinterview, asking informants to rate the “success of

TABLE 2Description of Informants

Firmand HQLocation Positiona Age

Nationof Birth

WhenJoined Firm

Prior TMTExperience

Description of OtherInformants’ Focal Informant

JacksonU.S.

Director of marketingMarketing managerCEO and founder

222360

U.S.U.S.China

Founding2003Founding

NoNoNo

Quick learnerKnows U.S and Chinese cultureSeasoned entrepreneur

RytiFinland

VP operationsCofounder/VP technologyCofounder/VP business

developmentCofounder/VP technology

282825

27

FinlandFinlandFinland

Finland

FoundingFoundingFounding

Founding

NoYesYes

Yes

Thorough and calmInnovative, get-it-doneSmart, proactive

Hardworking, straightforwardWeeSingapore

CEO and cofounderGM, SingaporeCEO, Malaysia

453836

Hong KongSingaporeMalaysia

FoundingFounding2001

YesNoYes

Dynamic, fast, great “side-view”Technically soundStrategic thinker, can talk at all levels

KallioFinland

Executive VP marketingCofounder, VP salesCEOCofounder

29273628

FinlandFinlandFinlandFinland

FoundingFoundingFoundingFounding

NoYesNoYes

Analytical, able to consolidate thoughtsAble to cope with new situationsFriendly, impulsiveDirect, technical

AdamsU.S.

Founder/ chairmanVP of internationalManager, U.K.VP and director, Australia

50484734

U.K.U.K.U.K.Australia


NoNoNoNo

Mr. InternationalProfessionalProcess-orientedGood entrepreneur

StahlbergFinland

CEO, U.S.CEOCFOFounderDirector, Central EuropePresident, Japan

555243334740

U.S.FinlandFinlandFinlandFinlandFrance

19972000FoundingFounding20022000

NoNoNoNoNoNo

A sales and marketing-oriented personStubborn, focusedAnalyticalBrilliant, intense, highly focusedUnconventional, maverickCould sell snow cones to Eskimos

PolkU.S.

CEO and PresidentDirector, U.K.CTO, founder

453950

U.S.New ZealandChina

19972003Founding

NoNoNo

ImpatientScrappy, likes a lot of balls in the airTechnical

TylerU.S.

Chairman and CEOPrincipal engineerStaff scientistHead of sales,Europe and Manager,

Southeast AsiaCSO, cofounderDirector of salesCOO

5350314955

564652

U.S.U.S.JapanU.S.Belgium

U.S.U.S.U.S.

2000Founding199720002002

Founding20001997

NoNoNoNoNo

NoNoNo

UpbeatIntelligentLaid-back, talented in engineeringHas a good perspectiveKnows how to do global business

Forward thinkerSales orientedDefensive

NairSingapore

CEO and founderVP of business developmentBusiness director, U.S.

383038

SingaporeSingaporeMalaysia


YesNoYes

Reserved, methodicalPolite, optimisticHard worker, concise in communication

a VP � “vice president,” CFO � “chief financial officer,” CTO � “chief technical officer,” CSO � “chief strategy officer,” COO � “chiefoperating officer.”


the firm in the new country after the first year” (0 �“very unsuccessful,” 5 � “moderately successful,”10 � “extremely successful”); we then computedthe mean response of informants in each firm. OurLikert measures are likely to be accurate as theyspan functional and hierarchical levels, therebyproviding assessment of country performance fromseveral vantage points. Overall, because we used asmall sample to ensure depth of understandingabout learning sequences our quantitative analysison performance is limited. However, because wetook multiple, independent measures of countryperformance, our study helps provide a more reli-able assessment of performance than is possiblefrom one performance measure alone (Zahra &Dess, 2001).

As an important additional step, we also assessedboth shorter and longer-term performance conse-quences resulting from following each learning se-quence. We did this because prior theoretical re-search on learning suggests that learning may haveboth immediate and distant consequences(Levinthal & March, 1993; March, 1991; Sapienza,Autio, George, & Zahra, 2006). To assess shorter-term performance consequences of learning se-quences we averaged the scores for the first andsecond country entries for each of the three perfor-mance metrics.4 Likewise, to assess longer-termperformance consequences, we averaged the scoresfor the third and fourth country entries for each ofthe three performance metrics.5

After within-case analysis, we began cross-caseanalysis, looking for the emergence of similarthemes and relationships related to learning inmultiple cases (Eisenhardt, 1989; Miles & Huber-man, 1994). During cross-case analysis, we used avariety of lenses, including grouping sample firmsas Finnish, U.S., or Singaporean and by executiveexperience and entry patterns. From the emergingpatterns, we formed tentative theoretical constructsand propositions. We then refined them throughreplication logic, frequently revisiting the data tosystematically compare and verify the occurrenceof specific learning sequences within each case. Wewere aware of the existing literature on commonlearning processes (i.e., trial-and-error learning, vi-carious learning, improvisational learning, and ex-

perimental learning) and so examined the data forthe emergence of these construct categories andtheir temporally ordered use in each country entry.But we also looked for unexpected learning pro-cesses. Thus, we combined theory elaboration (Lee,1999) and theory generation (Eisenhardt, 1989) inour analysis. We then iterated between theory anddata to clarify our findings and theoretical argu-ments. We also relied on research on learning, pro-cess management, and international entrepreneur-ship to sharpen the conceptual underpinnings ofour findings and visually depict them. For exam-ple, since sequences are a key focus in processresearch (Abbott, 1990; Langley, 2007; Van de Ven,1992) and our study examines learning processes,we decided to follow the schematic approach usedin other process research to depict our learningsequences. Following Langley’s (1989) use of theinitial letters of the names of people’s positions(i.e., M � “manager”; L � “line person”; and S �“staff person”) and arrows to depict interaction se-quences (e.g., L¡M¡S), we used the initial lettersof the names of types of learning processes (i.e., T �“trial-and-error learning”; V � “vicarious learn-ing”; E � “experimental learning”; and I � “impro-visational learning”) and connecting arrows to de-pict learning sequences (e.g., V¡T¡E). The overallanalysis involved iterations between data, theory,and later extant research until a strong match be-tween data and the theoretical framework occurred.

LEARNING SEQUENCES

Do Learning Sequences Exist, and Do They Matter?

The organizational learning literature generallyhas focused on particular learning processes, suchas trial-and-error learning or vicarious learning,one by one. Although this literature indicates thatthese different learning processes might work to-gether (Miner et al., 2001), it does not suggest thatlearning sequences exist. The data in our study,however, reveal their existence. We find that organ-izations temporally order the use of multiple learn-ing processes over time. Moreover, our data revealthe existence of several distinct learning sequences:“seeding” and “soloing.” We define seeding learn-ing sequences as those that begin with indirectlearning and then continue with direct learning.We define soloing learning sequences as those be-ginning with direct learning and then continuingwith direct learning. Seeding and soloing emergedfrom the data and reflect the ways that firms in ourstudy began to learn. That is, firms may begin learn-ing indirectly from others’ experience and so“seed” their subsequent direct learning. Alterna-

4 For example, if the time until the first sale in the firstcountry a firm entered was 10 months and the time untilfirst sale in the second country entered was 8 months, the“shorter-term time until first sale” would be 9 months.

5 For example, if the time until first sale in the thirdcountry entered was 20 months and the time until firstsale in the fourth country entered was 18 months, the“longer-term time until first sale” would be 19 months.


tively, firms may begin learning directly throughfirsthand experience (and so be “soloing,” in thatthey do not rely on others to learn).

We coded sequences as “seeding” if, in each newcountry entered, a firm first began using an indirectlearning process (e.g., vicarious learning) beforetransitioning to a direct learning process (e.g., trial-and-error learning). Likewise, we coded sequencesas “soloing” if in each new country entered, thefirm began using a direct learning process (e.g.,experimental learning) and then switched to an-other direct learning process (trial-and-error learn-ing or improvisational learning). Thus, a key dis-tinguishing feature of soloing learning sequences isthe absence of indirect learning. All nine of oursample firms developed seeding or soloing learningsequences in each of their country entries. See Ta-ble 3 for a summary of the sequence patterns weobserved and Appendixes A and B for more detailon the learning sequences in each firm over time.

We find evidence for two variations in seedingsequences. The first variation is vicarious learningfollowed by trial-and-error learning. As have otherresearchers, we assessed vicarious learning as achange in cognition and/or behavior resulting fromobserving others (Kim & Miner, 2007) and trial-and-error learning as a change in cognition and/or be-havior resulting from a completed action (Baum &Dahlin, 2007; Van de Ven & Polley, 1992). Yet,unlike prior work on vicarious learning, whichtends to report on one specific form of vicariouslearning, our analysis uncovered at least threeforms. Specifically, we show that vicarious learn-ing may take the form of (1) a modeling effect,wherein a firm imitates a competitor’s behavior, (2)an inhibitory effect, wherein a firm ceases behaviorafter observing another firm experience a negativeoutcome from pursuing that behavior, or (3) aneliciting effect, wherein a firm engages in a behav-ior similar to a competitor’s but in a different way.

Ryti offers an illustration of a modeling effect.Three young, inexperienced entrepreneurs (eachabout 25 years old) who had recently graduatedfrom Helsinki University of Technology foundedRyti, with the intention of creating software to ex-pedite drug discovery within the pharmaceuticalindustry. The founders developed technology thatallowed patients, research professionals, and datamanagers to quickly capture and report clinicaldata through PDAs, cell phones, and computersduring “phase three” clinical trials. Shortly afterfounding, the founders observed from Finnish com-petitor firms that accumulating trial experienceseemed key to gaining access to global customersand that experience appeared easy to get in nearbySweden. Given their lack of international experi-

ence to guide their actions, the founders copied theseemingly successful practice of competitors andentered Sweden. Trial-and-error learning then fol-lowed this vicarious learning. During project im-plementation with a Swedish firm, leaders in thelatter became frustrated with Ryti’s poor communi-cation with them. This outcome prompted Ryti ex-ecutives to improve their firm’s intranet and createa more effective dedicated e-mail list. One co-founder remarked, “Our customer got frustratedsince we were not actively sharing information on adaily basis. We learned to set up different mecha-nisms and improve the company intranet and thee-mail list to tackle that specific problem.”6

More intriguingly, our data show that beyond amodeling effect, vicarious learning may also takeother novel forms, such as an eliciting effect. Weeillustrates. The two founders of Singapore-basedWee had the goal of helping customers manageinformation security risks. When making their firstcountry entry, into Hong Kong, the founders de-cided to target banks as their primary customer.This choice was influenced by vicarious learning.The country manager of Singapore recalled, “Forthe majority of the banks [in Singapore] . . . theirphysical security, alarm and monitoring devices, isall taken care by Commercial Industry SecurityCorporation (CISCO). It has a monitoring service.Whenever there is a key or intrusion that takesplace, the physical security monitoring devicessend traffic back to the command center but in aphysical way. What we decided to do is exactly thesame thing in the cyber world.”

After engaging in vicarious learning, Wee execu-tives relied on trial-and-error learning. After enter-ing their first country (Hong Kong), these execu-tives used a sales approach of targeting IT groups inbanks. Yet leaders discovered that because newtechnology guidelines required senior executives tounderstand the risks associated with their technol-

6 Interviewees typically described learning as experi-enced by their executive teams (Daily et al. 2000). Al-though such executive team learning resembles teamlearning as described by Edmondson et al., 2001, weargue that it is more appropriately labeled “organiza-tional learning,” because the executive teams in our sam-ple constituted their entrepreneurial organizations’membership, leadership, and understanding. This viewis also consistent with the argument in the literature thatorganizational learning in an entrepreneurial firm is of-ten equivalent to individual learning, given that the firmconsists of a relatively small number of people and haslittle structure (Bingham & Eisenhardt, 2011; Kim, 1993;Zahra et al., 2000)—that is, entrepreneurial firms areoften equivalent to executive teams.


ogy, many Hong Kong banks had shifted responsi-bility for information security away from IT andinto audit. Given this outcome, Wee executives be-gan targeting audit groups instead of IT groups. The

CEO said, “There are a lot of organizations, includ-ing banks, which have transitioned from info-secu-rity under IT to info-security under the audit group.. . . So we changed.”

TABLE 3Learning Sequences

(3A) Country 1 Entry

Firm Sequence Country 1a International Experience of TMT at Time of Country 1

Ryti Seeding V¡T Low: 1 year“Among the founding team, there is a lot of inexperience.”

Jackson Seeding A¡T Low: 2 years“My role in China was to validate that the market is there, and that our solution iscompetitive.”

Wee Seeding V¡T Medium: 4 years“He [cofounder] lived in Australia for five years. . . . I worked for two years in theU.K. and a year an a half in New Zealand.”

Kallio Seeding A¡T¡I Low: �1 year“I [cofounder] lived for two months in Buenos Aires.” [“How long has the othercofounder lived outside Finland?”] “It’s countable in months, not years. He lived inthe Ivory Coast for one summer but that’s the longest.”

Tyler Soloing E¡T High: 6 years“John essentially had the long-term experience as an international salesperson. Hehad been in the industry for a long time and had a lot of sales contacts, basicallyknew how to process work. Hanz had worked for him in Sweden.”

Adams Soloing E¡T High: 13 years“Most of my jobs for the last 15–20 years have had an international context in thesense I have managed businesses outside my home country.”“He has lived and worked in South Africa, the U.K., Europe, and the United States.His experience has been exclusively global from the get-go.”

Stahlberg Soloing I¡T High: 6 years“We can speak the same language, we know the country, and we know thecompany.”“I have lived and worked outside my home country for almost 20 years.”“He had been a sales manager in Germany for Nokia previously for a year ortwo. . . . I also had some experience in dealing with Germany.”

Polk Soloing E¡T High: 10 years“We have international experience here, so we brainstorm a little bit about what weshould do and how we should do it. . . . I [CEO] have about 22 years ofinternational experience.”“Our manager for China has lived outside China for about 8–9 years.”

Nair Soloing E¡T¡E Medium: 4 years“I have been in the U.S. for four years. I also spent three years at the Center forNatural Products Research in Singapore.”

(3B) All Country Entries

Firm Sequence Country 1 Country 2 Country 3 Country 4

Ryti Seeding V¡T V¡A¡T¡E¡I A¡T¡E A¡D¡T¡EJackson Seeding A¡T A¡T¡D¡I A¡T¡I A¡T¡A¡IKallio Seeding A¡T¡I A¡T¡E A¡T¡E¡V A¡D¡T¡VWee Seeding V¡T V¡T¡I¡D A¡T¡I V¡T¡V¡ITyler Soloing E¡T T E¡T TAdams Soloing E¡T E¡T E¡T TStahlberg Soloing I¡T I¡T T TPolk Soloing E¡T I¡T I¡T TNair Soloing E¡T¡E E¡T¡E¡A T¡A T¡A

a V � “vicarious learning,” T � “trial-and-error learning,” A � “learning from external advice of others (via contact),” I � “improvi-sational learning,” E � “experiential learning,” D � “deviance-error learning.” Please see Appendixes A and B for more specific detail onlearning sequences used by each sample firm in each of its country entries.


Our study also indicates another variation inseeding sequences. Although the first variation isvicarious learning followed by trial-and-errorlearning, the second variation is advice from exter-nal entities, such as consultants, venture capital-ists, and partners, followed by trial-and-error learn-ing.7 Jackson illustrates the use of this secondseeding sequence variation (see Appendix A). Atthe outset of their entry into their first country,China, executives at Jackson hired a consultant tohelp understand how to enter the market. Regard-ing the China entry, one executive noted that theconsultant “spear-headed the whole thing.” Fromthe consultant, Jackson executives learned that thedue diligence process in China is long and that theyneeded to work with multiple distributors becausethe market is so big and regionally diverse. Jacksonexecutives entered the country and began workingwith multiple distributors as instructed by the con-sultant. As one vice president (VP) stated, “[Con-sultant] has been working with us to help us knowhow to develop relationships with distributors inChina.”

After beginning with advice from the externalconsultant, Jackson executives turned to trial-and-error learning. For example, after entering Chinaand trying to promote the firm’s wireless chips,executives discovered that Chinese firms didn’twant to buy chips alone. Instead, they wanted turn-key solutions. A senior leader stated, “We foundthat companies in China require a very complete,end-to-end solution.” On the basis of this new in-formation, Jackson executives changed their behav-ior and started providing turnkey solutions thatbetter set the firm apart from the competition. Thesenior leader continued, “Texas Instruments, In-tel—the big guys—they don’t really provide thatcomplete, end-to-end solution the way a start-uplike we can, so it allows us to differentiate.”

Although our data show evidence of seedinglearning sequences (in which firms start with anindirect learning process and then switch to a di-rect learning process), they also show evidence ofsoloing learning sequences (in which firms startwith a direct learning process and then switch toanother direct learning process). As Table 3 indi-cates, there were several variations in soloing se-quences (see Appendix B for more detail on these

learning sequences). Some firms began with impro-visational learning before moving to trial-and-errorlearning. More commonly, however, firms beganwith experimental learning before moving to trial-and-error learning. Intriguingly, however, thoughthe literature suggests that experimental learninggenerally occurs through deliberate, small-scaletests conducted “off-line” in controlled settings tohelp managers gain understanding (Miner et al.,2001; Pisano, 1994; Thomke, 2003), our data sug-gest that such experimental learning frequently oc-curs “online,” as managers try to learn while takingadvantage of transient and unpredictable windowsof opportunity.8

Adams provides an illustration of this soloingsequence variation. Adams is a U.S.-based firmwhose technology allows companies to integratereal-time information and personalized analyticsinto their corporate information portals, enterpriseapplications, and critical business processes. Lead-ers started with experimental learning when enter-ing their first country, Australia, which theyviewed as a market culturally similar to the U.S.and one in which leaders could learn to do inter-national business. One executive stated, “Australiais a good test bed. . . . It’s low risk and easy to seewhat drives profitability.” As part of the experi-ment, corporate leaders gave a highly experiencedcountry manager a lot of autonomy to run the Aus-tralian business. However, they discovered thatthis autonomy resulted in an Australian venturethat became too disconnected from corporate poli-cies. This experimental outcome helped corporateleaders see the need to ensure more control andoversight of foreign teams and ventures. After start-ing with some experimental learning, Adams exec-utives then learned through trial and error. Corpo-rate leaders began pushing the local countrymanager to use a “features and functions” sellingapproach (an approach in which the seller’s mes-sage is “I’ll tell you the features and functions. Youfigure it out whether it suits your need or not”).However, the country manager experienced verylittle success using that approach. Because of thisnegative outcome, the country manager and corpo-rate leaders decided to adopt a new selling ap-proach that emphasized solutions/consultative ser-

7 Both variations of seeding involve firms starting withan indirect learning process; the key difference betweenthe two variations is that vicarious learning does notinvolve a focal firm’s deliberate contact with other firms(i.e., distant observation), whereas learning from the ad-vice of external firms does involve contact withother firms.

8 Our study suggests that experimental learning maybe more nuanced than is described in the literature. Wefind that experimental learning often occurs online in theform of executives’ deliberately trying different sales ap-proaches in comparative contexts to see which is mosteffective, seeing which market responds first to inquiries,or trying a new ownership structure to see if it is moreeffective than what exists.


vices. The country manager explained, “Thecorporation was fairly infantile at the time wesigned on in Australia, so it did not necessarilyhave lots of relevant selling experience that wasuseful. There was a much greater tool- or product-selling mentality. We were better able to get intothe market with a solution-selling methodology.”

Overall, we find evidence for learning sequences.In general, this finding is important because priorprocess research on learning has not explored theconcept of sequences directly and so does not spec-ify if some learning processes may be used earlierversus later. A key question, though, is why firmsuse seeding sequences and soloing sequences. Theprior international experience of executives ap-pears to be relevant here (see Table 3). Conceptualresearch suggests that when entrepreneurial firmsenter new countries for the first time, they oftenlack the organizational structure needed to makecollective responses (Sapienza et al., 2006). Thisresearch also suggests that in these situations, pre-history resources, such as the prior experience ofexecutives, likely play a salient role (Helfat &Lieberman, 2002). Our empirical study showed thatthe prior international experience of top manage-ment team members9 appears to shape whetherfirms begin by either learning directly (throughfirst-hand experience) or indirectly (through oth-ers’ experience) and so whether firms use seedingsequences or soloing sequences.

Recall that in seeding sequences, firms start withan indirect learning process before transitioning toa direct learning process. They do so because theirexecutives are often inexperienced in the context inwhich knowledge is needed. Thus, our data showthat in firms whose top management team had little(if any) international experience, vicarious learningor learning from the advice of external parties wasused before trial-and-error learning in initial coun-try entries (see Table 3). In contrast, firms usingsoloing sequences did not rely on indirect learning.Rather, they focused exclusively on a sequence ofdifferent direct learning processes (e.g., experimen-tal learning or improvisational learning precedingtrial-and-error learning). Firms appeared to use so-loing learning sequences because their executiveswere more experienced in the context in which

knowledge was needed. Therefore, when its topmanagement team had more prior international ex-perience, a firm did not use indirect learning ininitial learning sequences (see Table 3). Several topmanagement team members at Adams, as one illus-tration, had extensive international experience andrelied more on this experience than on guidancefrom external sources. For example, the VP of In-ternational stated, “Most of my jobs for the last15–20 years have had an international context inthe sense I have managed businesses outside myhome country.” Further, this VP described thefirm’s CEO, a person with over 20 years of interna-tional experience, as “Mr. International,” saying“He has lived and worked in South Africa, the U.K.,Europe, and the United States. His experience hasbeen exclusively global from the get go.”

We also find that initial learning sequences mat-ter. Firms that use seeding sequences (i.e., startwith indirect learning before direct learning) do notappear to perform as well in the shorter-term asfirms that use soloing sequences. Specifically,firms that used seeding sequences in their first twocountry entries took more time to capture their firstsale, took more time to break even, and had loweroverall ratings of success than firms that used so-loing sequences (e.g., experimental or improvisa-tional learning before trial-and-error learning). Fig-ure 2 provides data and illustrative graphs. Withlittle international experience, leaders in thesefirms generally had less understanding of how tocoordinate internal activities such as sales andproduct adaptations in new foreign markets (Car-penter & Fredrickson, 2001; Carpenter, Sanders, &Gregersen, 2001; Daily, Certo, & Dalton, 2000). As aresult, these leaders appeared to first use indirectlearning; they looked to other firms around themfor clues about how to perform initial country entryactivities. Although such indirect learning is effi-cient, our data suggest that it may be less helpful forearly performance, as what is gained is often non-strategic surface level knowledge that is not tai-lored to a firm’s specific needs and situation. Forexample, prior to entering their second country (theU.S.), inexperienced leaders at the Finnish firmRyti learned indirectly from the advice of a Finnishgovernment agency named FinnPro about “practi-cal details regarding local contacts and then alsosome market information.” However, a VP notedthe restricted value of this indirect learning whenhe stated, “FinnPro was not very helpful. I think wecould have got that information from various othersources. That wasn’t particularly valuable informa-tion.” Moreover, the lack of direct experience onthe part of executives using vicarious learning oftenleads them to not fully grasp true causal links be-

9 In keeping with prior research, we define TMT mem-bers as those directly in charge of a firm’s strategic deci-sions and overall competitive positioning (Daily, Certo, &Dalton, 2000). Also in keeping with the literature, weassessed international experience as the number of yearsorganization members had lived and worked outsidetheir home country (Carpenter & Fredrickson, 2001; Car-penter et al., 2001).


tween others’ actions and outcomes (Lane & Lubat-kin, 1998). Vicarious learning may therefore lead topartially incorrect knowledge. Another cofounderof Ryti recalled how this was case when the firmentered the U.S.:

Eighty percent of the top 25 pharma companies areheadquartered in the United States and all of thecompetition was in the United States so we decidedto rent an apartment in Cambridge (MA). I boughtthe flight ticket for the guy and handed him a tele-phone book listing the pharma companies and senthim off. . . . We just thought that we would be able tocome and close the deals and build the market po-sition afterwards. But we realized that you reallyneed to build the market position and then you startclosing the deals. So we were coming at it from thewrong end in the beginning based on what we sawothers do. I worked for one year for McKinsey andthen for a year doing academic research at HUT, Iworked a couple of months for some pharma com-panies, but I didn’t have for myself any relevantbusiness experience. I think that this is the biggestflaw that we used to have. We are very young, veryenergetic, which is good, but . . . among the found-ing team, there is a lot of inexperience.

By contrast, firms that used soloing sequencesseem to perform better in the shorter-term (see Fig-ures 2 and 3). In their first two country entries, theytook less time to capture their first sale, less time tobreak even, and had higher overall ratings of suc-cess than the firms using a seeding sequence. Thesefirms appeared to perform better because of theprevious international experience of their execu-tive teams. Such experience among executives innew entrepreneurial firms seems to positively in-fluence the outcomes of early internationalizationfor several reasons. First, it can decrease the timeneeded to identify opportunities as well as the timeneeded to capture opportunities given existing net-works and access to resources. The founder of Polkprovided support for these points when he said, ofhis first country entry into China, “Because Iworked there I know that there is a gap. As a busi-nessman, you want to be the bridge and take aprofit when you connect two places together. . . .The first sale in China was quite easy. I have friendsthere.” Executives’ prior international experiencealso lowers the risk and so the cost of experimen-tation to uncover high-performance organizationalsolutions (Sapienza et al., 2006). For example,when discussing how his firm decided to go intoAustralia (in a first country entry) shortly afterfounding, the CEO noted the use of experimentallearning based on the prior experience of a topmanagement team member (“John”). Said he, “Westarted thinking about entering Australia based on

John. We knew he knew he had set up solutions inAustralia before. . . . The major experiment was let-ting John approach this thing on a much moresolutions oriented basis. To his credit, John notonly did it, but proved that it was indeed the ap-propriate way to market the product. It also endedup helping the U.S. as well.” The VP of Interna-tional concurred when he remarked, “John ap-proached us and said he wanted to go back toAustralia . . . with us he was essentially re-writinga business very similar to the one he started be-fore. . . . It was so clear that he knew what to do . . .it clicked from the beginning.” In summary, ourdata suggest that particular learning sequences ex-ist, that they are influenced by prior executive ex-perience, and that they appear to be consequentialto early performance. Collectively, these observa-tions lead to the following propositions:

Proposition 1. Firms use seeding or soloinglearning sequences.

Proposition 2. More executive experience atthe time of first entry is more likely to lead tothe use of a soloing sequence.

Proposition 3. Less executive experience at thetime of first entry is more likely to lead to theuse of a seeding sequence.

Proposition 4. Use of a soloing sequence leadsto higher performance in the shorter term thanuse of a seeding sequence.

Do Learning Sequences Evolve over Time, andDoes This Matter?

Our first section helps address two of our re-search questions (i.e., do learning sequences exist,and do they matter) by describing the existence oflearning sequences and how they do matter forshorter-term performance. This section now ad-dresses the remaining research question: Do learn-ing sequences evolve over time? Further, we de-scribe how the evolution of learning sequencesmatters for longer-term performance.

Two patterns related to evolution emerged fromthe data. The first is sequence expansion. Firmsthat used seeding sequences in their first countryentry expanded the number of learning processesused in subsequent country entries. By contrast,firms that used soloing sequences in their firstcountry entry contracted the number of learningprocesses used in subsequent country entries. Weassessed expansion and contraction of learning se-quences by tracking the addition and deletion oflearning processes over time in each new country


entry. Figure 1 and Table 3 summarize this finding,and Appendixes A and B provide more detail for it.

As Table 3 shows, there were variations in theexpansion of seeding sequences. An interestingvariation was firms iterating between indirect anddirect learning. During this iteration sometimesfirms would rely on different indirect learning pro-cesses (e.g., vicarious learning or learning from theadvice of external firms), whereas sometimes theywould rely on the same process. For example,when Wee leaders entered their fourth country(China) they used vicarious learning at the onset ofthe entry (i.e., entry of a competitor firm into thelarge market of China helped persuade Wee leadersto enter). After entering and learning through directtrial and error, Wee leaders then again relied onvicarious learning (i.e., they saw other foreign firmsexiting China when sales plummeted during theSARS epidemic and so decided to do the same).

A more common variation, however, was firms’expanding their number of direct learning pro-cesses after engaging in indirect learning. More-over, we found that firms appeared to be using anovel direct learning process, one that differs fromother direct learning processes discussed in theliterature (i.e., trial and error, experimental, andimprovisational). We call this new process devi-ance-error learning. Deviance-error learningemerged from the data, and we define it as breakingaway from a previously successful action pattern,the consequences of which are a drop in perfor-mance and then a return to the previously usedaction. Thus, with deviance-error learning, firmmembers learn the true importance of a prior actionpattern when they move away from it and see per-formance dip. Deviance-error learning is therefore a

different process from extant notions of trial-and-error learning in which firms only change currentaction patterns when performance falls below aspi-rations (Baum & Dahlin, 2007; Greve, 2003).10

Jackson illustrates. When leaders entered theirsecond country, Taiwan, they first relied on indi-rect learning in the form of advice from a hiredexternal consultant (“C”), who provided insightabout competitive positioning in Taiwan. One VPremarked, “We were looking at cell phones, auto-motive navigation systems, and PDAs. Of thosethree product segments, we asked [C] about thecompetitor companies—What do they make? Howwell do those products fit in to what we have tooffer? What is the value proposition that we canoffer x company, y company, z company? His (C’s)role has really been teaching us how to think aboutapproaching companies.” Deviance-error learningthen followed learning from advice. Leaders movedaway from using indirect sales through distributors(a successful action pattern established in their firstcountry, China) to focus on direct sales. However,lack of market responsiveness in Taiwan helpedJackson leaders realize the need to refocus on indi-rect sales in Taiwan and subsequent countries toimprove speed to market and bridge lack of localunderstanding. Leaders therefore returned to workwith distributors as they had in their first countryentry. Trial-and-error learning came after deviance-error learning and learning from advice. In discus-sions with Taiwanese firms, Jackson leaders foundout that their CEOs often were not the final deci-

10 Please see Table 4 for greater details on distinctionsamong learning processes.

FIGURE 1Learning Sequence Evolution

5

4

Number ofLearningProcesses

Used

3

2

1Country 1 Country 2 Country 3 Country 4

Soloing

Nair

Polk

Adams

Tyler

Stahlberg

Jackson

Ryti

Kallio

Wee

Seeding


FIG

UR

E2

Per

form

ance

Con

sequ

ence

sof

Lea

rnin

gS

equ

ence

sa

See

din

g

Sol

oin

g

Wee

Kal

lio

Ryt

i

Jack

son

Sta

hlb

erg

Tyle

r

Ad

ams

Pol

k

Nai

r

10 9 8 7 6 5 4 3 2 1

60 50 40 30 20 10

16 14 12 10 8 6 4 2

(2C

) S

ucc

ess,

Fir

st Y

ear

(2B

) T

ime

to B

reak

Eve

n(2

A)

Tim

e u

nti

l F

irst

Sal

eN

air

Pol

kA

dam

sTy

ler

Sta

hlb

erg

Sol

oin

g

444n.a

.3

13 8 6 4n.a

.

8877 3n

.a.

n.a

.14

1466 9

27 10 18

6 62 44

4

44

33

333 3

99

99

99

2 22

21

111

5

57

7 88

8T T T T T

AT A

I

TI

T

I

TI

TE

TE

TE T

E T

E

TE

T

TT

E

T

EE

T

E

A

Jack

son

6R

yti

8K

alli

oW

eeS

eed

ing

Fir

ms

5724 5 48 36

7574

45273 1218 29

6 9 5 96

6

88

87

77

75

2

2

3 33

3

31

11 1

5

6

6

A T

V T

A T

IV

T

A T

D I

A T

E V

IA

T E

V T

I D

A T

IA

T E

VA

T E

A T

IA

T A

IA

D T

EA

D T

VV

T V

I

AB

CD

EF

Cou

ntr

y 1

Cou

ntr

y 2

Cou

ntr

y 3

Cou

ntr

y 4

Sp

ecif

ic L

earn

ing

Seq

uen

ces

Ran

kin

g am

ong

Sam

ple

s(F

) L

onge

r-Te

rmS

ucc

ess,

Fir

stY

ear,

Cou

ntr

ies

3 an

d 4

(C)

Sh

orte

r-Te

rmS

ucc

ess,

Fir

stY

ear,

Cou

ntr

ies

1 an

d 2

(E)

Lon

ger

Tim

e to

Bre

akE

ven

,C

oun

trie

s 3

and

4

(B)

Sh

orte

rT

ime

to B

reak

Eve

n,

Cou

ntr

ies

1 an

d 2

(D)

Lon

ger

Tim

e u

nti

lF

irst

Sal

e,C

oun

trie

s 3

and

4

(A)

Sh

orte

r T

ime

un

til

Fir

st S

ale

Cou

ntr

ies

1 an

d 2

Mon

ths

Mon

ths

Lik

ert

Sca

le

Sh

orte

r T

ime

un

til

Fir

stS

ale

Lon

ger

Tim

eu

nti

l F

irst

Sal

e

Lon

ger

Tim

eto

Bre

ak E

ven

Sh

orte

r T

ime

to B

reak

Eve

nS

hor

ter-

Term

Su

cces

sF

irst

Yea

rL

onge

r-Te

rm S

ucc

ess

Fir

st Y

ear

aT

ime

isn

um

ber

ofm

onth

s.D

is“d

evia

nce

-err

orle

arn

ing”

;Vis

“vic

ario

us”

;Tis

“tri

al-a

nd

-err

orle

arn

ing”

;Eis

“exp

erim

enta

lle

arn

ing”

;Ais

“lea

rnin

gfr

omot

her

s(v

iaco

nta

ct)”

;I

is“i

mp

rovi

sati

onal

lear

nin

g.”

sion makers. Rather, it was someone at a lowerlevel. Consequently, Jackson executives changedtheir sales approach to appeal to lower levels too.Finally, after engaging in learning from advice, de-viance-error learning, and trial-and-error learning,Jackson relied on improvisational learning. Duringproduct delivery to a Taiwanese customer, Jacksonengineers had to reconfigure their product solutionon the fly when the customer told them they onlywanted development boards with documentationand support instead of the full solution Jacksonengineers had prepared. Through the improvisa-tional episode, leaders realized that this stripped-down, no-frills solution could be offered to moretechnologically savvy customers in later countries.

Contrary to the expansion of seeding sequences,wherein firm leaders began increasing the numberof learning processes used over time, we also findsupport for the contraction of soloing sequences,where firm leaders begin decreasing the number oflearning processes used over time (see Figure 1 andTable 3). For example, though executives at Stahl-berg relied on both trial-and-error learning and im-provisational learning in their first two countryentries, they relied only on trial-and-error learningin their next two. To illustrate, shortly after theFinnish firm entered the United States (third en-try), corporate executives moved marketing andsales functions there, perceiving the U.S. market tobe the firm’s most important one. However, execu-tives soon found that moving too many functions tothe U.S. too soon made Stahlberg overly “U.S. cen-tric.” As a result, corporate executives decided tomove marketing and sales back to Finland, wherethey could have greater control. Likewise, when thefirm entered Japan (fourth entry), corporate execu-tives wanted the local country manager to use acommission-based profit and loss system. Yet cor-porate executives found that the commission-basedsystem did not work well and so switched to trans-fer-based pricing. The country manager of Japanexplained the trial-and-error learning: “Corporatewanted to do a commission-based establishment.. . . So we did a commission agreement until they

[corporate] saw it would be better to do transfer-price-based accounting.”

Why do some learning sequences contract overtime, while others expand? Firms that use soloingsequences eliminate some learning processes overtime because they performed well in the shorterterm, and so executives became overconfident. Ex-ecutives seemed to assume they already knew howto do business in different countries and thereforeonly needed to draw on their own experientialwisdom. A Stahlberg leader conveyed the essenceof this point when he said, “I have a long experi-ence of many sectors, and therefore I have a wideunderstanding of many different problems. . . . Ipretty much know the culture in these countrieswhich I cover.” Similarly, another Stahlberg leaderremarked, “I knew pretty much everything before,so I haven’t learned many new things.” Executivesat Polk reflected similar sentiments. A Europeandirector asserted, “I obviously understand what ittakes to set up an office and create demand and alsoall the legal implications of doing everything here.”

By contrast, seeding sequences expand. Firm ex-ecutives increase the number of learning processesover time because the firms performed less well inthe shorter term and so executives do not becomeoverconfident. Because of their lower performancein initial country entries and their own relative lackof international experience, executives in thesefirms may feel that more learning processes areneeded for them to better understand how to act insubsequent country entries. One of the founders ofWee hinted at this when describing how his firmbegan to rely on more learning processes in theirthird country entry. Said he, “We knew that our biglimitation was that we only had the technicalknow-how and domain knowledge. What we didn’thave was the domestic country knowledge. That iswhat a local partner could give us.” Likewise, a VPat Jackson noted how his firm started using morelearning processes in later country entries:

It is now a mix of what we are learning on the field,what our marketing guy in Asia is telling us, pluslooking at reviews, looking at electronic press re-leases, looking at articles written about companies.We use lots of sources of information. If I were tobreak it down, I say the VC [venture capital] ones aregenerally the highest-quality sources of informationbut we learned they cannot always hit all the mainpoints. There are a lot of gaps.

A related question is if the expansion of seedingsequences and the contraction of soloing sequencesmatters. We find that the former seems to lead tohigher performance and the latter seems to lead tolower performance. Specifically, during their third

FIGURE 3Learning Sequence Performance Consequences

Seeding

Soloing + –

+–

Longer TermShorter TermLearning Sequence


and fourth entries, firms using seeding sequencestook less time to capture their first sale, took lesstime to break even, and had higher overall ratingsof success after the first year than firms using solo-ing sequences (see Figure 2). Hence, a key insight isthat although the use of soloing sequences leads tohigher performance than the use of seeding se-quences in the shorter term (the time it takes for afirm to achieve its first two country entries), thepattern is reversed for performance in the longerterm (time to the third and fourth entries) becauseof expansion and contraction.

The expansion of seeding sequences seems tolead to higher performance for several reasons.First, it provides greater opportunities to learn.This appeared to be the case for Wee when, in itsthird country entry, it learned from the advice of apartner (a learning process the firm had not useduntil that point). The firm’s general managerrecalled:

Our partner sent a group of engineers. They wouldbe in the room for days, going through the newfeatures we have, testing them out and doing theuser acceptance test. They point out things that Sin-gapore and Malaysia would miss. For example, aWindows screen. Sometimes there is a box we clickto say “okay.” They said that the box is not the samesize as the others. I am not kidding you. This is thelevel they go down to. The amount of improvementsthey can point out is tremendous. Because of this,the new version of our product, our monitoring sys-tem, benefited a lot of other countries as well.

The expansion of seeding sequences also appearsto increase performance by improving the reliabil-ity of what is learned. More learning processes mayprovide a multimodal method for comparing dataagainst each other. Any finding or conclusion car-ries more weight and is likely to be more convinc-ing if based on the pooling of several distinct butcorroboratory sources of information (Yin, Bate-man, & Moore, 1983). As one executive at Jacksonnoted about the increasing number of learning pro-cesses his firm relied on to make partner choices:

By looking at our competitors, we get a very goodgauge of who the partnering companies are. Thenwe obviously supplement that with all the standardweb site press. Also talking to analysts too. We havebeen in contact with analysts who specialize in au-tomotive aftermarket and OEMs and handsets ask-ing them what’s hot, what’s interesting, and who aregood companies coming up in this area.

More broadly, the use of many distinct learningprocesses may improve performance since it helpsaddress a fundamental trade-off between the speedof learning and the quality of what is learned. On

the one hand, processes such as experimentallearning and trial-and-error learning are time con-suming, resource intensive, and not very efficient.Yet the knowledge generated is often of high qual-ity and so likely to reduce the future probability ofmistakes (see Table 4). On the other hand, vicariouslearning is easy and efficient. But, because theknowledge generated is based on weak causal in-ferences drawn from others’ observable actions, itis of lower quality and so less likely to reduce thefuture probability of mistakes. The use of morelearning processes in sequences may therefore letfirms acquire quality information while also allow-ing for speed in action. Overall, although manystudies have suggested that greater performancefrom learning stems from enlarging the number(Anand & Khanna, 2000; Argote, 1999) or variety(Hayward, 2002; Schilling, Vidal, Ployhart, & Mar-angoni, 2003) of experiences to be observed, ourstudy suggests the complementary idea that it cancome from enlarging the number and variety oflearning processes used together in a singleexperience.

Data also suggest several reasons why the con-traction of soloing sequences might lead to lowerperformance. First, increased reliance on trial-and-error alone increases the likelihood of repeatingpast actions that resulted in positive outcomes butthat may be inappropriate for current experience. Asenior leader at Tyler provided support for thispoint when he said, “I don’t think we did anyexternal market research on Europe before entry . . .our foray has been ‘Well, this is what FDA ap-proved. Therefore, it must be good for Europe.’ . . .Guess what? We sold nearly nothing, that doesn’twork.” This leader later admitted, “We designedour product to meet U.S. requirements, becausethat was where we were and all the research thatwe did was catered to U.S. So to watch it overseasand hope it’s going to sell is hope.”

Second, because critical examination of cause-ef-fect relationships and the creation of new knowledgeare severely reduced when individuals rely only ontrial-and-error learning, performance also seems todecrease as executives become more influenced bypsychological proclivities that push them to attributenegative outcomes to external factors (Weiner, 1985).For example, explaining his firm’s poor outcomes inJapan (fourth country entered), the CFO of Stahlbergstated, “There has also been the problem that theAsian engineers are not as technically capable as theFinnish engineers.” Likewise, recounting his firm’slack of sales in Latin America, Tyler’s head of globalsales snidely remarked, “In European countries, theyset up a high standard of health care. In Latin Amer-ica, half of them still use voodoo.” In summary, less


TABLE 4Learning Processes

Characteristics

Direct Learning Processes Indirect Learning Processes

ExperimentalLearning

Trial-and-ErrorLearning

ImprovisationalLearning

Deviance-ErrorLearning

VicariousLearning

Learning fromExternal Advice

Definition Learning throughcontrolledsituations to testcausalpropositions andcreate newknowledge(Cook &Campbell, 1979).

Learning throughtheconsequences ofa firm’s previousactions (Baum &Dahlin, 2007).

Learning thatoccurs on the flyas design andaction converge(Miner et al.,2001).

Learning thatoccurs whenfirms breakaway from asuccessfulaction pattern.

Learningindirectly fromother firmsthroughobservation butwithout contact(Bandura,1977).

Learning fromothersinstructionthrough directcontact (Dyer &Nobeoka, 2000).

How learningoccurs

Through theintentionalmanipulation ofinputs andobservation ofoutputs firmsgain knowledgeandunderstandingof causalrelationships.

Firms undertake acourse of actionand theconsequences ofthat completedaction lead tochange ininferences ofaction.

Firms adjust theirbeliefs and/orbehavior in realtime (withoutwaiting forconsequences ofaction) in orderto solveunexpectedproblems orcapturesurprisingopportunities.

Firms deviate froma previouslysuccessfulaction pattern;the result is aperformancedrop and then areturn to thepreviously usedaction. Firmslearn the trueimportance of aprior actionpattern whenthey deviatefrom it and seeperformancedip.

May take the formof (1) amodeling effect(a focal firmreplicates acompetitor’sbehavior), (2) aninhibitory effect(i.e., ceasingbehavior afterobservinganother firmexperience anegativeoutcome forpursuing thatbehavior), or (3)an elicitingeffect (i.e.,engaging in anaction similar toa competitor ‘sbut doing itdifferently).

Firms adjust theirunderstandingsand/or actionthroughinstruction fromexternal parties,includingpartners, VCs,and industryassociations.

Explicit intentto learn

High: Goal is todevelop newunderstandingsthat can then beincorporatedinto ongoing or-ganizationalactivities.

Medium: Trial anderror may beused as adeliberate formof learning. Trialand error mayalso be “blind”and involvelittle intent tolearn.

Low: Goal is moreto addresssurprisingproblems and/oropportunitiesand less to gainknowledgeabout action/outcomerelationships.

Medium: May be adeliberate formof learning. Mayalso be lessdeliberate;deviance from asuccessful pastaction patternmay havecapturing anovelopportunity asits goal, notlearning.

High: Goal is togain thebenefits ofaccumulatedknowledgewhile avoidingthe expense ofaccumulatedexperience.

High: Firms hireoutside othersand/or asktargetedquestions tohelp coverparticulardeficiencies ininternal stock ofknowledge.

Reliance onwhat wasdone in thepast

High: Outcomesfrom off-lineand onlineexperiments arecarefullycompared withwhat the firmshas done in thepast.

Medium: Firmsrepeat pastactions thatresulted inpositiveoutcomes. Ifoutcomes arenot positive,they revisebeliefs and/oractions asneeded.

Low: Design andaction convergein time andresult in a novelproduction thatis idiosyncraticto time andplace.

Low: Firms moveaway fromeffective actionsused in the past.But doing sohelps thembetterunderstand thereason for thoseactions.

Low: Firms usevicariouslearning tojump-startaction sincethey often donot have anexperience baseto draw upon.

Low: Firms moveaway from ownstock ofknowledge builtfrom the pastand rely moreheavily on theknowledge ofothers.

(Continued)


reflection about causal relationships stemming fromthe increased use of trial and error alone appears toincrease the probability that assessments of unfavor-able outcomes slip into a pattern of finger-pointingrather than occasions for learning. Consequently, un-derlying but unresolved problems propagate overtime so that performance gradually decreases withthe accumulation of experience. As support, the co-founder of Tyler sadly admitted, “Here we are yearslater and we have not yet found the key to completelysucceeding with this particular product.” Similarly,another member of the founding team disclosed, “Wehave the same problems in Asia as in Europe, and we

haven’t even started to get to that one yet.” Collec-tively, these observations lead to our second group ofpropositions.

Proposition 5. Seeding learning sequences ex-pand with use in later experience.

Proposition 6. Soloing learning sequences con-tract with use in later experience.

Proposition 7. Use of a soloing learning se-quence leads to lower performance in the lon-ger term than use of a seeding learningsequence.

TABLE 4(Continued)

Characteristics

Direct Learning Processes Indirect Learning Processes

ExperimentalLearning

Trial-and-ErrorLearning

ImprovisationalLearning

Deviance-ErrorLearning

VicariousLearning

Learning fromExternal Advice

Relation withother learningprocesses

May be used toguide trial-and-error learning.

May drive outindirect organiza-tional learning.

May lead to longerterm trial-and-error learning(Miner et al.,2001).

May be a viewedas a form ofunplannedexperimentallearning.

May be used toseed directlearning.

May seed directlearning andbecome usedmore thanvicariouslearning.

Exploration/exploitationemphasis

Exploration:Decision makersdeliberatelymanipulateinputs todiscover newknowledge andpractices.

Exploitation: Ownexperience isusually not asource of newideas but a basisfor refinement ofexisting ones.

Exploitation:Intention is tomake do withmaterials athand to create asolution for anemergentproblem and/oropportunity.

Exploration:Decision makersdeviate frompractices thatprovedsuccessful inpriorexperience.

Exploration:Decision makersengage in non-local searchwhen they lookbeyond theirboundaries fornew ideas andpractices. Oftenideas comefrom firms inthe sameindustry.

Exploration: Firmsengage in non-local search.May be moreglobal thanvicariouslearning sinceexternal sourcesof informationare notnecessarily inthe sameindustry.

Potentialbenefits anddetriments

Benefit:Knowledgegenerated maybe generalizableover time giventhat it containsunderstandingabout main andinteractioneffects.

Detriment:Experimentationcan be costly interms ofresources andtime away fromcore activities.

Benefit: Learningcan graduallyandsystematicallybuild off errorscommitted inthe past.

Detriment: Errorsmay have beenavoided throughexperimentationor vicariouslearning.

Benefit: Allowsfirms to rapidlyrespond toattractiveemergentopportunitiesfaster than thecompetition.

Detriment:Knowledge isshort-term andidiosyncratic toparticularexperiences andso lessgeneralizableover time.

Benefit:Knowledgegeneratedreflects moreunderstandingof whysuccessfulactions used inthe past aresuccessful.

Detriment: Can becostly in thatfirms deviatefrom successfulactions only tosee a drop inperformance andthen a return tothat successfulaction.

Benefit: Expediteslearning byavoiding directtrial and error.

Detriment:Knowledgegained could beless usefulgiven that it isbased onmaking weakcausalinferences fromobservation ofbehavior (i.e.,involvesdrawinginferences fromnoisy data).

Benefit: Expediteslearning byavoiding directtrial and error.Also improvesweak inferencesmade throughvicariouslearning.

Detriment: Costs tohire externalsources to helpinstruct andcoverinformationaldeficienciescould be high.Also requiresthat firms knowdeficiencies apriori.


DISCUSSION

Organizational learning is of fundamental impor-tance because it enables innovation, adaptation, andimprovement in efficiency and productivity (Argote,1999). Yet most research generally explores how oneparticular learning process is used while underex-ploring whether firms use multiple learning pro-cesses together over time in temporally ordered ways.Using data on the accumulated country entries ofentrepreneurial firms, we address this gap. Our find-ings have implications for several research areas, in-cluding process management, organizational learn-ing, and international entrepreneurship.

Process Research and Learning Sequences

Our primary contribution is to establish “se-quences” as a meaningful focus and concept inprocess research on learning. Process research cen-ters on understanding how things happen over timeand why they happen this way (Langley, 1999,2007). Whereas variance theories offer explana-tions for the world in terms of relationships be-tween independent and dependent variables (e.g.,more of A leads to more of B), process theories offerexplanations in terms of sequences of events, activ-ities, and choices (Langley, 1999). With processtheory, the concept of sequence thus takes centerstage. Intriguingly, although much process theoryon organizational learning exists (on, for example,trial-and-error learning or vicarious learning), theconcept of sequences has been noticeably absentfrom extant research. This absence may be partlyattributable to the fact that organizational learninghas a fluid character that makes it difficult to iso-late distinct learning processes and their temporalordering. However, by exploiting the benefits ofinductive multiple case methods for time seriesanalysis (Eisenhardt & Graebner, 2007; Yin, 1994),we are able to overcome some of these challengesand develop the concept of sequence in organiza-tional learning.

In line with other organizational research explor-ing sequences, our research uses the term to refer totemporally ordered elements (Abbott, 1990; Lang-ley, 1999). In our study, we are concerned with theorder of discrete learning processes used in firms.In particular, we develop the concept of learningsequences by addressing the central questions inrepresentative organizational process research onsequences, which include (1) whether sequencepatterns exist, (2) what influences those patterns,and (3) what the patterns affect (Abbott, 1990).

First, our study helps address the question ofexistence. Do learning sequences exist and, if so,

are there typical patterns? Our data reveal two pri-mary patterns: seeding and soloing. With seedingsequences, executives begin by using an indirectlearning process before using a direct learning pro-cess. By contrast, with soloing sequences execu-tives begin by using one direct learning process andthen switch to another direct learning process. Ourdata also reveal variations of seeding and soloingsequences. Thus, beyond showing support for theexistence of learning sequences more generally, wealso show support for common versions. For exam-ple, the most common version of soloing learningsequences used in initial country entries was firmsstarting with experimental learning and then tran-sitioning to trial-and-error learning. A less commonversion was firms starting with improvisationallearning and then transitioning to trial-and-errorlearning. Finally, our study provides some insightinto the internal interdependencies in learning se-quences and so helps address the question ofwhether certain orderings of learning processes areseldom or never used. As one illustration, datashow that experimental learning did not followimprovisational learning in any of the learning se-quences uncovered in our sample firms. One expla-nation is that with improvisational learning there isa lower explicit intention to learn, whereas withexperimental learning there is a higher explicit in-tention to learn (Miner et al., 2001). Hence, becausethe goal of improvisational learning is more to ad-dress surprising problems and/or immediate op-portunities, and less to gain information aboutcausal laws, it is less likely that experimental learn-ing, which centers on generating generalizableknowledge, will follow it.

Second, our study helps address the question ofwhy initial learning sequence patterns exist. Wetherefore explore factors that seem to influencewhether seeding or soloing sequences get used. Wefind that less international experience among TMTmembers makes firms more likely to use a seedingsequence in their initial country entry, whereasmore international experience among TMT mem-bers makes firms more likely to use a soloing se-quence. Indirect learning therefore appears to“seed” direct learning when firms lack experience.Hence, while extant studies contribute by high-lighting a range of potential learning processes (Hu-ber, 1991; Miner et al., 2001), this study contributesby suggesting that sometimes constraints (e.g., lackof international experience) may shape the order inwhich firms use those processes together over time.

Similarly, the presence or lack of internationalexperience also appears to influence why laterlearning sequence patterns exist. Unexpectedly, wefind that initial learning sequences both contract


and expand with continued use. Soloing sequencescontract. Executives with international experienceappear to get overconfident. Overconfident execu-tives overestimate the likelihood they can rely ontheir existing stock of knowledge and personal abil-ities for success in later country entries. As oneexecutive at Stahlberg said smugly, “I know what todo . . . selling is selling, whatever you sell even ifit’s a toothbrush. You have the same points whichyou have to go through. In each new country I’musing the same model, selling exactly the sameway.” This finding provides insight into the findingof other learning scholars as to why firms tend toexploit similar domain expertise when continuingto expand abroad (Vermeulen & Barkema, 2001).Seeding sequences, alternatively, expand. Execu-tives with less international experience recognizetheir lack of knowledge and so come to rely onmore learning processes when entering new coun-tries to address this deficiency. For example, afterthe first country entry, one senior executive re-marked, “We are no longer just relying on the in-formation that we think we have.”

Finally, our study helps answer questions aboutthe consequences of learning sequences. We exam-ined whether and how the order of learning pro-cesses used in a country entry influences countryperformance. We found that learning sequencesdiffer in their shorter-term versus longer-term per-formance impacts. Soloing sequences appear betterthan seeding sequences in the shorter term. Duringthis time frame, soloing sequences involve moreexperimental learning than seeding sequences. Theuse of more experimental learning seems to be par-ticularly performance enhancing early in experi-ence (see Figure 2), as the purpose of experimentallearning is to gain new knowledge and practicesthat can then be incorporated into organizationalactivities. Thus, we find some empirical supportfor the conjectures of others that experimentallearning may be useful in guiding subsequent trial-and-error learning (Miner et al., 2001).

The use of soloing sequences, however, appearsto result in lower longer-term performance than theuse of seeding sequences. One possible reason forthis is that continued use of soloing sequences re-flects a drift toward more local search (i.e., empha-sis on just trial-and-error learning) over time. Incontrast, the ongoing use of seeding sequencesseems to reflect a drift toward more global search(i.e., emphasis on different types of indirect learn-ing processes). Prior research suggests that suchglobal search is important for performance becauseit gives executives a fuller perspective and thusmake them less susceptible to learning biases (Ka-tila & Ahuja, 2002; Levinthal & March, 1993).

Therefore, although it may take a firm a while tointegrate the knowledge generated from a globalsearch, once integrated it appears to give firms aknowledge edge in the longer term. Hence, unlikeprior studies that suggest firms may emphasize in-direct learning more as performance falls belowaspirations (Baum & Dahlin, 2007) our study sug-gests indirect learning may become more empha-sized as performance nears or exceeds aspirations.Indeed, the learning sequences that appeared tolead to the highest performance in the longer termwere ones that reflected more, and more distincttypes of, indirect learning processes (see Figure 2).

In sum, we contribute by showing that particularlearning sequence patterns are present in organiza-tions. We also address what influences the choiceof and change in those sequences, and what thosesequences mean for important organizational out-comes such as performance. Overall, our studyhelps reveal the existence, effects, and evolution oflearning sequences. Such revelation is important,as existing process research on learning does notanswer questions about whether there are learningsequences, let alone questions dealing with theircauses and consequences. Thus, we contribute toprocess research on learning by developing theconcept of sequences, a concept that lies at theheart of process research (Burgelman, 1996; Graeb-ner, 2004; Langley, 1989; Van de Ven & Polley,1992) and one that scholars have noted as critical inthe development of theory about the temporal dy-namics of strategy-related phenomena such aslearning (Langley, 1999, 2007; Van De Ven, 1992;Van de Ven & Poole, 1995).

Organizational Learning

Besides establishing sequences as a meaningfulfocus and concept in process research on learning,our study also adds to organizational learning re-search by shedding more nuanced light on the na-ture of common learning processes discussed in theliterature (see Table 4).

First, our study provides an expanded view ofhow experimental learning occurs. Extant researchdescribes how experimental learning takes place asleaders intentionally manipulate inputs off-lineand then observe outputs to gain knowledge andunderstanding of causal relationships (Cook &Campbell, 1979; Huber, 1991; Pisano, 1994;Thomke, 2003). Our data indicate that althoughexperimental learning may occur off-line throughthe use of controlled situations to test causal prop-ositions and create new knowledge (which canthen be implemented in ongoing organizational ac-tivities), it frequently also occurs online, as execu-


tives deliberately try variations of practices andproducts as they go. This is because the uncertaintyassociated with the technology-based environmentin which our firms did business increased the timepressure to take advantage of serendipitous oppor-tunities faster than the competition. A senior exec-utive lent credence to this point when he remarkedabout his firm’s online experimental learning pro-cess of deciding which countries to enter: “Wethrow out our seeds and see which one will germi-nate fastest.” Thus, finite attention and time, un-predictable windows of opportunity, and limitedcapital increase the likelihood that fewer resourceswill be allocated to off-line experimental learningabout projects with only potential future value andthat more resources will be allocated to online ex-perimental learning about possibilities for immedi-ate revenue.

Second, our study provides an expanded view ofhow vicarious learning occurs. Current studiestend to describe one particular form of such learn-ing—modeling, defined as replication of a compet-itor’s behavior (Denrell, 2003; Kim & Miner, 2007).However, although this study shows support forvicarious learning via a modeling effect, it alsoshows that vicarious learning may take the form ofseveral different, less well understood, effects. Dataindicate that sometimes vicarious learning has aninhibitory effect, defined as ceasing behavior afterobserving another firm experience a negative out-come for pursuing that behavior. For example, inone firm, executives learned to create a more light-weight product prior to entering the U.S. afterwatching competitors suffer with a product thatwas not lightweight. The country manager of theU.S. said, “We actually learned a lot from the pio-neers out there. We have learned that the total costof ownership is an issue when loading a lot ofheavy-duty software for companies. So scalabilityfor them has been an issue. We decided we weregoing to go for lightweight software. . . . Havingseen where firms have gone and not really suc-ceeded helps us be better.”

Vicarious learning may also have an eliciting ef-fect, defined as engaging in a behavior similar tothat of a competitor firm but in a different way. Toillustrate, by watching their U.S.-based competi-tors, leaders in one firm realized the importance ofhaving a U.S. presence so that they could haveadded legitimacy for Asian customers. A seniorexecutive remarked, “We realized that it was im-portant to have an American base. . . . One of ourcompetitors in Taiwan at the time was [firm] andthey had a strong American presence. They had aTaiwanese reseller for them. They were an Ameri-can company with a Taiwanese reseller. So, it

somehow worked out really well for them, becausetheir Taiwanese reseller was able to leverage theAmerican image in the Taiwanese market.”

International Entrepreneurship

We also contribute to the growing literature oninternational entrepreneurship. Many studies inthis research stream describe how new firms inter-nationalize shortly after founding in pursuit of per-formance advantages (Autio, George, & Alexy,2011; Autio, Sapienza, & Almeida, 2000; Bingham,2009; Sapienza et al., 2006; Zahra et al., 2000). Ourstudy improves understanding of how these perfor-mance advantages might be realized and when.

First, we contribute by showing how determi-nants of performance advantages in the shorterterm may lead to performance disadvantages in thelonger term. Previous studies have suggested thatthe first entry of an entrepreneurial firm can bechallenging owing to liabilities of both newnessand foreignness (Zaheer & Mosakowski, 1997). Butin theoretical work, Sapienza and colleagues (2006)posited that the prior international experience ofexecutives may partially substitute for lack of or-ganizational experience and so mitigate the afore-mentioned liabilities. They thus suggest that entre-preneurial firms whose executives have previousinternational experience will perform better intheir first country entry than entrepreneurial firmswhose executives do not have that experience. Ourempirical study supports and extends this work.We find that entrepreneurial firms whose execu-tives had more previous international experiencegenerally exhibited higher average performance intheir first two country entries than entrepreneurialfirms whose executives had less previous interna-tional experience. However, our study also suggeststhat early performance advantages in initial coun-try entries may lead to overconfidence among ex-perienced executives and so lead to less learningand lower performance in later country entries.Hence, the prior international experience of exec-utives seems to generate shorter-term performanceadvantages but longer-term performance disadvan-tages. In general, our finding on the dampeningeffect of overconfidence on learning and perfor-mance over time helps address the question in theinternational entrepreneurship literature ofwhether the “imprinting” of early executives pro-vides continued performance advantages (Autio etal., 2011). It is also consistent with other researchon entrepreneurial firms suggesting that executiveteams with greater confidence tend to deprive theirfirms of important opportunities for learning andtherefore increase the likelihood that their firms


will underperform relative to their industries (Hay-ward, Shepherd, & Griffin, 2006).

Second, we contribute by showing how determi-nants of performance disadvantages in the shorterterm may lead to performance advantages in thelonger term. We find that entering a new countryfor the first time is a costly exercise for entrepre-neurial firms, especially when executives do nothave previous international experience. Lack of ex-perience causes some leaders to begin learning in-directly (see Table 3). Yet because indirect learningoften consists of making weak causal inferences foreffective actions based on distant observations ofothers’ behaviors (Kim & Miner, 2007), it can resultin incomplete and even inaccurate understandingsthat can lead to lower performance in initial coun-try entries relative to the performance of entrepre-neurial firms in which executives have previousinternational experience. However, our data alsoshow that relatively inexperienced firms withlower performance in initial country entries re-bound by adding more learning processes (see Ta-ble 3) and so perform better in later country entries.This finding on the increasing number and diver-sity of learning processes used over time mayprovide additional insight into why some entre-preneurial firms are able to develop larger andmore diverse action repertoires for new countryentry over time (Autio et al., 2011). More broadly,this finding helps extend the important conceptof “learning advantages of newness” discussed inthe international entrepreneurship literature(Autio et al., 2000) to executive, and not justorganizational, experience. Autio and colleaguesargued that younger firms are better able to inter-nationalize than older firms because older firmshave more structural and other institutional con-straints that make them increasingly resistant tochange. Similarly, we find that executives withless international experience prior to their cur-rent entrepreneurial venture appear to do morelearning over the longer term (i.e., increase thenumber of learning processes used in new coun-try entries) than those executives with more priorinternational experience and so outperform themin later country entries. By highlighting howlearning advantages of newness might well berelevant at the individual level of analysis (inaddition to the firm level), our work suggests theintriguing notion that inexperienced foundersmay constitute a more important long-termsource of competitive advantage for entrepre-neurial firms entering new markets than previ-ously theorized.

Limitations

Like all studies, ours has limitations that suggestopportunities for future research. To more accu-rately portray learning sequences over multiple ex-periences, we restricted our analysis to nine firms.Although we found intriguing patterns, more workis needed to examine learning sequences in a largernumber of firms and a wider range of industries.Likewise, we focus on learning across a series ofcountry entry experiences. It would therefore bevaluable to explore how our findings generalize (ordo not) to experience with other strategically rele-vant motors for growth, such as alliances, acquisi-tions, and product development. In addition, oursample consists of small, young firms in whichlearning may be more critical to survival than it isfor more established firms. Although a focus onyounger firms may allow for greater transparency oflearning, better understanding of the existence,causes, and consequences of learning sequences inolder firms is also needed. Similarly, we focus onthe information technology industry. It may be thatthis dynamic setting increases the number andrange of opportunities to learn. Finally, althoughwe identify particular learning sequences that seemto lead to higher performance than others in theshorter and longer terms, future research is neededto continue exploring which sequences are betterand under what conditions. More generally, all re-search designs involve trade-offs due to the practi-cal limits of data collection. We chose a small sam-ple to allow rich examination of learning sequencesand their potential causes and consequences. How-ever, although this choice increases the likelihoodthat findings will be fresh and internally valid, itdoes so at the expense of generalizability and ex-ternal validity. Thus, an important next step is tosubmit our findings and emergent propositions torigorous empirical tests.

Conclusions

Scholars know much about the importance oflearning and particular learning processes that or-ganizations use (e.g., trial-and-error learning, vicar-ious learning). But too little is known aboutwhether multiple learning processes get used to-gether in learning sequences. Our work is a firststep in addressing this gap. It is also part of a largerresearch program that addresses not only howlearning occurs in entrepreneurial firms and in dy-namic environments (Bingham, 2009; Bingham &Kahl, 2012), but also what is learned (Bingham &Eisenhardt, 2011), the impact of that learned con-tent (Bingham, Eisenhardt, & Furr, 2007; Davis,


Eisenhardt, & Bingham, 2009; Eisenhardt, Furr, &Bingham, 2010), and why some firms learn morethan others (Bingham & Eisenhardt, 2008; Bingham& Haleblian, 2012).

From our exploration of rich field data, we findthe existence of learning sequences. Moreover,we find that learning sequences seem to reflecttwo broad patterns: seeding and soloing. Thesetwo patterns show variation across firms, and theextent of executives’ international experience in-fluences their adoption. Our study also suggeststhat learning sequences evolve over time, but inopposing ways. Seeding sequences expand, andsoloing sequences contract. Further, our studysuggests that the performance benefits associatedwith each learning sequence are contingent uponwhen it is used: Soloing sequences seem to leadto higher performance than seeding sequences inthe shorter term, whereas the reverse appearstrue for the longer term. In sum, this study pro-vides an emergent and empirically groundedmodel for the existence, evolution, and effect oflearning sequences. More broadly, our studyhighlights the sequential nature of key strategicorganizational processes, thereby suggesting thatcareful, in-depth analysis may reveal intriguingnew insights in many other areas of managementresearch beyond learning.

REFERENCES

Abbott, A. 1990. A primer on sequence methods. Organ-ization Science, 1: 375–392.

Abbott, A. 1995. Sequence-analysis—New methods forold ideas. In J. Hagan & K. Cook (Eds.), Annualreview of sociology, vol. 21: 93–113. Palo Alto, CA:Annual Reviews.

Adair, W. L., & Brett, J. M. 2005. The negotiation dance:Time, culture, and behavioral sequences in negotia-tion. Organization Science, 16: 33–51.

Amis, J., Slack, T., & Hinings, C. R. 2004. The pace,sequence, and linearity of radical change. Academyof Management Journal, 47: 15–39.

Anand, B. N., & Khanna, T. 2000. Do firms learn to createvalue? The case of alliances. Strategic ManagementJournal, 21: 295–315.

Argote, L. 1999. Organizational learning: Creating, re-taining and transferring knowledge. Norwell, MA:Kluwer Academic.

Argyris, C., & Schön, D. 1978. Organizational learning:A theory of action perspective. Reading, MA: Ad-dison Wesley.

Autio, E., George, G., & Alexy, O. 2011. Internationalentrepreneurship and capability development-qual-

itative evidence and future research directions. En-trepreneurship: Theory and Practice, 35: 11–37.

Autio, E., Sapienza, H., J., & Almeida, J. G. 2000. Effectsof age at entry, knowledge intensity, and imitabilityon international growth. Academy of ManagementJournal, 43: 909–924.

Bandura, A. 1977. Social learning theory. Upper SaddleRiver, NJ: Prentice-Hall.

Barkema, H. G., Bell, J. H., & Pennings, J. M. 1996. For-eign entry, cultural barriers, and learning. StrategicManagement Journal, 17: 151–166.

Baum, J. A. C., & Dahlin, K.. B. 2007. Aspiration perfor-mance and railroads’ patterns of learning from trainwrecks and crashes. Organization Science, 18: 368–385.

Baum, J. A. C., & Haveman, H. A. 1997. Love thy neigh-bor? Differentiation and agglomeration in the Man-hattan hotel industry, 1898–1990. AdministrativeScience Quarterly, 42: 304–338.

Baum, J. A. C., & Ingram, P. 1998. Survival-enhancinglearning in the Manhattan hotel industry. Manage-ment Science, 44: 996–1017.

Baum, J. A. C., Li, S. X., & Usher, J. M. 2000. Making thenext move: How experiential and vicarious learningshape the locations of chains’ acquisitions. Admin-istrative Science Quarterly, 45: 766–801.

Bingham, C. B. 2009. Oscillating improvisation: Howentrepreneurial firms create success in foreign mar-ket entries over time. Strategic EntrepreneurshipJournal, 3: 321–345.

Bingham, C. B., & Eisenhardt, K. M. 2008. Position,leverage and opportunity: A typology of strategiclogics linking resources with competitive advan-tage. Managerial and Decision Economics, 29:241–256.

Bingham, C. B., & Eisenhardt, K. M. 2011. Rational heu-ristics: What firms explicitly learn from their processexperiences. Strategic Management Journal, 32:1437–1464.

Bingham, C. B., Eisenhardt, K. M., & Furr, N. R. 2007.What makes a process a capability? Heuristics, strat-egy, and effective capture of opportunities. StrategicEntrepreneurship Journal, 1(1–2): 27–47.

Bingham, C. B., & Haleblian, J. 2012. How firms learnheuristics? Uncovering missing components in or-ganizational learning. Strategic EntrepreneurshipJournal: In press.

Bingham, C. B., & Kahl, S. 2012. The process of schemaemergence: Assimiliation, deconstruction, unitiza-tion and the plurality of analogies. Academy ofManagement Journal: In press.

Brown, S. L., & Eisenhardt, K. M. 1997. The art ofcontinuous change: Linking complexity theory


and time-paced evolution in relentlessly shiftingorganizations. Administrative Science Quarterly,42: 1–34.

Brush, C. G., & Vanderwerf, P. A. 1992. A comparison ofmethods and sources for obtaining estimates of newventure performance. Journal of Business Ventur-ing, 7: 157–170.

Burgelman, R. A. 1994. Fading memories: A process the-ory of strategic business exit in dynamic environ-ments. Administrative Science Quarterly, 39: 24–56.

Burgelman, R. A. 1996. A process model of strategicbusiness exit: Implications for an evolutionary the-ory of strategy. Strategic Management Journal, 17:193–214.

Carpenter, M., & Fredrickson, J. 2001. Top manage-ment teams, global strategic posture, and the mod-erating role of uncertainty. Academy of Manage-ment Journal, 44: 533–545.

Carpenter, M., Sanders, W. G., & Gregersen, H. B. 2001.Bundling human capital with organizational con-text: The impact of international assignment experi-ence on multinational firm performance and pay.Academy of Management Journal, 44: 493–511.

Chang, S.-J. 1995. International expansion strategy of Jap-anese firms: Capability building through sequentialentry. Academy of Management Journal, 38: 383–407.

Chuang, Y.-T., & Baum, J. A. C. 2003. It’s all in the name:Failure-induced learning by multiunit chains. Ad-ministrative Science Quarterly, 48: 33–59.

Cook, T. D., & Campbell, D. T. 1979. Quasi-experimen-tation: Design and analysis. Boston: Houghton Mif-flin.

Cyert, R. M., & March, J. G. 1963. A behavioral theory ofthe firm. Englewood Cliffs, NJ: Prentice Hall.

Daily, C. M., Certo, S. T., & Dalton, D. R. 2000. Interna-tional experience in the executive suite: The path toprosperity? Strategic Management Journal, 21:515–523.

Darr, E., Argote, L., & Epple, D. 1995. The acquisition,transfer, and depreciation of knowledge in serviceorganizations: Productivity in franchises. Manage-ment Science, 41: 1750–1762.

Davis, J. P., & Eisenhardt, K. M. 2011. Rotating leadershipand collaborative innovation: Recombination pro-cesses in symbiotic relationships. AdministrativeScience Quarterly, 54: 413–452.

Davis, J. P., Eisenhardt, K. M., & Bingham, C. B. 2009.Optimal structure, market dynamism, and the strat-egy of simple rules. Administrative Science Quar-terly, 56: 159–201.

Delios, A., & Beamish, P. W. 2001. Survival and profit-

ability: The roles of experience and intangible assetsin foreign subsidary performance. Academy of Man-agement Journal, 44: 1028–1038.

Denrell, J. 2003. Vicarious learning, undersampling offailure, and the myths of management. OrganizationScience, 14: 227.

Dunning, J. H. 1980. Toward an eclectic theory of inter-national production: Some empirical tests. Journalof International Business Studies, 11: 9–31.

Dyer, J. & Nobeoka, K. 2000. Creating and managing ahigh-performance knowledge-sharing network: TheToyota case. Strategic Management Journal, 21:345–367.

Eisenhardt, K. M. 1989. Building theories from casestudy research. Academy of Management Review,14: 532–550.

Eisenhardt, K. M., Furr, N. R., & Bingham, C. B. 2010.Microfoundations of performance: Balancing effi-ciency and flexibility in dynamic environments. Or-ganization Science, 21: 1263–1273.

Eisenhardt, K., & Graebner, M. 2007. Theory buildingfrom cases: Opportunities and challenges. Academyof Management Journal, 50: 25–32.

Geringer, J. M., & Herbert, L. 1990. Measuring perfor-mance of international joint ventures. Journal ofInternational Business Studies, 22: 249–264.

Graebner, M. 2004. Momentum and serendipity: Howacquired leaders create value in the integration oftechnology firms. Strategic Management Journal,25: 751–777.

Greve, H. R. 1998. Managerial cognition and the mimeticadoption of market positions: What you see is whatyou do. Strategic Management Journal, 19: 967–988.

Greve, H. R. 2003. Organizational learning from per-formance feedback: A behavioral perspective oninnovation and change. Cambridge, U.K.: Cam-bridge University Press.

Haleblian, J., & Finkelstein, S. 1999. The influence oforganizational acquisition experience on acquisitionperformance: A behavioral learning perspective. Ad-ministrative Science Quarterly, 44: 29–56.

Haunschild, P. R., & Miner, A. S. 1997. Modes of interor-ganizational imitation: The effects of outcome sa-lience and uncertainty. Administrative ScienceQuarterly, 42: 472–500.

Haunschild, P. R., & Sullivan, B. N. 2002. Learning fromcomplexity: Effects of prior accidents and incidentson airlines’ learning. Administrative Science Quar-terly, 47: 609–643.

Hayward, M. L. A. 2002. When do firms learn from theiracquisition experience? Evidence from 1990-1995.Strategic Management Journal, 23: 21–39.


Hayward, M. L. A., Shepherd, D. A., & Griffin, D. 2006. Ahubris theory of entrepreneurship. ManagementScience, 52: 160–172.

Helfat, C. E., & Lieberman, M. B. 2002. The birth ofcapabilities: Market entry and the importance of pre-history. Industrial and Corporate Change, 11: 725–760.

Henisz, W. J., & Delios, A. 2001. Uncertainty, imitation,and plant location: Japanese multinational corpora-tions, 1990-1996. Administrative Science Quar-terly, 46: 443–475.

Hoang, H., & Rothaermel, F. T. 2005. The effect of generaland partner-specific alliance experience on jointR&D project performance. Academy of Manage-ment Journal, 48: 332–345.

Hofstede, G. 1980. Culture’s consequences. BeverleyHills, CA: Sage.

Huber, G. P. 1991. Organizational learning: The contrib-uting processes and the literatures. OrganizationScience, 2: 88–115.

Ingram, P. 2002. Interorganizational learning. In J. A. C.Baum (Ed.), Companion to organizations: 642–663.Malden, MA: Blackwell.

Jick, T. D. 1979. Mixing qualitative and quantitativemethods: Triangulation in action. AdministrativeScience Quarterly, 24: 602–611.

Kalnins, A., Swaminathan, A., & Mitchell, W. 2006.Turnover events, vicarious information, and the re-duced likelihood of outlet-level exit among smallmultiunit organizations. Organization Science, 17:118–131.

Katila, R., & Ahuja, G. 2002. Something old, somethingnew: A longitudinal study of search behavior andnew product introduction. Academy of Manage-ment Journal, 45: 1183–1194.

Kim, J.-Y., Miner, A., & S.. 2007. Vicarious learning fromthe failures and near-failures of others: Evidencefrom the U.S. Commercial banking industry. Acad-emy of Management Journal, 50: 687–714.

Kraatz, M., & S.. 1998. Learning by association? Interor-ganizational networks and adaptation to environ-mental change. Academy of Management Journal,41: 621.

Lane, P. J., & Lubatkin, M. 1998. Relative absorptivecapacity and interorganizational learning. StrategicManagement Journal, 19: 461–477.

Langley, A. 1989. In search of rationality: The pur-poses behind the use of formal analysis in organ-izations. Administrative Science Quarterly, 34:598–631.

Langley, A. 1999. Strategies for theorizing from processdata. Academy of Management Review, 24: 691–710.

Langley, A. 2007. Process thinking in strategic organiza-tion. Strategic Organization, 5: 271–282.

Leavitt, B., & March, J. G. 1988. Organizational learning.American Review of Sociology, 14: 319–340.

Lee, T. W. 1999. Using qualitative methods in organi-zational research. Thousand Oaks, CA: Sage.

Leonard-Barton, D. 1992. Core capabilities and core ri-gidities: A paradox in managing new product devel-opment. Strategic Management Journal, 13: 111–125.

Levin, D. Z. 2000. Organizational learning and the trans-fer of knowledge: An investigation of quality im-provement. Organization Science, 11: 630–647.

Levinthal, D. A., & March, J. G. 1993. The myopia oflearning. Strategic Management Journal, 14: 95–112.

Levitt, B., & March, J. G. 1988. Organizational learning.American Review of Sociology, 14: 319–340.

Lieberman, M. B. 1984. The learning curve and pricing inthe chemical processing industries. Rand Journal ofEconomics, 24: 441–452.

March, J. G. 1991. Exploration and exploitation in or-ganizational learning. Organization Science, 2:71–87.

Miles, M. B., & Huberman, A. M. 1994. Qualitative dataanalysis (2nd ed.). Thousand Oaks, CA: Sage.

Miner, A. S., Bassoff, P., & Moorman, C. 2001. Organiza-tional improvisation and learning: A field study. Ad-ministrative Science Quarterly, 46: 304–337.

Miner, A. S., & Moorman, C. 1998. Organizational impro-visation and organizational memory. Academy ofManagement Journal, 23: 698–723.

Pisano, G. P. 1994. Knowledge, integration, and the locusof learning: An empirical analysis of process devel-opment. Strategic Management Journal, 15: 85–100.

Rindova, V., Ferrier, W. J., & Wiltbank, R. 2010. Valuefrom gestalt: How sequences of competitive actionscreate advantage for firms in nascent markets. Stra-tegic Management Journal, 31: 1474–1497.

Root, F. R. 1994. Entry strategies for international mar-kets. San Francisco: Jossey-Bass.

Sapienza, H., Autio, E., George, G., & Zahra, S. 2006. Acapabilities perspective on the effects of early inter-nationalization on firm survival and growth. Acad-emy of Management Review, 31: 914–933.

Schilling, M. A., Vidal, P., Ployhart, R. E., & Marangoni,A. 2003. Learning by doing something else: Varia-tion, relatedness, and the learning curve. Manage-ment Science, 49: 39–56.

Schwab, A. 2007. Incremental organizational learning


from multilevel information sources: Evidence fromcross-level interactions. Organization Science, 18:233–251.

Shaver, J. M., Mitchell, W., & Yeung, B. 1997. The effect ofown-firm and other-firm experience on foreign directinvestment survival in the United States, 1987-92.Strategic Management Journal, 18: 811–824.

Srinivasan, R., Haunschild, P., & Grewal, R. 2007. Vicar-ious learning in new product introductions in theearly years of a converging market. ManagementScience, 53: 16.

Thomke, S. 2003. Experimentation matters. Boston:Harvard Business School.

Tsang, E. W. 2002. Acquiring knowledge by foreign part-ners from international joint ventures in a transitioneconomy: Learning-by-doing and learning myopia.Strategic Management Journal, 23: 835–854.

Van De Ven, A. H. 1992. Suggestions for studying strat-egy process: A research note. Strategic Manage-ment Journal, 13: 169–188.

Van de Ven, A. H., & Polley, D. E. 1992. Learning whileinnovating. Organization Science, 3: 92–116.

Van de Ven, A. H., & Poole, M. S. 1995. Explainingdevelopment and change in organizations. Academyof Management Review, 20: 510–540.

Vermeulen, F., & Barkema, H. G. 2001. Learning throughacquisitions. Academy of Management Journal, 44:457–476.

Weick, K. 1998. Improvisation as a mindset for organiza-tional analysis. Organization Science, 9: 543–555.

Weiner, B. 1985. An attributional theory of achievementmotivation and emotion. Psychological Review, 92:548–573.

Yin, R. K. 1994. Case study research: Design and meth-ods (2nd ed.). Thousand Oaks, CA: Sage.

Yin, R. K., Bateman, P. G., & Moore, G. B. 1983. Casestudies and organizational innovation: Strength-ening the connection. Washington, DC: Cosmos.

Zaheer, S., & Mosakowski, E. 1997. The dynamics of theliability of foreignness: A global study of survival infinancial services. Strategic Management Journal,18: 439–464.

Zahra, S. A., & Dess, G. G. 2001. Entrepreneurship as afield of research: Encouraging dialogue and debate.Academy of Management Review, 26: 8–10.

Zahra, S. A., & George, G. 2002. Absorptive capacity: Areview, reconceptualization, and extension. Acad-emy of Management Review, 27: 185–203.

Zahra, S. A., Ireland, R. D., & Hitt, M. A. 2000. Interna-tional expansion by new venture firms: Internationaldiversity, mode of market entry, technological learn-ing, and performance. Academy of ManagementJournal, 43: 925–950.

Christopher B. Bingham ([email protected]) is an assis-tant professor and Philip Hettleman Fellow of strategy andentrepreneurship at the Kenan-Flagler Business School,University of North Carolina at Chapel Hill. He received hisPh.D. from Stanford University with a focus on strategy,organizations, and entrepreneurship. His research interestsfocus on learning, capability development, and strategy innew ventures and firms in unpredictable environments aswell as psychological foundations of organizations.

Jason P. Davis ([email protected]) is an assistant professorof strategy at the Sloan School of Management at theMassachusetts Institute of Technology. He received hisPh.D. from Stanford University with a focus on strategyand organizations. His research interests focus on thestructures and processes that enable organizations toadapt and on how organizations develop technologicalinnovations in dynamic and interdependent environ-ments such as the computer industry.

This article continues with appendixes.


APPENDIX ASeeding Learning Sequencesa, b

Firm Country 1 Country 2 Country 3 Country 4

Ryti V¡TV. Saw from other firms in the industrythat trials were key and easy to get inSweden, so leaders started to get trialsthere too.T. During project implementation withSwedish customer (action) the customerbecame frustrated with poorcommunication (outcome). Leadersworked to improve corporatecommunication with customers throughcompany intranet and dedicated emaillists (change in behavior).

V¡A¡T¡E¡ IV. Watched where competitor firms werelocated in the U.S. and then set up anoffice in that location (Cambridge, MA).A. FinnPro helped establish an office, screenthe market. and create introductions tolawyers, venture capitalists, and banks.T. Leaders established an office andcontacted customers (action) but were toldthat their image wasn’t sophisticated enough(outcome). As a result, leaders hired a PRfirm and marketing agency and moved to anicer office (change in behavior).E. Leaders intentionally tried different salesapproaches (mailing, direct sales, e-mailshots, trade events) to determine which wasmost appropriate. Direct sales was best andso used more (change).I. In negotiations with customer, leadersdiscovered that the standard front-endpayment schedule wasn’t accepted and sothey altered it on the fly to be back-endloaded (change in behavior).

A¡T¡EA. Relied on FinnPro for advice aboutwhere to establish an Eastern Europeanoffice. FinnPro said that muchpharmaceutical activity was occurring inthe Czech Republic and so leaders enteredthere.T. Had interest from potential customersin the Czech Republic (action), but fewhad money for solutions (outcome).Leaders changed qualification questions touncover the ability to pay early indiscussions (change in behavior).E. Tested many European markets to finda beachhead. Leaders saw that Germanyhad lots of pharmaceutical firmheadquarters and so decided to move toGermany (change in behavior).

A¡D¡T¡EA. Relied on FinnPro for basic market data.D. Management had always sent a Finn toopen a new country who spoke thelanguage. This time they did not, and itcaused communication problems. Sendinga non-German- speaking sales lead clarifiedthe importance of sending in a Finn whospoke the local language (change incognition).T. Tried to do business with Germanpharmaceutical firms but realized thatcustomers were risk averse (outcome).Leaders switched to German contracts tooffset fears of doing business with a non-German firm (change in behavior).E. Leaders deliberately decided not tocreate a legal entity for the German venturebut to try using a satellite office. Theyfound that the satellite office reducedcomplexity and seemed to be a bettersolution for new country entries.

Jackson A¡TA. Learned from consultant that firmsneed to work with multiple distributors inChina as the market is so big and diverseacross regions. Leaders began workingwith multiple distributors (change inbehavior)T. When trying to do business withChinese (action), leaders learned that thatChinese didn’t want to buysemiconductors (outcome); they onlywanted turnkey solutions. Leaders beganproviding turnkey solutions (change inbehavior).

A¡T¡D¡IA. Consultant provided insight about theright firms to target when entering Taiwan.Leaders targeted these firms when entering(change in behavior)T. In discussions with Taiwanese firms(action), leaders found out that their CEOsoften were not the final firm decisionmakers (outcome); rather it was someone ata lower level. Leaders changed salesapproach to appeal to lower level too(change in behavior)D. Moved away from using indirect salesthrough distributors (as in first countryentered, China) to focus on direct sales forentry into Taiwan. Realized in Taiwan theneed to refocus on indirect sales in futuremarkets to improve speed to market andbridge lack of local understanding (change incognition).I. Improvised product during delivery toTaiwanese customer that just wanteddevelopment boards with documentation andsupport instead of a complete turnkeysolution (change in behavior). Firm leadersoffered this no frills solution in latercountries.

A¡T¡IA. Consultant told firm about whichKorean distributors to work with. Leaderspursued these distributors (change inbehavior).T. During discussions with potentialcustomers (action), leaders discoveredKorean firms wanted to start design rightaway and not wait (outcome). Leaderexpedited their design phase (change inbehavior)I. In the middle of sales pitch to aprominent Korean firm, the VPimprovised to more strongly emphasizefeatures after Korean customer appearedless concerned with price and moreconcerned with features. Leaders relied onthe new sales pitch in later meetings withcustomers.

A¡T¡A¡IA. Spoke to prototypical large Japanesefirms to gain insight about demands inJapan. Leaders learned that in Japanreliability is the first concern and price isthe second.T. Leaders started using local distributorsafter entering (action). This wasn’t effective(outcome) and so they started using large,global implementation partners (change inbehavior)A. Japanese implementation partners helpedleaders better understand the factors localfirms would consider (e.g., ISO 9000certification, escrow accounts). Leadersadjusted their sales timeline (change incognition) because they saw it would takelonger than anticipated to close deals.I. Improvised projection figures in meetingafter learning that firms were interested inprice of chips at time of shipping, notcurrent pricing. Learned to better clarifyexpectations.

Wee V¡TV. Focused on information securitysoftware by watching what a prominentfirm in Singapore did in the physicalworld (change in behavior).T. Leaders used sales approach ofapproaching IT leaders in customer organi-zations (action). This did not work out, asmany firms had transferred responsibilityfor IT security out of IT and into audit(outcome). Leaders began targeting auditgroups instead of IT groups (change inbehavior).

V¡T¡I¡DV. Decided to promote solutions on 24/7services by watching a few firms in theU.S. and Europe (change in behavior).T. Entered Malaysia promoting 24/7security monitoring solutions (action). Butthe limited IT infrastructure made firmsreluctant to make purchases (outcome).Leaders backward-integrated into securityinfrastructure (change in behavior).I. Inability to get government accountcaused leaders to improvise a newconceptualization of their target customeras “large firms with proprietary data andthat ability to pay” so that they couldquickly capture an emergent opportunitywith an insurance firm.d: Entered Malaysia without funding ofjoint venture (JV) partners (had funding infirst country entered, Hong Kong). Leaderslearned they needed JVs to provideresources for in country growth and sodecided to get a JV partner and use one forfuture entries (change in behavior).

A¡T¡IA. Relied on local partner in Japan to helplearn what to change in their product forthe Japanese market (change in behavior)T. Leaders tried to promote sales but didnot have the track record in Japan toestablish legitimacy and so customers feltless comfortable about outsourcing theirIT security (outcome). Leaders began torely more on local partner for selling(change in behavior).I. Leaders improvised, changing theirconceptualization of their target customerto capture an emergent opportunity with alarge manufacturing firm (change incognition). Learned a broader view oftarget customers.

V¡T¡V¡IV. Entry of competitor firms into largemarket of China helped persuade firm to dothe same (change in behavior).T. CEO created an alliance with a localpartner to help promote sales (action). Butfew sales were achieved (outcome), andleaders realized that because of the size ofthe country, having a cross-regional partnerwas ineffective. So they decided to startusing multiple partners for differentgeographies (change in behavior).V. Leaders saw other foreign firms exitingChina during the SARS epidemic, whensales plummeted. They followed suit(change in behavior).I. When packaging its software for acustomer, leaders heard that the customerwould only pay for hardware, so engineersdecided to bundle its software in a physicalbox (change in action). Used physical boxoption in later entries.

Kallio A¡T¡IA. FinnPro provided market knowledge andcontacts for potential customers in Italy.T. After implementing pilot in Italy(action), where infrastructure was poor(outcome), leaders decided to do futurepilots in host countries (change in action).I. Engineers discovered they needed todevelop ten more features than what waspiloted. Leaders learned to only offerfeatures that had previously been tested(change in cognition).

A¡T¡EA. FinnPro provided market knowledgeabout the potential customers in the country.Leaders pursued these when entering.T. Project leader negotiated features withcustomer without R&D input (action). R&Dresisted making changes (outcome), and theleader learned to not promise a feature notalready developed and tested (change inbehavior).E. Leaders tried different negotiation tactics(i.e., with and without a senior member).Having the senior member provided morecredibility to customers and so leadersstarted bringing him to future meetingswith client.

A¡T¡E¡VA. Relied on documentation ofimplementation partner.T. Working with a partner created customercommunication problems (outcome).Leaders started communicating moretransparently and actively (change inbehavior).E. Tried a new way to manage customeraccounts. Compared to old method, thenew method helped protect R&D fromcustomer complaints and so wascontinued in each new entry.V. Firm compared itself with U.S.competitor start-ups expanding abroad.Leaders learned they had accomplishedthe same thing with six times lessfunding.

A¡D¡T¡VA. Spoke to competitors such as Wistowand Seven to understand how they createdthe same value proposition with differentapproaches.D. Moved away from focus on Europe toexplore merger with a U.S. firm. Leadersreceived low valuations and so decided torefocus on Europe. They realized that the realreason for remaining focused on Europe wasto master the continent (change in cognition).T. When trying to buy a U.K. firm, leadersmet resistance from Swedish investors(outcome). Leaders realized they needed tospend more time convincing constituents(change in cognition).V. Leaders imitated Blackberry’s approachbut in an open European market (change inbehavior).

a V � “vicarious learning,” T � “trial-and-error learning,” A � “learning from external advice of others (via contact),” I � “improvisational learning,” E� “experimental learning,” D � “deviance-error learning.”

b We note “action,” “outcome,” and other terms in parentheses to denote the learning process.


APPENDIX BSoloing Learning Sequencesa, b


Tyler E¡TE. Entered Sweden as it wasa country with nationalizedmedicine, and firm leadersbelieved doctors in thesecountries would be morereceptive to technology(controlled situation to testproposition). After trying toget trials started (action),leaders saw skepticism fortheir technology (outcome)and so decided to bettereducate local doctors(change in behavior).T. Tried to get doctors whowere excited about thetechnology to create apurchase order (action), butthey didn’t want to buy(outcome). Discovered thatdoctors often didn’t makedecisions. Rather agovernment official madedecisions, and so leadersneeded to solicit them(change in cognition).

TT. Entered Netherlandsbecause leaders knew a fewprominent radiologists.After demonstrating analogtechnology with radiologists(action), leaders saw thatlocal firms were notinterested in their analogproduct (outcome). Instead,leaders discovered localradiologists wanted a digitalCAD solution (change incognition). Because theydidn’t have a digital CADsolution, leaders decided topush their analog productmore aggressively (changein behavior).T. Managers tried to betterpromote analog product tolocal doctors (action) butwere told that their productwas too big for the mobileunits many doctors wereusing (outcome). Decidedthat local medical systempromoted efficiency overeffectiveness (change incognition).

E¡TE. Entered Germany becauseit was believed to be a bigmarket, with nationalizedmedicine and manyprominent doctors withmoney to pay fortechnology (controlledsituation in which to testproposition). Leaders foundthat though the market wasbig, doctors did not want touse clinical trial data fromother countries (outcome).This helped leaders realizethat the sales process wouldtake much longer thananticipated (change incognition).T. Had local doctors beginusing technology to conducttrials (action). Trials seemedto suggest benefits, but fewdoctors were willing tomake purchases (outcome).Managers saw that manyGerman doctors waited tofollow the actions of theregion’s most prominentdoctors (change incognition).

TT. After entering Japan,leaders discovered thatdoctors just wanted print-outs, not machines(outcome). They learnedtheir product was too big(change in cognition).T. Leaders used a localdistributor to promotetechnology (action), but fewdoctors respondedpositively (outcome). Aftertalking with a few doctors,leaders realized that manydoctors felt distributorswere corrupt and so did notlike to work with them(change in cognition).Leaders began to look for aconsulting company thatcould perform the same role(change in behavior).

Adams E¡TE. Australia was seen as atest bed in which leaderscould learn to dointernational business.Leaders gave manager lotsof autonomy (action) butfound that this resulted in adisconnect from corporate(outcome), so leaders addedoversight (change inbehavior).T. Corporate leaders pushedlocal country manager touse a features and functionsselling approach (action).After little success(outcome), the countrymanager persuadedcorporate to adopt a moresolutions/consultativeselling approach (change inbehavior).

E¡TE. Entered U.K. with a verysmall team to gauge marketinterest. Leaders sawinterest and so entered.T. Local country managerworked to leverage existingrelationships from the U.S.to U.K. (action), but theyrealized that relationshipsdidn’t translate to Europe(outcome). This causedleaders to see they needed agood implementationpartner to build valueproposition. They thushired PWC (change incognition and thenbehavior).

E¡TE. Entered France to see ifEuropean expansion waspossible. Leaders realizedthey were early to themarket (outcome). As aresult, corporate decided toslow entry into France(change in behavior).T. Leaders signed acustomer in France on thebasis of stating that theywould be putting a team ofFrench nationals in place infairly short order (action).However, the Americaneconomy and the businesswere starting to slow downglobally, and so there wasnot enough resources toexpand (outcome). Thus,corporate decided toconcentrate on largermarkets (change inbehavior).

TT. Leaders entered countryafter a German customerexpressed interest (action).They found it hard to closedeals, given lack ofresources from the U.S.corporate office (outcome),and so decided to pull outof the country and focus onbuilding up major marketsin the U.K. and U.S.(change in behavior).

ContinuedContinued


APPENDIX B(Continued)


Stahlberg I¡TI. Leaders entered Swedento do business with a bigcustomer, but duringcontract negotiation, thecustomer wanted more thaninitially agreed upon. Thecountry manager thus hadto improvise a noveldiscount policy. Leaderslearned that big referenceaccounts required largerdiscounts than they hadexpected.T. After signing, thecustomer complained aboutlocal support. Dealing withthe challenge helpedleaders see they neededmore of an infrastructure inSweden (change inbehavior).

I¡TI. When country managerfound out that Germancustomer wanted more thana license deal, he created anovel solution sale as well.Firm learned another wayto promote its products.T. Leaders found thatrunning the operation fromFinland proved difficult toclose subsequent deals(outcome). Leaders realizedthat they should haveconducted more up-frontresearch and have morelocal country contacts. Theydecided to hire a localcountry manager (change incognition and behavior).

TT. Leaders entered the U.S.and, since the market wasperceived to be so big, theymoved many functionsthere, including marketingand sales (action). However,the leaders found thatmoving too many functionstoo soon made the firm tooU.S. focused (outcome). HQmanagement decided tomove some functions backto Finland, where it couldhave greater control (changein behavior).

TT: Leaders began with acommission-based profitand loss statement (action).But the commission-basedsystem did not work well(outcome), and so leadersswitched to transfer pricing(change in behavior).

Polk E¡TE. Experimented afterentering China by sending atrial version of the softwareaggressively everywhere tosee which companies wouldbe most interested.Discovered that systemsintegrators appearedparticularly attractive.T. Used a sales approachwith potential customersthat relied on using U.S.references (action). Chinesefirms did not readily acceptthose references (outcome).Leaders came to understandthat China-based referenceswere more important(change in cognition).

I¡TI. When trying to create aunique product solution forCompaq in Germany,engineers uncovered avaluable product feature themanagement didn’t knowexisted (change in protocolbased on location). Thisfeature was then promotedas a key feature (newactivity).T. During the project forCompaq (action), projectmanagement difficultiesassociated with culturaldifferences sometimes arose(outcome). HQ realized thatthe firm should have putpeople on the ground in thecountry with knowledge ofthe culture and language(change in cognition).

I¡TI. Had to adjust in real timeseveral features of its “proofof concept” for firstcustomer.T. While trying to completecustomer solution inSwitzerland, the firm reliedon a “fly in” model andsent people over toSwitzerland for periods oftime to work with customer(action). Leaders at HQbegan to see that this modelof doing sales and supportfrom the U.S. was veryinefficient (outcome). Theydecided that countriesentered should have localoffices and countrymanagers and so they hiredlocal people (change inbehavior).

TT. Hired local sales managerand established an office inthe U.K. to generate localsales (action). However,sales were much slowerthan expected (outcome).HQ leaders realized thattheir marketing messagewas too technical and didnot say enough about abusiness solution. As aresult, they altered theirsales approach to make itmore understandable(change in behavior).

ContinuedContinued


APPENDIX B(Continued)


Nair E¡T¡EE. India seen as a test bedfor developing newsolutions, as developmentcosts were much lower thanin other countries.T. Worked hard to closedeals (action) but had fewsales (outcome). Leadersrealized that local firmswanted “Indian pricing”and so leaders had to givebigger discounts (change inaction).E. Experimented withpromoting differentproducts (i.e., trialsolutions, educationalproducts, high-throughputscreening) to see whichwould be in most demand.High-throughput screeningseemed best and so wasemphasized.

E¡T¡ E¡AE. “Threw out seeds to seewhich [country] wouldgerminate fastest.” Firms inJapan seemed interested, sothe firm entered thatcountry with a partner.T. Leaders created acontract with a firm (action)but drastically overpricedtheir solution (outcome),and so learned to notoverprice (change inaction).E. Experimented sellingboth an integrated solutionand component parts ofsolutions. Japanesecustomers liked thecomponent parts more andso leaders began to offermore modular solutions(change in behavior).A. Learned from partnerhow to better negotiate withclients: “What are thethings they fight for andwhat are the things theyleave alone.”

T¡AI. Leaders entered Australiaand began working withIBM to showcase solutionsat conferences andworkshops (action). Afterlittle success (outcome),leaders realized that toomuch “educating” waswasting time and should besignificantly cut back(change in behavior).A. From Australianacademics, leaders gainedinformation on marketopportunities in Australia—specifically, which projectsthe government wasfunding. This impacted howthe firm positioned itsproducts (change inbehavior).

T¡AT. After entering Malaysia,the firm tried to promote itsbioinformatics solutions toseveral promising customers(action), but firms in theMalaysian market were notthat familiar withbioinformatics and so werereluctant to purchasesolutions (outcome). Thiscaused leaders to spendmore time doing“roadshows” to helpeducate others (change inbehavior).A. Lack of sales pushedleaders to rely more on alocal partner to do“relationship management”so firm could gain access totop accounts (change inaction).

a V � “vicarious learning,” T � “trial-and-error learning,” A � “learning from external advice of others (via contact),” I � “improvi-sational learning,” E � “experimental learning,” D � “deviance-error learning.”

b We note “action,” “outcome,” and other terms in parentheses to denote the learning process.


Documents

LEARNING SEQUENCES: THEIR EXISTENCE, EFFECT, AND …jasond/learningsequences.pdfscribed for the unit. Experimental learning is another direct learning process. Experimental learning