simultaneous experimentation as a … experimentation as a learning strategy: business model...

SIMULTANEOUS EXPERIMENTATION AS ALEARNING STRATEGY: BUSINESS MODELDEVELOPMENT UNDER UNCERTAINTY

PETRA ANDRIES*, KOENRAAD DEBACKERE, and BART VAN LOOYCentre for R&D Monitoring, Department of Managerial Economics, Strategy,and Innovation, Katholieke Universiteit Leuven, Leuven, Belgium

Ventures operating under uncertainty face challenges defining a sustainable value proposition.Six longitudinal case studies reveal two approaches to business model development: focusedcommitment and simultaneous experimentation. While focused commitment positively affectsinitial growth, this commitment and lack of variety jeopardize long-term survival. Simultaneousexperimentation implies lower initial growth levels, but facilitates long-term survival byenacting variety in a resource-effective manner. This article enriches organizational learningtheory by demonstrating that not only distant search but also simultaneous experimentationresults in variety. Moreover, simultaneous experimentation implies effectual behavior andreconciles the apparent juxtaposition between ‘action’ and ‘planning.’ Copyright © 2013Strategic Management Society.

INTRODUCTION

Successful exploitation of an entrepreneurial oppor-tunity requires translating that opportunity into aviable business model (Amit and Zott, 2001).However, for many ventures, this is not straightfor-ward, given the considerable levels of uncertaintywith which they are confronted, both in terms oftechnical and market feasibility (Anderson andTushman, 1990). This is especially true in emergentindustries where what the market will becomedepends on multiple decisions by various stakehold-ers, and clarity will only arise when entrepreneurialactivities result in the industry’s development (Duttaand Crossan, 2005). As a consequence, the set offeasible opportunities and viable business models isoften not predictable in advance (Alvarez andBarney, 2007).

Recent work (e.g., Gruber, MacMillan, andThompson, 2008; Andries and Debackere, 2007)concludes that experimenting with and redefining,business models—based on consumer and marketreactions—is highly instrumental under such uncer-tain circumstances. According to Sarasvathy (2001,2008), traditional planning will not suffice inrapidly changing and uncertain environments.Instead, an effectual logic, in which entrepreneursproceed with available resources to envisage arange of possible business opportunities and try toimplement some of these opportunities throughpartnerships, is more suitable for leveraging unex-pected events and feedback. If the outcomes of thisexperimentation process are negative, the initialbusiness model is redefined and a new experimentis launched (Minniti and Bygrave, 2001; Vohora,Wright, and Lockett, 2004). Hence, experimentingwith a variety of business models appears crucialunder uncertainty.

However, current research offers little insight intowhat these business model experimentation pro-cesses actually look like and how they can be orga-nized effectively. Organizational learning theory

Keywords: business models; uncertainty; organizational learn-ing; experimentation; variety*Correspondence to: Petra Andries, Centre for R&D Monitor-ing, Katholieke Universiteit Leuven, Waaistraat 6, 3000Leuven, Belgium. E-mail: [email protected]

bs_bs_banner

Strategic Entrepreneurship JournalStrat. Entrepreneurship J., 7: 288–310 (2013)

Published online in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/sej.1170

Copyright © 2013 Strategic Management Society

suggests that different approaches to experimenta-tion exist, but these have not been studied in thecontext of new ventures’ business model develop-ment. This article, therefore, seeks to investigatehow ventures develop their business models throughexperimentation. In particular, it intends to study(1) whether different experimentation and learningapproaches exist and, if so, (2) what the rationale andimplications of these approaches are.

Addressing these questions calls for fine-grainedinsights into the venture processes unfolding overtime (Yin, 1994). Hence, we opt for a longitudinalcase study design. We analyze in-depth six venturesactive in various industries, all of them faced withconsiderable levels of market and technologicaluncertainty and displaying variety in terms of busi-ness model development. Relying on rich dataobtained from 28 interviews, 17 business plans, 75press articles, and 250 pages of other internalcompany documents, we analyze how these six ven-tures develop their business models over time.

Our study makes several contributions. Besidesfocused commitment to one single business model,our cases reveal a previously undocumented entre-preneurial learning strategy labeled ‘simultaneousexperimentation,’ entailing the cost effective pursuitof a portfolio of business model experiments. Ana-lyzing the rationale and implications of both learningstrategies shows that, while focused commitmentis initially instrumental in acquiring dedicatedresources, it reduces the variety of business modelexperiments being pursued and complicates thefinancing and execution of subsequent compulsorysearches, thereby jeopardizing long-term survival.Simultaneous experimentation, implying a portfolioof related but diverging searches, facilitates long-term survival by enacting variety in a resource effec-tive manner.

Our research extends organizational learningtheory by demonstrating that not only distant searchbut also simultaneous experimentation leads tovariety in terms of business models. In addition, itsuggests that simultaneous experimentation can beorganized in a resource effective way. Due to thiscombination of resource efficiency and variety,simultaneous experimentation presents itself as arelevant learning strategy for entrepreneurial ven-tures alongside focused commitment and previouslydocumented bootstrapping approaches.

The identification of a simultaneous experimenta-tion approach adds to the effectuation literature sinceit is one of the first illustrations of how an effectual

logic translates into effectual behavior. As thisapproach combines experimentation with carefulselection, deliberate planning, and efficient execu-tion of specific experiments, it also reconciles theapparent juxtaposition between ‘action’ and ‘plan-ning.’ While planning has traditionally been con-ceived as implying the ex ante selection of onespecific business model, our findings reveal thatsimultaneous experimentation implies deliberateplanning and selection, as well as effectual actionand experimentation.

THEORETICAL BACKGROUND

As a conceptual framework for studying the businessmodel development of ventures, we rely on two mainliterature streams: the literature on business modelsand their components on the one hand and organiza-tional learning theory on the other. The businessmodel literature formed the initial basis for thisresearch, while the relevance of organizational learn-ing theory emerged from the case study analyses (seerecommendations by Suddaby, 2006). For reasons ofclarity, both are discussed up front.

The business model and its components

Since the work of Amit and Zott (2001), researchershave defined the business model concept in a ratherbroad way, referring to a large variety of firm char-acteristics and decision variables, which translateentrepreneurial opportunities into particular configu-rations that create and capture value. Afuah (2003),for example, defines a business model as ‘the set ofactivities a firm performs, how it performs them, andwhen it performs them as it uses its resources toperform activities, given its industry, to create supe-rior customer value . . . and put itself in a position toappropriate the value’ (Afuah, 2003: 9).

As shown by Zott, Massa, and Amit (2011), thereis a consensus in the literature that business modelscombine a broad variety of components or ‘ingredi-ents,’ reflecting the fact that value creation arisesfrom multiple sources, including: (1) transactionefficiency as a major way of reducing costs; (2) dif-ferentiation raising buyers’ utility; (3) the combina-tion and development of a set of scarce, durable, anddifficult to imitate resources and capabilities; and(4) the density, centrality, size, and governance ofthe company’s strategic networks.

New Ventures’ Business Model Development Under Uncertainty 289

Copyright © 2013 Strategic Management Society Strat. Entrepreneurship J., 7: 288–310 (2013)DOI: 10.1002/sej

According to a detailed literature review byMorris, Schindehutte, and Allen (2005), the mostconsistently emphasized business model compo-nents are: (1) factors related to the offering;(2) market factors; (3) internal capabilities; (4) com-petitive strategy factors; (5) economic factors, orhow the venture makes money; and (6) personal/investor factors (i.e., the venture’s time, scope, andsize ambitions). A business model’s performancedepends not only on the content of these individualcomponents, but also on the fit between them(Hamel, 2000). Just like a recipe, the business modelincludes both the main elements of a firm’s activityand their integration. A change in one componentinteracts with many other components to determinethe value of the business model as a whole. A busi-ness model can, hence, be described as a configura-tion of interdependent business model components.

Developing a business model throughexperimentation: insights from organizationallearning theory

It is now generally accepted that entrepreneurialventures—when operating under uncertainty—should experiment with a range of business models(Gruber et al., 2008; Andries and Debackere, 2007).Through experimentation, the initial value proposi-tion evolves into a viable business model by meansof ‘a series of trial and error changes pursued alongvarious dimensions’ (Nicholls-Nixon, Cooper, andWoo, 2000: 496). By experimenting with a specificbusiness model and incorporating feedback fromthe environment, entrepreneurial ventures adopt anactive stance to learning about the environment. Ifoutcomes are negative, the initial business model isredefined and a new experiment is launched (Minnitiand Bygrave, 2001), implying ventures will deviatefrom their initial business model configurations asthey learn about and incorporate information thatbecomes available during the entrepreneurial trajec-tory (Gruber et al., 2008).

The literature on organizational learning proposestwo main approaches to learning and experimenta-tion under uncertainty. Both approaches differ in theway configurations are adapted over time. On theone hand, organizations can develop throughstepwise, incremental changes. In each step of thelearning process, they alter one single componentof their configuration and verify whether thischange improves performance (Levinthal, 1997).This process is called ‘local search’ or ‘related

search.’As a result of these incremental changes, theset of solutions in the immediate neighborhood ofthe existing configuration is examined (March andSimon, 1958; Cyert and March, 1963) and the orga-nization will evolve toward a configuration thatis closely related to its initial one (Minniti andBygrave, 2001). On the other hand, organizationscan make radical changes to their configurations bysimultaneously altering multiple components oftheir configuration in each learning step. This learn-ing process is called ‘distant search,’ ‘search throughlong jumps’ (Levinthal, 1997), or ‘path-creatingsearch’ (Ahuja and Katila, 2004) because it leadsorganizations to experiment with configurationsthat differ to a large extent from their initialconfigurations.

In line with the literature on organizational learn-ing, it would then appear that ventures can eitherchange their business model configurations throughlocal search, i.e., by changing one single businessmodel component at a time, or through distantsearch, i.e., by altering multiple business modelcomponents in each new experiment. However,which of the two approaches is more suitable forentrepreneurial ventures operating under uncertaintyremains unclear. With this article, we contribute tothe literature by analyzing: (1) how ventures developand redefine business models under uncertainty;(2) why they opt for a specific search approach;and (3) how these choices affect the developmenttrajectory of the venture.

METHODOLOGICAL APPROACH

Case selection

Given our focus on the business model developmentprocess, we have opted for longitudinal case studies(Eisenhardt, 1989), with ventures as the unit ofanalysis. We started by studying a sample of 21 newventures in the framework of a broader researchproject on incubation, venture creation, and venturefinancing. These 21 ventures were initially selectedto reflect industry variety. Since the current contri-bution focuses on different approaches to businessmodel development under uncertainty, we haveretained only the subset of ventures that, at the timeof their launch, were clearly faced with uncertaintyregarding their technology and markets while, at thesame time, displaying variety in terms of businessmodel development. Indeed, as argued by Eisenhardt

290 P. Andries, K. Debackere, and B. Van Looy


and Graebner (2007), including a diverse set of casesis more likely to result in robust insights (for a recentillustration in the field of entrepreneurship, seeClarysse, Bruneel, and Wright, 2011).

This approach resulted in a set of six ventures, allfaced with considerable degrees of uncertainty at thetime of founding, and active in diverse industries,including machine visions systems, software foradvanced laser applications, E-commerce, and soft-ware development related to emerging Internet use(see Table 1 for an overview of the cases). It turnsout that the six cases also vary in terms of success,although this was not a selection criterion. Imageand L-goritm1 identified viable business models,turned breakeven, went public, and eventuallybecame global players in their market niches.@Music and OOPs both failed to identify a viablebusiness model and went bankrupt. At the time ofdata collection, it was not yet clear whether or notSiS’ and Regmed’s efforts would result in sustain-able revenues.

Data collection

For each venture, we have documented the periodfrom the emergence of the idea to establish acompany until: (1) the point in time at which theventure ceases to exist (in our cases, either throughbankruptcy or voluntary closure); (2) the point intime where the company turns breakeven; or (3) forfirms that still exist but have not yet turnedbreakeven, the point in time when the data collectionis terminated (i.e., the first quarter of 2005). Thisimplies that, for some companies, we observe moreyears than for others (see Table 1). Also, one casein our sample—namely Image—was significantlyolder than the other five cases, and we had to go backfurther in time to document the initial businessmodel development.2

A historical description of each company has beencreated, based on the information from semi-structured interviews and document analysis (seeTable 1). Data collection and interviews were con-ducted from the first quarter of 2004 to the firstquarter of 2005. Available documents (including

company Web sites and press articles) were analyzedin preparation for the interviews. Twenty-eight inter-views were conducted (see Table 1), lasting approxi-mately two hours on average, with the shortesttaking about 50 minutes and the longest takingalmost three hours. For each case, we interviewed atleast one founder and one person not part of thefounding team. Wherever possible, we includedexternal views such as those of investors, consul-tants, and technology transfer officers involved. Theinterviewees provided complementary documentsafter the interviews. For each case, interviews anddocument analysis were performed until a consistentcase account could be constructed. When inconsis-tencies between interviews and documents emerged(e.g., in @Music and OOPs), additional interviewswere conducted to clarify pending issues.

In total, four company Web sites, 17 businessplans, 75 press articles (including online articles),and approximately 250 pages of other internal docu-ments have been analyzed. The resulting historicaldescriptions were presented to the interviewees toassess accuracy and completeness. In some cases,information has been added, refined, or corrected.We triangulated the documentation of our centralconstructs and involved key informants in eachventure to review our historical descriptions in orderto increase construct validity (Gibbert, Ruigrok, andWicki, 2008). This analysis results in narrativedescriptions of the entrepreneurial trajectory.

Central constructs: business models andtheir relatedness

When engaging in case study analysis, the use of aset of central constructs is beneficial in documentingphenomena of interest systematically (Eisenhardt,1989). As we aim to analyze ventures’ approaches tobusiness model development, we document andinterpret the business model concept as it changesover time. This construct stems directly from ourresearch questions and is, hence, specified a priori.

Whereas conceptually business models arebroadly conceived as the entirety of interrelatedactivities performed to create and capture value(Afuah, 2003), most empirical research on businessmodels adopts a more narrow measurement of theconcept. As summarized by Zott et al. (2011),researchers often adopt idiosyncratic business modeldescriptions and typologies. While this fits specificresearch questions, it hinders cumulative progress inthe field. In addition, most existing approaches are,

1 The names of the ventures in our case study have beenchanged for reasons of confidentiality.2 We are convinced that—due to the availability of detailedbusiness plans and privileged access to two consecutiveCEOs—retrospective biases for the Image case are comparableto the other five cases.



Table 1. Case study overview

Company Activity Uncertainty at founding Data sources Time period

@Music E-commerce Market: awareness of critical issues,such as impact of Internet onindividuals’ buying behavior,musical genres, geographical scope;however, rate of diffusion unknown;regulation for sales through Internetnot existing

Technology: technical solutions forsales through Internet not existing

Three business plans, 250±pages of internal documents(e-mail correspondence,meeting reports, financialreports), four press articles,11 interviews with threefounders (one wasinterviewed twice), twoinvestors, and five employees

Sept 1998 toJuly 2002

OOPs SW productsthat enableE-commerce

Market: implicit assumption thatB-to-C will be considerable market;unawareness of the distributionapproach, of market interest inB-to-B

Technology: unawareness of thediscrepancy between thetechnological difficulties in adaptingstandardized solution to real settings

Two business plans, three pressarticles, six interviews withone investor, one consultant,one employee, and twofounders (one wasinterviewed twice)

Jan 1998 toFeb 2003

Image Machine visionsystems

Market: difficult to estimate size ofmarket and subsegments; highnumber of applications in differentindustries possible, but unclearwhich one is commerciallyinteresting

Technology: high number ofapplications in different industriespossible, but unclear which one isfeasible from technical point of view

Company Web site, fivebusiness plans, 37 pressarticles, two interviews withtwo former CEOs (of whichone was the founder)

Jan 1983 toOct 1988

L-goritm SW foradvancedlaserapplicationsin productdesign

Market: unawareness of relevance ofsales approach and geographicalscope

Technology: impact of software versushardware technology unknown

Company Web site, threebusiness plans, 21 pressarticles, three interviews withone founder/CEO, onecofounder, and one financemanager

Dec 1995 toDec 2001

SiS Securecommunicationservices

Market: high expectations, but alsohigh uncertainty regarding impact ofInternet on individuals’ buyingbehaviors; uncertainty regardingmarket potential of three potentialapplications; lack of knowledge ofinternational marketing and lack ofknowledge of fierce competition onhome market

Technology: technology platform stillunder development; uncertaintyregarding technical feasibility ofthree potential applications

Company Web site, twobusiness plans, three pressarticles, three interviews withone founder (interviewedtwice), and one technologytransfer officer

May 2000to May2005

RegMed Biomedicalregenerativemedicine

Market: lack of market knowledge;FDA testing procedures forbiomedical products not existing

Technology: uncertain results ofclinical studies

Company Web site, twobusiness plans, seven pressarticles, three interviews withtwo founders, and onetechnology transfer officer

Jan 2000 toJan 2005



to some extent, sector specific, making them lessappropriate for mapping and analyzing the businessmodel evolution process across different industries(Demil and Lecocq, 2010).

Consequently, limited progress has been made inempirically operationalizing the business modelconcept in line with its definition as a broad set ofdecision variables related to the creation of a com-petitive advantage. An important exception is thedetailed coding scheme by Morris et al. (2005) com-prising a wide variety of business model compo-nents. For our empirical analysis, we rely on thiscoding scheme because it has three important advan-tages. First, it represents multiple sources of valuecreation and value capturing and, therefore, fits theconceptual view of the business model as a system-level, holistic approach to explaining how firms dobusiness (Zott et al., 2011). Second, even though thiscoding scheme is very detailed, it can be appliedacross industries. And finally, it allows for docu-menting changes over time.

For each case, we document and analyze changesin the business model components listed by Morriset al. (2005): (1) the (nature of) the offering, includ-ing the distribution approach; (2) the market;(3) internal capabilities; (4) the competitive strategy;(5) economic factors, i.e., the venture’s revenue/margin model; and (6) personal/investor factors, i.e.,the venture’s ambitions. We adapt the list of sub-items provided by Morris et al. (2005) in a way thatallows us to code not only changes in the businessmodel, but also the number of business modelsexperimented with.

Based on the adapted coding scheme in Table 2,we document changes in sub-items for each venture.We regard any observed combination of these sub-items as one specific business model. If a companyexperiments with two or more such combinations atthe same time, we classify these as different businessmodels. For example, if a venture simultaneouslysells a product directly and indirectly (through dis-tribution partners), we regard both activities as twodifferent business models. If a venture evolves fromoffering services only to offering both products andservices, we consider this the introduction of oneadditional business model. A similar logic applies ifthe venture changes from fixed to both fixed andflexible pricing. If a venture switches from fixed toflexible pricing or from business-to-business tobusiness-to-consumer activities, we regard this asthe abandonment of one business model and theadoption of another business model.

In line with the organizational learning literature,a second central construct adopted is the relatednessof a venture’s experiments.3 We analyze whetherventures develop their business models throughlocal or distant business model searches, and wealso map the overall variety of business modelsbeing explored over time. To do so, we compareeach newly introduced business model to otherbusiness model(s) on the venture’s developmentpath. More specifically, we calculate four distancemeasures for each new business model, representingits distance to the venture’s most similar and leastsimilar business models (developed previously, aswell as simultaneously).

First, we compare the new business model withthe most similar business model used up to that pointin time. We count on how many business modelcomponents they differ. This implies comparisonwith (1) business models used both before and afterthat specific point in time, as well as (2) businessmodels that had been used up to that specific point intime but were abandoned afterward. So, if a ventureat a specific point in time is experimenting with threebusiness models and introduces a new businessmodel, we calculate how many components this newbusiness model differs from each of the three previ-ously used business models, irrespective of whetheror not these three are discontinued. We take thelowest value of these three differences as an indi-cation of the new business model’s relatedness toprevious experiments.

Second, we compare the new business modelwith the most similar business model used fromthat point in time onward. We count on how manybusiness model components they differ. Thisimplies comparison with (1) business models thathad been introduced beforehand and were stillmaintained at that specific point in time, as well as(2) business models that were introduced at thatspecific point in time but had not been usedbeforehand. This gives us an indication of the newbusiness model’s relatedness to simultaneouslyconducted experiments.

Third, we perform similar coding to map the dif-ference between each new business model and itsleast related counterpart. More precisely, we count

3 The inclusion of relatedness as a central concept emergedwhen conducting the case study analysis. Thus, our method-ological approach combines longitudinal case methods guidedby central constructs with more ‘grounded’ approaches in linewith recommendations advanced recently by Suddaby (2006).



the number of business model components for whichthe new business model differs from the least similarbusiness model used up to that point in time and,finally, we count the number of components forwhich it differs from the least similar business modelused from that point in time onward. Whereas thefirst two distance measures allow us to assess the useof local and distant search, the latter two are indica-tive of the level of overall business model varietyenacted by the venture.

ANALYSIS AND FINDINGS

Not only does longitudinal case study researchbenefit from the use of central constructs, but italso implies different levels of analysis (Pentland,1999)—from describing the actual process of busi-ness model development within each venture to inter-preting the observed findings by means of the centralconstructs and by situating the findings in relation tothe current literature. Moving from description to

Table 2. Subcomponents of the business model (adapted from Morris et al., 2005)

OfferingHow does the company create value? (select one from each set)■ product / service■ standardized / some customization / high customization■ internal manufacturing or service delivery / outsourcing / licensing / reselling / value-added reselling■ direct distribution / indirect distributionMarketWho does the company create value for? (select one from each set)■ type of customer (b-to-b / b-to-c)■ local / regional / international■ position of customer in the value chain: upstream supplier / downstream supplier / government / institutional /

wholesaler / retailer / service provider / final consumer■ broad market / niche market■ transactional / relationalInternal capabilitiesWhat is the company’s source of competence? (select one or more)■ production / operating systems■ selling / marketing■ information management / mining / packaging■ technology / R&D / creative or innovative capability / intellectual■ financial transactions / arbitrage■ supply chain management■ networking / resource leveragingCompetitive strategyHow does the company competitively position itself? (select one or more)■ image of operational excellence / consistency / speed■ product or service quality / selection / features / availability■ innovation leadership■ low cost / efficiency■ intimate customer relationship / experienceEconomic factorsHow does the company make money? (select one from each set)■ pricing and revenue sources: fixed / flexible■ operating leverage: high / medium / low■ volumes: high / medium / low■ margins: high / medium / lowPersonal / investor factorsWhat are the company’s ambitions? (select one)■ subsistence model / income model / growth model / speculative model



analysis implies an iterative process consisting ofwithin-case analysis and between-case comparison(Eisenhardt, 1989). We analyze business modeldevelopment in each venture in our sample andcompare the development pattern over the six cases.

Based on the historical description constructed foreach case and using the coding scheme in Table 2,we coded whether and when the business models ofthe six ventures in our sample changed. Thesechanges include instances in which new businessmodels were introduced and/or previous businessmodels were abandoned. In a next step, we codedhow many business models were added and droppedin each instance. We then used the coding scheme inTable 2 to map each new business model in detail. Ina final step, we used these detailed mappings tocalculate the distance measures discussed earlier inthe Methodology section.

Coding was performed independently by two ofthe authors, who identified the same 23 changes overall six ventures; i.e., the same 23 points in time atwhich the business model changed. Of these 23events, 19 were initially coded as the exact samenumber of abandoned and adopted business models.Taking into account the different categories usedby the two coders, this implies a Cohen’s kappa of0.59. The differences in coding have been resolvedthrough discussion and by obtaining additionalinformation from the ventures, resulting in consen-sus for all events in a second and final round. In anext step, the two authors independently mappedeach business model in detail and calculated its relat-edness to other business models, immediately reach-ing full agreement. For each case, the coding wasvisualized in (1) a timeline representing the numberof business models added or dropped over time and

(2) a detailed mapping of each new business modeland its relatedness to the venture’s previously andconcurrently developed business models. Figures 1and 2 and Tables 3 and 4 illustrate this for the casestudies of @Music and Image.4 A summary for allcases can be found in Table 5.

A first striking observation is that, for most ven-tures, consecutive and concurring business modelsdiffer on multiple aspects (see median distance mea-sures in Table 5). For all ventures except RegMed,the majority of business models differ on three ormore components from the least-related businessmodel developed previously and also from the least-related business model developed concurrently.These differences indicate that the ventures generatea considerable variety of experiments when search-ing for a viable business model.

However, visual comparison of the timetables ofthe different cases indicates that the six venturesdiffer significantly with respect to how they arrive atcreating variety. Four of them (@Music, OOPs, SiS,and RegMed) focus on one (in the case of RegMed,two) business model(s) very early on and commit tothis decision for several years (see Figure 1 for@Music). Only after a considerable time period do@Music, OOPs, and SiS abandon their initial busi-ness models and launch new business model experi-ments. So, whereas these four ventures focus on andcommit to one specific business model initially, threeof them switch to new business model experimentslater.

4 Due to space limitations, the figures and tables for the fourother cases are not included in this article. They are availablefrom the authors on request.

09/98initial idea

01/99founding

01/00 01/02 07/02@Music files for bankruptcy

01/01

BM1

BM2

BM3

BM4

Figure 1. Event history of @Music



Image’s and L-goritm’s timelines display distinctpatterns (see Figure 2 for Image). They spend two tofive years developing a large portfolio of businessmodel experiments simultaneously, which they thengradually narrow down to one business model. Asindicated by the median distance measures inTable 5, the majority of these business model experi-ments differ with respect to only one or two businessmodel components from the most similar previousexperiments in the portfolio and with respect to onlyone business model component from the mostsimilar concurrent business model experiment.Stated otherwise, newly introduced business modelsare highly related to already existing businessmodels in the portfolio.

Image and L-goritm are, hence, developing port-folios of simultaneous, local, business modelsearches. While these search paths start from thesame experiment initially, they spread out in variousdirections over the performance landscape. So,although individual search paths in the portfolioconsist of related business model experiments, theventures end up experimenting with a set of businessmodels that, in the end, vary significantly. This canbe seen in the median distant measures in Table 5:while a new business model differs, in most cases,only with respect to one component from the mostsimilar business model previously or concurrently

developed, it often differs on three components fromthe least similar business model in the portfolio.

Consequently, we identified two distinct experi-mentation approaches in our sample: (1) focusedcommitment followed by consecutive businessmodel search and (2) simultaneous experimentation.In a next step, we carried out an in-depth analysis ofthese two approaches. More specifically, we revis-ited the historical case descriptions and investigatedwhy each venture opted for a specific approach andwhat the implications, advantages, and disadvan-tages were. We then compared our findings across allcases adhering to the same experimentation strategy.

Focused commitment followed by a period ofbusiness model experimentation

We find that some ventures select one specific busi-ness model very early on and then commit to thisbusiness model for several years. One could considerthis an extreme case of local, path-deepening search(Ahuja and Katila, 2004) in which the venture keepsexperimenting with the exact same model in thehope that it will become viable. Only when, after asignificant period of time, initial assumptions fail tomaterialize, do these ventures begin to experimentwith alternative business models.

01/83founding 01/84 01/85 01/86 01/87 01/88 10/88

BM1 BM2

BM3

BM4

BM5

BM6

BM7

BM8

BM9

BM10

Figure 2. Event history of Image



Tabl

e3.

Bus

ines

sm

odel

sus

edby

@M

usic

BM

Off

erin

gM

arke

tIn

tern

alca

pabi

litie

sC

ompe

titiv

est

rate

gyE

cono

mic

sPe

rson

al/

inve

stor

fact

ors

Dis

tanc

eto

clos

est

prev

ious

BM

Dis

tanc

eto

clos

est

curr

ent

BM

Dis

tanc

eto

furt

hest

prev

ious

BM

Dis

tanc

eto

furt

hest

curr

ent

BM

1do

wnl

oada

ble

nonc

omm

erci

alqu

ality

mus

ic;

high

cust

omiz

atio

n;va

lue-

adde

dre

selli

ng;

dire

ctsa

les

b-to

-c;

inte

rnat

iona

l;ni

che

mar

ket;

tran

sact

iona

l

oper

atin

gsy

stem

s;W

ebde

sign

;m

arke

ting

inno

vatio

nle

ader

ship

com

mis

sion

per

song

sold

;m

ediu

mop

erat

ing

leve

rage

;hi

ghvo

lum

es;

low

mar

gins

grow

thm

odel

n.a.

n.a.

n.a.

n.a.

2cu

stom

ized

CD

s;so

me

cust

omiz

atio

n;va

lue-

adde

dre

selli

ng;

dire

ctsa

les

b-to

-b;

natio

nal;

nich

em

arke

t;tr

ansa

ctio

nal

mar

ketin

gpr

oduc

tfe

atur

esfle

xibl

epr

icin

g;m

ediu

mop

erat

ing

leve

rage

;m

ediu

mvo

lum

es;

med

ium

mar

gins

grow

thm

odel

5to

BM

15

toB

M1

5to

BM

16

toB

M3,

BM

4

3co

ncer

ts;

high

cust

omiz

atio

n;in

tern

alse

rvic

ede

liver

ing;

dire

ctsa

les

b-to

-c;

natio

nal;

nich

em

arke

ts;

tran

sact

iona

l

netw

orki

ngse

rvic

equ

ality

flexi

ble

pric

ing;

low

oper

atin

gle

vera

ge;

med

ium

volu

mes

;m

ediu

mm

argi

ns

inco

me

mod

el6

toB

M1

4to

BM

46

toB

M1

6to

BM

1,B

M2

4W

ebde

sign

serv

ices

;hi

ghcu

stom

izat

ion;

inte

rnal

serv

ice

deliv

erin

g;di

rect

sale

s

b-to

-c;

natio

nal;

broa

dm

arke

t;tr

ansa

ctio

nal

oper

atin

gsy

stem

s;W

ebde

sign

serv

ice

qual

ityfle

xibl

epr

icin

g;lo

wop

erat

ing

leve

rage

;m

ediu

mvo

lum

es;

low

mar

gins

inco

me

mod

el6

toB

M1

4to

BM

36

toB

M1

6to

BM

1,B

M2



Tabl

e4.

Bus

ines

sm

odel

sus

edby

Imag

e

BM

Off

erin

gM

arke

tIn

tern

alca

pabi

litie

sC

ompe

titiv

eSt

rate

gyE

cono

mic

sPe

rson

al/

inve

stor

fact

ors

Dis

tanc

eto

clos

est

prev

ious

BM

Dis

tanc

eto

clos

est

curr

ent

BM

Dis

tanc

eto

furt

hest

prev

ious

BM

Dis

tanc

eto

furt

hest

curr

ent

BM

1G

ener

alpu

rpos

em

achi

nevi

sion

syst

ems;

stan

dard

ized

;in

tern

ally

deve

lope

d;di

rect

sale

sto

user

s

b-to

-b;

natio

nal;

broa

d;re

latio

nal

imag

epr

oces

sing

tech

nolo

gyin

nova

tion

lead

ersh

ipfle

xibl

epr

icin

g;hi

ghop

erat

ing

leve

rage

;m

ediu

mvo

lum

es;

med

ium

mar

gins

grow

thm

odel

n.a.

n.a.

n.a.

n.a.

2M

achi

nevi

sion

syst

emfo

rflo

wer

insp

ectio

n;cu

stom

ized

;in

tern

ally

deve

lope

d;di

rect

sale

sto

user

s

b-to

-b;

natio

nal;

nich

e;re

latio

nal

imag

epr

oces

sing

tech

nolo

gyin

nova

tion

lead

ersh

ipfle

xibl

epr

icin

g;hi

ghop

erat

ing

leve

rage

;m

ediu

mvo

lum

es;

med

ium

mar

gins

grow

thm

odel

2to

BM

11

toB

M3,

BM

42

toB

M1

3to

BM

5,B

M6

3M

achi

nevi

sion

syst

emfo

rfr

uit

insp

ectio

n;cu

stom

ized

;in

tern

ally

deve

lope

d;di

rect

sale

sto

user

s

b-to

-b;

natio

nal;

nich

e;re

latio

nal

imag

epr

oces

sing

tech

nolo

gyin

nova

tion

lead

ersh

ipfle

xibl

epr

icin

g;hi

ghop

erat

ing

leve

rage

;m

ediu

mvo

lum

es;

med

ium

mar

gins

grow

thm

odel

2to

BM

11

toB

M2,

BM

42

toB

M1

3to

BM

5,B

M6

4M

achi

nevi

sion

syst

emfo

rau

tom

otiv

epa

rts

insp

ectio

n;cu

stom

ized

;in

tern

ally

deve

lope

d;di

rect

sale

sto

user

s

b-to

-b;

natio

nal;

nich

e;re

latio

nal

imag

epr

oces

sing

tech

nolo

gyin

nova

tion

lead

ersh

ipfle

xibl

epr

icin

g;hi

ghop

erat

ing

leve

rage

;m

ediu

mvo

lum

es;

med

ium

mar

gins

grow

thm

odel

2to

BM

11

toB

M2,

BM

32

toB

M1

3to

BM

5,B

M6

5M

achi

nevi

sion

syst

emfo

rau

tom

otiv

epa

rts

insp

ectio

n;st

anda

rdiz

ed;

inte

rnal

lyde

velo

ped;

dire

ctsa

les

tosy

stem

inte

grat

ors

b-to

-b;

inte

rnat

iona

l;ni

che;

tran

sact

iona

l

imag

epr

oces

sing

tech

nolo

gyin

nova

tion

lead

ersh

ipfix

edpr

icin

g;hi

ghop

erat

ing

leve

rage

;hi

ghvo

lum

es;

high

mar

gins

grow

thm

odel

3to

BM

11

toB

M6

3to

BM

13

toB

M2,

BM

3,B

M4

6M

achi

nevi

sion

syst

emfo

rph

arm

aceu

tics;

stan

dard

ized

;in

tern

ally

deve

lope

d;di

rect

sale

sto

syst

emin

tegr

ator

s

b-to

-b;

inte

rnat

iona

l;ni

che;

tran

sact

iona

l

imag

epr

oces

sing

tech

nolo

gyin

nova

tion

lead

ersh

ipfix

edpr

icin

g;hi

ghop

erat

ing

leve

rage

;hi

ghvo

lum

es;

high

mar

gins

grow

thm

odel

3to

BM

11

toB

M5

3to

BM

13

toB

M2,

BM

3,B

M4

7M

achi

nevi

sion

syst

emfo

rau

tom

otiv

epa

rts

insp

ectio

n;st

anda

rdiz

ed;

inte

rnal

lyde

velo

ped;

indi

rect

sale

sth

roug

hO

EM

s

b-to

-b;

inte

rnat

iona

l;ni

che;

rela

tiona

l

imag

epr

oces

sing

tech

nolo

gyin

nova

tion

lead

ersh

ipfix

edpr

icin

g;hi

ghop

erat

ing

leve

rage

;hi

ghvo

lum

es;

high

mar

gins

grow

thm

odel

2to

BM

5,B

M6

1to

BM

93

toB

M2,

BM

3,B

M4

3to

BM

2,B

M3,

BM

4,B

M8

8M

achi

nevi

sion

syst

emfo

rse

mic

ondu

ctor

insp

ectio

n;st

anda

rdiz

ed;

inte

rnal

lyde

velo

ped;

dire

ctsa

les

tosy

stem

inte

grat

ors

b-to

-b;

inte

rnat

iona

l;ni

che;

tran

sact

iona

l

imag

epr

oces

sing

tech

nolo

gyin

nova

tion

lead

ersh

ipfix

edpr

icin

g;hi

ghop

erat

ing

leve

rage

;hi

ghvo

lum

es;

high

mar

gins

grow

thm

odel

1to

BM

5,B

M6

1to

BM

5,B

M6

3to

BM

2,B

M3,

BM

4

3to

BM

2,B

M3,

BM

4,B

M7

9M

achi

nevi

sion

syst

emfo

rse

mic

ondu

ctor

insp

ectio

n;st

anda

rdiz

ed;

inte

rnal

lyde

velo

ped;

indi

rect

sale

sth

roug

hO

EM

s

b-to

-b;

inte

rnat

iona

l;ni

che;

rela

tiona

l

imag

epr

oces

sing

tech

nolo

gyin

nova

tion

lead

ersh

ipfix

edpr

icin

g;hi

ghop

erat

ing

leve

rage

;hi

ghvo

lum

es;

high

mar

gins

grow

thm

odel

2to

BM

5,B

M6

1to

BM

73

toB

M2,

BM

3,B

M4

3to

BM

2,B

M3,

BM

4

10‘s

econ

dop

tical

’m

achi

nevi

sion

syst

emfo

rse

mic

ondu

ctor

insp

ectio

n;st

anda

rdiz

ed;

inte

rnal

lyde

velo

ped;

indi

rect

sale

sth

roug

hO

EM

s

b-to

-b;

inte

rnat

iona

l;ni

che;

rela

tiona

l

imag

epr

oces

sing

tech

nolo

gyin

nova

tion

lead

ersh

ipfix

edpr

icin

g;hi

ghop

erat

ing

leve

rage

;hi

ghvo

lum

es;

high

mar

gins

grow

thm

odel

1to

BM

91

toB

M9

2to

BM

82

toB

M8



In the case of OOPs, the founders—soon afterinception in 1998—commit to the development of‘Spoot,’ a software product for B-to-C applications.Previous consulting activities were discontinued in1999 to concentrate solely on product developmentand sales. Even though many technical problemswere encountered and sales did not materialize asplanned, the company stuck to its business modeluntil 2001. At that point in time, the foundersstarted to analyze the lack of sales in-depth. Theyradically changed the business model’s offering,the competitive strategy it entailed, and its econom-ics. About a year later, they refined this businessmodel by identifying a niche market. Again, salesdid not materialize and yet another redirectionimposed itself.

In SiS, a very similar story unfolded. Although thefounders initially identified three possible applica-tions, they decided to develop communication ser-vices only. After two years of disappointing sales, aradical shift to storage applications took place. Thisimplied a complete overhaul in terms of envisagedcustomer segment, technology, and geographicalmarket. Also @Music (see Figure 1 and Table 3)focused from the very beginning on one specificbusiness model, namely online B-to-C sales of nichemusic labels. When, after three years, sales had stillnot materialized, they started to experiment withadditional business models, such as concert organiz-ing, Web design, and the production of custom-madeCDs (B-to-B). RegMed was slightly different fromthe other three cases in the sense that (1) it commit-ted very heavily to two parallel business models and(2) although these business models were still notgenerating sales after four years, no redirectionswere planned. The fact that RegMed was active in

biotech, a sector with long time horizons, probablyexplains this persistence.5

Rationale for focused commitment

From the historical descriptions, we find that earlycommitment by the ventures in our sample was notdue to a lack of awareness regarding uncertainty; itwas, in fact, a deliberate choice in the face of uncer-tainty. Founders and investors actually acknowl-edged at the start that many factors affecting thebusiness model’s viability were uncertain. Theyacknowledged technical uncertainties, uncertaintyabout how the market would evolve in terms of cus-tomer behavior and competition, and uncertaintyregarding regulation (see Table 6). Nevertheless,they decided to persistently focus on the develop-ment of one (or in the case of RegMed, two) specificbusiness model(s).

In our interviews (see Table 6), stakeholdersmentioned two main reasons for committing to onespecific business model despite the presence ofuncertainty. First, they focused early to capitalize onlearning effects. Both founders and investors wereconvinced that by putting all their efforts into onesingle business model, they would learn quicklyabout technical and market issues. They believed thiswould enable them to stay ahead of the competitionand capture a large market share. As one founder ofOOPs put it: ‘We just had to be first on that market.

5 The case of RegMed, which pursues multiple business modelsover longer time frames, is an example of what Westhead andWright (1998) call ‘portfolio entrepreneurship.’ Of course, suchpersistent adherence to multiple business models multipliesresource requirements compared to a persistent focus on onesingle business model.

Table 5. Overview of business model changes

Case Approach Average # ofnew business

models per year

Mean distanceto closest

previous BM

Mean distanceto closest

current BM

Mean distanceto furthest

previous BM

Mean distanceto furthest

current BM

@Music Focused commitment 1.00 6 4 6 6OOPs Focused commitment 0.83 4 4 4 4SiS Focused commitment 0.50 5 n.a. 5 n.a.RegMed Focused commitment 0.40 n.a. 1 n.a. 1Image Simultaneous

experimentation1.67 2 1 3 3

L-goritm Simultaneousexperimentation

1.57 2 1 3 3



Tabl

e6.

Quo

tatio

nson

focu

sed

com

mitm

ent

Rat

iona

leIm

plic

atio

ns

Ack

now

ledg

emen

tof

unce

rtai

nty

Incr

ease

dm

obil

izin

gpo

wer

Pro

mis

eof

achi

evin

gle

arni

ngef

fect

san

dfir

st-m

over

adva

ntag

es

Ded

icat

edor

gani

zati

onal

stru

ctur

ean

dco

nsid

erab

leex

pend

itur

es

Dif

ficul

ties

inre

orga

nizi

ngw

hen

assu

mpt

ions

fail

Dif

ficul

ties

inac

quir

ing

exte

rnal

inve

stm

ent

whe

nas

sum

ptio

nsfa

il

@M

usic

•W

eha

dth

isvi

sion

that

the

era

ofth

etr

aditi

onal

mus

icbu

sine

ssw

ould

beov

erso

on.I

tw

asex

pect

edth

atth

roug

hout

the

wor

ld,t

heIn

tern

etw

ould

com

plet

ely

chan

geco

nsum

ers’

buyi

ngbe

havi

or.O

fco

urse

,no

body

knew

whe

nan

dho

wth

isw

ould

happ

en.

•W

ew

ent

toa

big

inte

rnat

iona

lm

usic

fair

...A

llth

ese

labe

lsju

stlo

ved

us.W

eca

me

back

with

sign

edag

reem

ents

from

all

impo

rtan

tla

bels

...T

hein

vest

ors

wer

eth

rille

d.

•T

hegu

ysfr

omth

eau

thor

righ

tsau

thor

itydi

dn’t

know

wha

tto

dow

ithus

.The

yha

dne

ver

deal

tw

ithon

line

mus

icsa

les

...W

ew

ere

actu

ally

sitti

ngw

ithth

em,w

ritin

gre

gula

tions

abou

tau

thor

righ

tspr

otec

tion

and

tari

ffs

for

onlin

esa

les.

Itw

as,o

fco

urse

,agr

eat

adva

ntag

eto

doth

at.

•T

hein

vest

ors

had

this

idea

that

toge

ton

line

sale

sgo

ing,

we

shou

ldla

unch

abi

gm

arke

ting

cam

paig

n.A

nin

tern

atio

nal

mar

ketin

gm

anag

erw

asat

trac

ted

todo

that

.We

wer

ebu

rnin

gca

shlik

ehe

ll.B

utw

ew

ere

told

that

that

’sth

ew

ayyo

ush

ould

doit.

•T

hese

thin

gsha

dno

thin

gto

dow

ithth

ein

itial

idea

.It

was

not

wha

tI

sign

edup

for.

Iw

ante

dto

wor

kfo

ra

com

pany

selli

ngm

usic

onlin

e,go

ing

com

plet

ely

agai

nst

the

trad

ition

alm

usic

indu

stry

...

Iha

dle

ftm

yjo

bfo

rth

at.

Iw

asn’

tin

tere

sted

inW

ebde

sign

orin

orga

nizi

ngco

ncer

ts.

•T

hey

wer

ede

sper

atel

ytr

ying

toge

nera

tere

venu

esfr

omco

ncer

tor

gani

zing

,W

ebde

sign

and

othe

rst

uff.

The

sew

ere

not

the

type

sof

activ

ities

we

wan

ted

toin

vest

inas

vent

ure

capi

talis

ts.

OO

Ps•

Bas

edon

ago

vern

men

tsu

bsid

y,w

edi

da

proo

fof

conc

ept

for

the

Spoo

tpr

oduc

t.T

here

sults

wer

epr

omis

ing,

but

alo

tof

tech

nica

lis

sues

rem

aine

dun

solv

ed..

.It

was

uncl

ear

whe

ther

we

coul

dpu

llth

isof

f.N

ever

thel

ess,

we

aban

done

dou

rpr

ofit-

mak

ing

cons

ultin

gac

tiviti

esan

dfo

cuse

dal

lou

ref

fort

son

deve

lopi

ngth

ispr

oduc

t.

•W

ew

ere

quite

happ

yw

ithth

ere

venu

esco

min

gfr

omou

rco

nsul

ting

activ

ities

.At

leas

tit

prov

ided

som

eki

ndof

inco

me

...t

hat

coul

dpa

rtia

llysu

ppor

tth

ede

velo

pmen

tof

Spoo

t.B

utth

ein

vest

ors

wer

eno

tin

tere

sted

inye

tan

othe

rco

nsul

ting

com

pany

.The

yw

ante

dus

topu

tal

lef

fort

son

prod

uct

deve

lopm

ent.

•W

eju

stha

dto

befir

ston

that

mar

ket.

We

wou

ldbe

way

ahea

dof

com

petit

ors

byth

etim

eth

eyev

enun

ders

tood

the

oppo

rtun

ity.

•T

hein

vest

ors

brou

ght

ina

big

shot

with

tons

ofsa

les

expe

rien

cefr

oma

big

firm

...H

ebr

ough

thi

sem

ploy

ees

with

him

.Su

dden

ly,w

eha

dan

incr

edib

lyex

pens

ive

sale

ste

amof

20pe

ople

.Stil

lno

thin

gso

ld.

•W

eha

dse

para

teR

&D

and

serv

ice

divi

sion

s.Pe

ople

from

thes

edi

visi

ons

wer

eke

ptap

art.

The

yw

ere

inop

posi

tesi

des

ofth

ebu

ildin

g..

.We

had

this

idea

that

they

shou

ldno

tin

terf

ere

with

each

othe

r,bu

tsh

ould

focu

son

thei

rm

ain

task

.

•T

hesa

les

man

ager

and

his

entir

esa

les

team

wer

efir

ed.

Inst

ead,

the

mar

ketin

gte

ambe

cam

ere

spon

sibl

efo

rsa

les.

Toim

prov

eou

rin

sigh

tsin

cust

omer

need

s,al

lde

velo

pers

had

tost

art

wor

king

dire

ctly

for

the

cust

omer

...A

lthou

ghI

still

thin

kth

isw

asa

good

deci

sion

,it

also

pose

dpr

oble

ms.

Som

eof

thes

ede

velo

pers

wer

ere

ally

tech

nica

lpe

ople

,typ

ical

‘ner

ds’

let’

ssa

y.N

otth

ety

peof

peop

leyo

uw

ant

tose

ndto

acu

stom

er.

•Su

dden

ly,o

neof

the

inve

stor

s..

.com

esup

with

the

idea

tofo

rma

join

tve

ntur

e..

.He

just

didn

’tbe

lieve

anym

ore

that

we

coul

dm

ake

iton

our

own.

Now

,whi

leth

ism

erge

rpr

obab

lyw

ould

bebe

nefic

ial

for

him

,the

rew

asno

thin

gin

ther

efo

rm

e,an

dI

didn

’tag

ree

...

Ica

me

upw

ithan

idea

for

aw

hole

new

mar

ket

...

The

rew

asno

supp

ort

wha

tsoe

ver

onth

ebo

ard.

Iim

med

iate

lyre

aliz

edth

atth

eco

mpa

nyco

uld

not

goon

with

out

the

supp

ort

ofth

ein

vest

ors

...T

wo

mon

ths

late

r,I

filed

for

bank

rupt

cy.

SiS

•O

fco

urse

ther

ew

asa

lack

ofcl

arity

abou

tw

hat

the

mar

ket

wou

ldlo

oklik

e.N

obod

yre

ally

knew

whe

ther

peop

lew

ould

buy

onlin

e.W

ekn

ewth

atth

isw

ould

bea

key

dete

rmin

ant.

•It

was

atig

htpl

an.E

very

body

coul

dea

sily

unde

rsta

ndw

hat

they

wan

ted

todo

...T

hest

ory

was

easy

tose

ll.

•O

fco

urse

ther

ew

asa

lack

ofcl

arity

abou

tw

hat

the

mar

ket

wou

ldlo

oklik

e..

.But

inan

yca

se,y

ouw

ante

dto

beth

ere

first

—to

obta

inm

arke

tsh

are

inca

seit

beca

me

asu

cces

s.

•W

ithth

em

oney

from

the

first

finan

cing

roun

d,w

eim

med

iate

lyst

arte

dto

hire

peop

le..

.to

exec

ute

our

plan

.

•T

his

was

real

lya

com

plet

eov

erha

ul.W

ew

ould

now

focu

son

this

new

appl

icat

ion,

but

we

didn

’tha

veth

eex

pert

ise.

We

had

tost

art

sour

cing

inte

chno

logy

from

exte

rnal

part

ners

,bec

ause

we

didn

’tha

veth

eca

pabi

litie

sto

deve

lop

itou

rsel

ves.

•O

urin

vest

orw

asn’

tin

tere

sted

anym

ore.

Luc

kily

,thi

sw

asal

lbe

fore

the

dotc

omcr

ash.

We

wer

eab

leto

attr

act

new

inve

stor

san

dw

ithth

isne

wca

pita

lin

ject

ion,

we

basi

cally

star

ted

all

over

agai

n.

Reg

Med

•T

his

isa

biot

ech

star

t-up

,ri

ght.

The

re’s

alw

ays

unce

rtai

nty

abou

tw

heth

erth

ete

chno

logy

will

wor

k,w

heth

eryo

u’ll

get

FDA

appr

oval

,and

thin

gslik

eth

at.

•Pr

ofes

sor

L.c

ame

back

from

the

Stat

esw

ithth

isve

rysp

ecifi

can

dcl

ear

idea

ofw

hat

hew

ante

dto

do..

.He

cam

eto

us(i

.e.,

the

TT

O).

We

have

confi

denc

eth

athe

can

pull

this

off.

He’

sin

tern

atio

nal

top

inth

isfie

ldan

dhe

real

lykn

ows

wha

the

’sdo

ing.

•W

ere

ally

need

toin

vest

all

poss

ible

effo

rts

inth

ispr

oduc

t.It

coul

dbe

com

eth

efir

stbi

otec

hpr

oduc

tge

tting

FDA

appr

oval

.Thi

sw

ould

open

upin

cred

ible

mar

ket

oppo

rtun

ities

.

•Y

ouca

n’t

just

hire

gene

ralis

tsfo

rth

isty

peof

wor

k.Y

oune

edre

sear

cher

sw

ithsp

ecifi

ckn

owle

dge

abou

tjo

ints

and

cart

ilage

.

(did

not

occu

rin

peri

oddo

cum

ente

d)(d

idno

toc

cur

inpe

riod

docu

men

ted)



We would be way ahead of competitors by the timethey even understood the opportunity.’ Persistentfocus is, hence, seen as a way to quickly reduceuncertainty over a specific business model and tocreate first-mover advantages.

Second, our interviews with founders and inves-tors further reveal that focus on one specific businessmodel improves the clarity of the value propositionfor different stakeholders. The fact that the compa-nies have well-delineated business models allowsthem to attract motivated employees and to convincestrategic partners. In all four cases, the independentventure capitalists (VCs) involved were very muchin favor of a persistent focus on a specific businessmodel. In the case of OOPs, the investors were actu-ally pushing the founders to focus solely on productdevelopment, as one founder recalled: ‘We werequite happy with the revenues coming from our con-sulting activities. At least it provided some kind ofincome . . . that could partially support the develop-ment of Spoot. But the investors were not interestedin yet another consulting company. They wanted usto put all efforts on product development.’

Implications of focused commitment

Our historical case descriptions and our interviews(see Table 6) demonstrate that focusing on a specificbusiness model results not only in mobilizing powerbut also in clarity regarding the venture’s organiza-tional configuration. Objectives and activities aredeveloped in line with the envisaged value proposi-tion, and employees whose competencies reflectthe objectives outlined in the business model areattracted. As an OOPs employee testified: ‘Theinvestors brought in a big shot with tons of salesexperience from a big firm . . . He brought hisemployees with him.’ This results in a dedicatedorganizational structure, consisting of different func-tional departments, as indicated by that sameemployee: ‘We had separate R&D and service divi-sions. People from these divisions were kept apart.They were in opposite sides of the building.’ Thedevelopment of these dedicated activities impliessignificant expenditures. Our case studies clearlyshow that external investors are supportive of thesecostly initiatives, believing these will enable ven-tures to achieve fast growth. One @Music founderpointed out that: ‘The investors had this idea thatto get online sales going, we should launch a bigmarketing campaign. An international marketingmanager was attracted to do that. We were burning

cash like hell. But we were told that that’s the wayyou should do it.’

Over time, the venture develops the chosen busi-ness model and thereby reduces uncertainty pertain-ing to the underlying assumptions. This process canyield outcomes ranging from full confirmation ofinitial assumptions to the complete opposite. In threeout of the four case studies—OOPs, @Music, andSiS—crucial assumptions proved to be too optimis-tic. As the founder of SiS testified: ‘The competitionon the home market turned out to be much fiercerthan expected (because) this one big player decidedto become active.’ In multiple cases, including@Music, consumer buying behavior turned out to bedifferent than expected, as one employee recalled:‘Only later, we realized that people were just down-loading to listen, but were simply not willing to buy.’

Our case studies show that when ventures arefaced with discrepancies between initial assump-tions and unfolding reality, change becomes inevi-table for their survival (see Ambos and Birkinshaw,2010, on the role of cognitive dissonance in ven-tures’ development processes). From our analysis, itbecame clear that OOPs, @Music, and SiS turned toconsecutive business model search. These redirec-tions had serious implications for the ventures’ inter-nal organization. These ventures have attractedspecialized personnel and have introduced dedicatedorganizational structures—commitments, all in linewith their initial focus. Redefining the businessmodel then becomes complex, as it requires chang-ing the mind-set of the entrepreneurial team and itsstakeholders. It implies choosing a different set ofactivities, skills, and structures (Miller and Friesen,1984). Some of the specialized competencies maybecome obsolete, while others may be missing. Theventures’ core capabilities might, therefore, turn intocore rigidities (see also Dougherty, 1995, andAmbos and Birkinshaw, 2010, on disruptive transi-tions). In the case of OOPs, founders decide to dras-tically adapt their development and sales approachafter having undertaken an in-depth analysis of nega-tive sales results: ‘To improve our insights in cus-tomer needs, all developers had to start workingdirectly for the customer . . . Although I think thiswas a good decision, it also posed problems. Some ofthese developers were really technical people,typical “nerds” let’s say. Not the type of people youwant to send to a customer.’ Moreover, our casestudies confirm that employees themselves are notalways willing to take part in this change process.Or, as one @Music employee indicated: ‘These



things had nothing to do with the initial idea. It wasnot what I signed up for. I wanted to work for acompany selling music online, going completelyagainst the traditional music industry . . . I had leftmy job for that. I wasn’t interested in Web design orin organizing concerts.’

In addition, the case studies demonstrate that it isdifficult to convince investors to fund reconfigura-tions, since these investors question whether theinitial business model was wrong or whether theproblems are simply due to lousy execution bythe venture’s founders (cf. Bhide, 1992). Thisweakens investor confidence. As one OOPs founderexplained (see Table 6): ‘Suddenly, one of the inves-tors . . . comes up with the idea to form a jointventure with another company in his portfolio. Hejust didn’t believe anymore that we could make it onour own. Now, while this merger probably would bebeneficial for him, there was nothing in it for me, andI did not agree . . . I came up with an idea for a wholenew market . . . There was no support whatsoever onthe board. I immediately realized that the companycould not go on without the support of the investors. . . Two months later, I filed for bankruptcy.’ OOPsand @Music did not succeed in mobilizing resourcesfor executing consecutive business model searchesand went bankrupt.

Summary and propositions on focused commitment

Our case studies show that some ventures persistentlyfocus on a specific business model. This focus is adeliberate choice in the face of uncertainty since it isinstrumental in mobilizing different stakeholders(among whom independent VCs figure prominently),and ventures believe it can result in first-mover advan-tages. This approach entails significant expenditures:launching dedicated campaigns, developing an elabo-rate organizational structure, and hiring specializedpersonnel with skills matching the chosen businessmodel. Given that focused commitment is instrumen-tal in mobilizing resources, it facilitates rapid organi-zational growth during the initial phases of theventure’s life.

However, our case studies illustrate that, in reality,these first-mover advantages might not materialize(see also the work by Dowell and Swaminathan,2006, showing that the occurrence of first-moveradvantages in emergent industries is indeed a rareevent). When the initial assumptions underlying thebusiness model prove to be incorrect, the ventures inour study turn to consecutive searches and experi-

ment with new business models. As our cases illus-trate, such change processes are hazardous andpainful since they require the venture to becomeflexible again with regard to its assumptions and itsorganizational configuration. Given the specializedcapabilities and dedicated organizational structure,this becomes very difficult. Our case studies showthat founders and employees are not always capableor willing to do so and that investors are reluctant tofinance such changes. The ventures’ organizationalrigidities and financing problems reduce the possi-bility of exploring new business models, therebyhampering the venture’s ability to identify a viablebusiness model and, hence, also its long-term sur-vival. Based on these findings, we propose that:

Proposition 1: Focused commitment has a posi-tive effect on the initial growth of venturesoperating under uncertainty.

Proposition 2: Focused commitment jeopardizesthe long-term survival of ventures operatingunder uncertainty.

Simultaneous experimentation

We find that some ventures do not commit earlyon to a specific business model, but develop diverg-ing search paths by engaging in a series of relatedbusiness model experiments. L-goritm initiallyexperimented with a variety of business modelssimultaneously. In the first year of its existence, thecompany developed software enabling the use ofadvanced laser applications in product design,including applications for reverse engineering aswell as for quality control. The latter were sold incombination with quality control services. Duringthe entire period of 1997 to 1999, the companyexperimented with a combination of indirect anddirect sales of these software products. From 2000onward the company discontinued some of theseactivities, ending up with one business model by2001. This business model was viable in the sensethat it generated sufficient income for the companyto break even, grow internationally, and becomeworld leader in its niche market.

Image, however, began with one single businessmodel but immediately developed it into a portfolioof five different business models (see Figure 2 andTable 4). After four years, the venture gradually nar-rowed down its business model portfolio and, by1988 (i.e., five years after inception), one viable



business model remained. In this case too, the busi-ness model was the start of a highly prosperousinternational trajectory and a successful IPO.

Rationale for simultaneous experimentation

From interviews with stakeholders (see Table 7), wefound that multiple search paths of related businessmodel experiments were executed in parallel for tworeasons. First, although each new experiment wasrelated to another experiment in the portfolio, thefact that these search paths developed in differentdirections implies that ventures end up experiment-ing with a variety of business models. Ventures con-sider this instrumental in addressing uncertainty. Itallows them to learn and, hence, reduce uncertaintyabout a relatively broad range of opportunities. Ifone business model proves unviable, there is still achance that one of the other options may prove suc-cessful. In doing so, the ventures manage risk andretain strategic flexibility (Raynor, 2007; deWeerd-Nederhof et al., 2008). As one founder ofImage related: ‘It was impossible to foresee whichmarket niches would emerge and which ones wouldbecome profitable. We just had to spread ourchances.’ Or, as the CEO of L-goritm testified: ‘The(indirect) sales approach didn’t work . . . We had toabandon it. That wasn’t a problem really, becausewe already had experience with direct sales.’ Theventures in our sample used a combination of mul-tiple search paths to generate a variety of businessmodel experiments and, thereby, avoided prelimi-nary commitment to a suboptimal business model.

Second, learning about one option can also reduceuncertainty about other options, especially whennew business model experiments are related to atleast one other business model in the portfolio. Weobserve that ventures use the knowledge acquiredfrom exploring one option to reassess and redefinethe nature of, and the priorities for, the activity port-folio as a whole. L-goritm gathered knowledge andexpertise from its service activities and—based onthis experience—decided to add hardware compo-nents to its product offerings, as the CEO testified:‘What we saw when doing consulting activities isthat customers didn’t realize that it is the soft-ware that determines total performance. They werealways talking about the hardware. We realized thatif we wanted to develop a profitable software tool, wewould have to include hardware in order to sell itand to drive up the price.’ Hence, ventures explicitlyregard engagement in an additional business model

experiment as instrumental in re(de)fining otherbusiness models in the portfolio.

Our case study analysis reveals that also investorswere following this rationale. They did not regardthese experiments as resource-consuming errors thatneeded to be avoided, but as appropriate and neces-sary ways to select the most interesting entrepreneur-ial opportunity (see interview quotes in Table 7). Inthe cases of L-goritm and Image, both public sectorand independent VCs were willing to invest in ven-tures that explicitly acknowledge uncertainty anddevelop various related business model search pathsin parallel. L-goritm’s investors—when closing theinvestment agreement—were fully aware and sup-portive of the venture’s plan to experiment withvarious business models, as L-goritm’s cofoundertestified. ‘The investors did not mind. On the con-trary. They understood that we could not know inadvance what would work.’ In addition, the casestudies reveal that investors were closely monitoringprogress and—instead of providing a major capitalinjection upfront—were staging their investmentsaccordingly. As L-goritm’s cofounder observed: ‘. . .they were monitoring pretty closely what we weredoing in all these areas, just to make sure that wewere making progress overall. They didn’t provideus with a large investment sum you sometimes saw inother companies. Instead, they infused additionalcapital as some of our projects turned out to bepromising.’

Implications of simultaneous experimentation

The cases reveal that, although investors are support-ive of experimentation with a portfolio of businessmodels, they refrain from injecting large amounts ofcapital in the initial stages of the ventures’ life andinstead stage their investments as progress is made.We observe that to deal with these initial resourcelimitations, ventures adopt cost-effective strategies.First, business models are included only if they entailsignificant technical or commercial overlap withother business models in the portfolio. By develop-ing multiple search paths of related business models,ventures can redeploy activities in different businessmodel logics, thereby creating spillovers betweenthe different activities undertaken. Second, they uselow-cost probing strategies to find out which prod-ucts and services the markets prefer (cf. Brown andEisenhardt, 1997), as the founder of Image recalled:‘In these early days, our projects consisted mainly ofpilots. We set up a prototype and checked whether it



Tabl

e7.

Quo

tatio

nson

sim

ulta

neou

sex

peri

men

tatio

n Rat

iona

leIm

plic

atio

ns

Ris

km

anag

emen

tK

now

ledg

esp

illo

vers

Ext

erna

lin

vest

ors’

supp

ort

Cos

tef

ficie

ncy

Gra

dual

evol

utio

nfr

ompr

ojec

t-ba

sed

orga

niza

tion

tom

ore

elab

orat

est

ruct

ure

Imag

e•

Itw

asim

poss

ible

tofo

rese

ew

hich

mar

ket

nich

esw

ould

emer

gean

dw

hich

ones

wou

ldbe

com

epr

ofita

ble.

We

just

had

tosp

read

our

chan

ces.

•O

fco

urse

we

didn

’tju

stdo

anyt

hing

.We

delib

erat

ely

sele

cted

proj

ects

whe

rew

eco

uld

gene

rate

som

eki

ndof

cros

s-fe

rtili

zatio

n,w

here

the

thin

gsyo

ule

arn

inon

epr

ojec

tca

nbe

used

for

refin

ing

your

othe

rac

tiviti

es.

•W

ere

ceiv

edse

vera

lra

ther

limite

dca

pita

lin

fusi

ons

over

time,

depe

ndin

gon

the

iden

tifica

tion

ofpr

omis

ing

activ

ities

and

the

prog

ress

we

had

mad

e.•

We

neve

rha

dan

ytr

oubl

ede

fend

ing

our

proj

ects

toth

eex

tern

albo

ard

mem

bers

.

•In

thes

eea

rly

days

,ou

rpr

ojec

tsco

nsis

ted

mai

nly

ofpi

lots

.We

set

upa

prot

otyp

ean

dch

ecke

dw

heth

erit

coul

ddo

the

job.

•In

the

begi

nnin

g,w

ew

ere

doin

gev

eryt

hing

ours

elve

s.I

was

help

ing

him

tode

velo

pth

eso

ftw

are,

but

Iw

asal

sose

lling

and

doin

gqu

ality

cont

rol

serv

ices

.Iw

asa

sale

sman

,aco

nsul

tant

,and

apr

oduc

tde

velo

per

all

inon

epe

rson

.•

By

2001

,we

had

deve

lope

da

prof

essi

onal

stru

ctur

e,in

clud

ing

anR

&D

team

,asa

les

man

agem

ent

team

,am

arke

ting

and

sale

ssu

ppor

tte

am,a

nda

finan

cean

dad

min

istr

atio

nun

it.L

-gor

itm•

The

(ind

irec

t)sa

les

appr

oach

didn

’tw

ork

...W

eha

dto

aban

don

it.T

hat

was

n’t

apr

oble

mre

ally

,bec

ause

we

alre

ady

had

expe

rien

cew

ithdi

rect

sale

s.

•W

hat

we

saw

whe

ndo

ing

cons

ultin

gac

tiviti

esis

that

cust

omer

sdi

dn’t

real

ize

that

itis

the

soft

war

eth

atde

term

ines

tota

lpe

rfor

man

ce.T

hey

wer

eal

way

sta

lkin

gab

out

the

hard

war

e.W

ere

aliz

edth

atif

we

wan

ted

tode

velo

pa

profi

tabl

eso

ftw

are

tool

,we

wou

ldha

veto

incl

ude

hard

war

ein

orde

rto

sell

itan

dto

driv

eup

the

pric

e.

•T

hein

vest

ors

did

not

min

d.O

nth

eco

ntra

ry.T

hey

unde

rsto

odth

atw

eco

uld

not

know

inad

vanc

ew

hat

wou

ldw

ork.

But

they

wer

em

onito

ring

pret

tycl

osel

yw

hat

we

wer

edo

ing

inal

lth

ese

area

s—ju

stto

mak

esu

reth

atw

ew

ere

mak

ing

prog

ress

over

all.

The

ydi

dn’t

prov

ide

usup

fron

tw

itha

larg

ein

vest

men

tlik

eyo

uso

met

imes

saw

inot

her

com

pani

es.

Inst

ead,

they

infu

sed

addi

tiona

lca

pita

las

som

eof

our

proj

ects

turn

edou

tto

bepr

omis

ing.

•T

his

soft

war

ew

ould

take

time

tode

velo

p.So

we

need

edot

her

activ

ities

toge

nera

tem

oney

.O

ffer

ing

serv

ices

for

qual

ityco

ntro

lw

asth

eea

sies

tw

ayto

doth

at.

•O

fco

urse

we

had

diff

eren

tpe

ople

doin

gdi

ffer

ent

thin

gs.

But

ther

ew

asno

tre

ally

ahi

erar

chic

alst

ruct

ure.

The

rew

ere

nocl

ear

divi

sion

s..

.It

was

mor

epr

ojec

tba

sed.

Lat

eron

,onc

ew

eha

da

bette

rvi

ewof

wha

tou

rco

rem

arke

tw

ould

be,w

eof

cour

seor

gani

zed

ina

mor

epr

ofes

sion

alw

ay.Y

oukn

ow,

with

real

func

tiona

ldi

visi

ons

and

thin

gslik

eth

at.W

eal

soat

trac

ted

mor

epe

ople

.



could do the job.’ And finally, they include businessmodel experiments with short-term cash-generatingpotential in their portfolio. Similarly, the founder ofL-goritm stated: ‘This software would take time todevelop. So we needed other activities to generatemoney. Offering services for quality control was theeasiest way to do that.’

The development of the ventures’ organizationalstructures is in line with the chosen portfolioapproach. An organic project-based configurationemerges, which resembles the project-based natureof R&D departments and provides flexibility(Fiegenbaum and Karnani, 1991). During the firstyears, L-goritm’s founders took care of all businessactivities themselves. After awhile, three employeeswere hired, but the structure remained project ori-ented rather than functional. Over time, specificproject teams were formed, each of them focusing onbusiness models considered worthwhile pursuing.The founder’s quotations in Table 7 illustrate this:‘In the beginning, we were doing everything our-selves. I was helping him to develop the software, butI was also selling and doing quality control services.I was a salesman, a consultant and a product devel-oper all in one person.’

Since these ventures acknowledged that thevarious conceivable business models are all charac-terized by uncertainty, they postponed the decisionto commit to one option until more information withrespect to a range of value propositions becameavailable. As uncertainty decreased, the companiesgradually narrowed down the range of businessmodels until a viable business model remained.L-goritm gradually narrowed down its activity rangefrom 2000 onward and ended up with one successfulbusiness model by the end of 2001. A similar processtook place in Image. This portfolio reductionentailed organizational changes. The venturesevolved from loose configurations resembling R&Ddepartments’ project-based characteristics to moreelaborated structures, as they narrowed down therange of business models. Or as the CEO of L-goritmtestified: ‘Later on, once we had a better view ofwhat our core market would be, we of course orga-nized in a more professional way. You know, withreal functional divisions and things like that. We alsoattracted more people.’

Summary and propositions on simultaneousexperimentation

Our case studies reveal that some ventures developeda portfolio of business model experiments in order to

learn about a broad variety of potential businessmodels. Both public sector and independent inves-tors are supportive of this approach, as they considerit appropriate to select the most interesting entrepre-neurial opportunity. However, instead of injectinglarge amounts of capital in the initial stages of theventures’ life, they stage their investments as prog-ress is made. This limits initial organization growthand leads to the development of a project-basedorganization of limited scale.

To deal with these initial resource limitations,ventures deploy low-cost probing strategies, includebusiness model experiments with short-term cash-generating potential, and enact knowledge and costspillovers. The latter is achieved by including onlybusiness models that are related to at least one otherexperiment in the portfolio. The ventures, therefore,do not select experiments at random, but do so in adeliberate manner. Although each new experiment isrelated to at least one other experiment in the port-folio, the parallel search paths develop in variousdirections. As a result, the ventures reduce uncer-tainty with respect to a wide variety of businessmodels. This uncertainty reduction facilitates theselection of viable options over time. Staged invest-ments and the evolution of the ventures’ organiza-tional structures from informal and project-based tomore formal and divisional, reflect this process. Thefact that ventures learn about a wide variety of busi-ness models in a cost-effective manner increasestheir chances of identifying viable business modelsand, hence, facilitates their survival in the long run.Based on these findings, we propose that:

Proposition 3: Compared to focused commitment,simultaneous experimentation reduces the initialgrowth of ventures operating under uncertainty.

Proposition 4: Compared to focused commit-ment, simultaneous experimentation facilitateslong-term survival of ventures operating underuncertainty.

DISCUSSION

In this article, we have analyzed how ventures con-fronted with uncertainty develop their businessmodel through experimentation. In particular, weinvestigated (1) whether different experimentationand learning approaches exist and, if so, (2) what therationale and implications of these approaches are.



In doing so, we drew on business model literatureand organizational learning theory. The latter pro-poses two main experimentation approaches tolearning under uncertainty: (1) local or relatedsearch versus (2) distant search or search throughlong jumps. While local search is expected to resultin a low to moderate variety, distant searches areoften deemed necessary to generate considerablelevels of variety.

Some ventures in our sample indeed tried toexplore a broader range of business models througha sequence of distant business model searches, oncetheir initial business models proved to be unviable.However, altering business models after a period offocused commitment is hazardous and painful. Theventures in our study all had considerable difficultiesin mobilizing stakeholders and additional resourcesfor these subsequent experiments, leading to bank-ruptcy in several cases.

At the same time, our research shows that thereare less hazardous ways to enact variety in terms ofbusiness models. Ventures can develop multiplediverging search paths of related business modelexperiments. These simultaneous search paths ini-tially start from the same business model but spreadout in various directions. As a result, ventures end upexperimenting with different business models. Theparallel pursuit of different business models createsa considerable variety of (real) options, whichincreases the odds for survival and growth (GuntherMcGrath, 1999). This finding enriches organiza-tional learning theory by going against the assump-tion that only distant search can lead to considerablelevels of variety. Our findings reveal that a portfolioof related search paths is equally instrumental inenacting variety.

Our findings not only reveal how simultaneousexperimentation generates variety in terms of busi-ness models, but also demonstrate how it does this ina cost-effective manner. By including only businessmodels that are related to at least one other experi-ment in the portfolio and adhering to low-costprobing strategies, the ventures can enact spilloversefficiently. Simultaneous experimentation, therefore,presents itself as a relevant learning strategy for ven-tures operating under uncertainty, alongside previ-ously documented bootstrapping approaches (Bhide,1992; Winborg and Landström, 2001).

Our work not only extends organizational learningtheory, it also contributes to the lively discussion oncausation and effectuation. Sarasvathy (2001) dis-cusses the relevance of causation and effectuation

processes to firm creation. Causation processesimply that entrepreneurs choose specific means (aspecific organizational structure, specific employeeskills, etc.) to create a desired effect (such as aspecific business model). Effectuation processes,however, imply that ventures draw on their knowl-edge and networks and select between possibleeffects that can be created with this set of means.Whereas existing work discusses effectuation ascognition or logic, research on how effectual logictranslates into effectual behavior is scarce (excep-tions include recent contributions by Fisher, 2012,and Mauer, 2009). In an effectual logic, the basis fortaking action is means driven, investment decisionsinvolve a commitment limit in terms of affordableloss, and stakeholders and unforeseen events heavilyinfluence the development direction (Read et al.,2009; Sarasvathy, 2008).

In our study, ventures deliberately develop portfo-lios of business model experiments, starting frominitial ideas, available capabilities, and experiences.They add specific business models to the portfolioand redefine others based on experiences and infor-mation gathered in the course of previous businessmodel experiments. Simultaneous experimentationconsequently represents an effectual logic, since ittakes a set of means as a given and focuses on select-ing between possible effects that can be created withthis set of means. Focused commitment, however,follows a causal logic where ventures choose spe-cific means (a specialized organizational structure,specialized personnel, and significant expenditures)to create a desired effect (namely, the implementa-tion of one, ex ante chosen business model).

In addition, the fact that simultaneous experimen-tation involves careful selection of related experi-ments suggests a reconciliation of the notions of‘action’ and ‘planning.’ In Sarasvathy’s work(Sarasvathy, 2001), effectuation is partly driven bycoincidence: ‘Whoever first buys . . . becomes, bydefinition, the first target customer. By continuallylistening to the customer and building an ever-increasing network of customers and strategic part-ners, the entrepreneur can then identify a workablesegment profile’ (Sarasvathy, 2001: 247). Otherauthors have also attributed an important role to‘initial coincidences’ in the creation of entrepreneur-ial opportunities (Alvarez and Barney, 2007).However, we find that the ventures in our casestudy do not just ‘go with the flow’ in developing aportfolio of business model experiments. Instead,they consciously select and design these experiments.



Ventures deliberately select business model experi-ments that are related and that offer cross-fertilizationopportunities, allowing them to exploit their meansand strengths efficiently across a variety of businessmodels. Simultaneous experimentation, therefore,implies both effectual experimentation building onthe venture’s own means and strengths and the con-scious planning and selection of specific businessmodel experiments to be included in the portfolio.

Whereas ‘planning’ and ‘action’ or ‘planning’ and‘learning’ have been juxtaposed in management lit-erature (e.g., Liao and Gartner, 2006), our findingssuggest that these two notions can be reconciled ascomplementary aspects constituting the simultane-ous experimentation approach. Moreover, our find-ings suggest that this combination of action andplanning results not only in substantial variety, but itcan also be organized in a cost-effective manner and,therefore, it offers good prospects for the survivaland growth of entrepreneurial ventures operatingunder uncertainty. Simultaneous experimentation,thus, reconciles Sarasvathy’s (2001) preference foreffectual cognition with Liao and Gartner’s (2006)preference for planning as a way to deal withuncertainty.

Our work also demonstrates that some investors(including both public sector and independent VCs)are supportive of this combination of planning andaction. They finance ventures while being fullyaware of their intentions to experiment with multiplebusiness models. Those investors do not regard theseexperiments as resource-consuming errors that needto be avoided but as necessary efforts to select themost interesting entrepreneurial opportunity. Thus,these findings refute the widely held belief that VCsare willing to finance only ventures with a focusedbusiness plan. Whereas investment in focused com-panies requires managing risk at the level of theportfolio, financing ventures engaged in simultane-ous experimentation implies that risk is alsomanaged within the boundaries of the venture. Wesuggest that this approach has the potential toincrease the overall success rate and, hence, thevalue of VCs’ investment portfolios.

Finally, we believe our work contributes to thedevelopment of empirical studies on businessmodels in general and on business model innovationin particular. As shown by Zott et al. (2011),researchers’ use of idiosyncratic definitions andcoding schemes hinders cumulative scientific prog-ress. While our coding scheme is highly demandingin terms of information gathering, it allows for ana-

lyzing changes in a wide variety of business modelcomponents across various industries and facilitatesthe development of an encompassing view on thebusiness model as the firm’s underlying architec-ture. We hope it can be an inspiration for otherresearchers.

LIMITATIONS AND FUTURERESEARCH

In discussing our research, some limitations should,of course, be kept in mind. It is clear that one cannotgeneralize empirically from six cases and that thepropositions developed in this article require furthertesting. Our case studies suggest that focused com-mitment fosters short-term growth but limits variety,thereby jeopardizing long-term venture survival. Inaddition, we suggest that simultaneous experimenta-tion restricts initial growth but introduces higherlevels of requisite variety in a resource-effectivemanner which, in turn, fosters long-term survival.Further research is certainly warranted on how dif-ferent approaches to business model developmentinfluence short-term growth and long-term survival.Such assessments would involve translating our casestudy findings into an adequate panel-oriented lon-gitudinal research design comprising both focusedventures and ventures engaged in simultaneousexperimentation (e.g., De Carolis et al., 2009). Atthe firm level, indicators of economic growth couldbe regressed on the number of business models beingexplored simultaneously, controlling for human andfinancial resources invested as well as for comple-mentary success factors identified in previousresearch (such as industry characteristics, foundingteam characteristics, legitimacy, and available assets,including patents, alliances, and geographical loca-tion). Hazard models could be used to assess theimpact of the number of simultaneous businessmodel experiments on survival, again controlling forresources and other firm and industry characteristics.

Moreover, we observe only two distinct (combi-nations of) experimentation approaches in ourcase studies: (1) simultaneous experimentation and(2) focused commitment followed by consecutivebusiness model search. Although we have sampledventures with quite different business model devel-opment trajectories, it goes without saying thatfurther engagement in in-depth case studies mayresult in additional insights with respect to both thedelineation of additional, relevant approaches and



the combination of different approaches over time.For example, it is conceivable that ventures maybegin with a sequence of distant business modelsearches—thereby identifying the most promisingdirection for subsequent development throughrelated business model searches—or they may use asequence of related business model searches toidentify a first client, which puts them in a betterposition to attract external investment and focus sin-gularly on that one specific business model. Othertemporal combinations can be imagined, and furtherresearch aimed at identifying these approaches,their rationales, and their implications seems highlyappropriate.

Finally, our research also points to the need andrelevance of further inquiry into the nature and con-sequences of different types of investor behavior.When and why are investors willing to supportsimultaneous experimentation versus when and whyare they inclined to restrict funding to ventures witha clear focus? (How) does this depend on VC fundcharacteristics and investment managers’ humancapital characteristics, as can be expected from exist-ing literature (Knockaert et al., 2010; Wright,Vohora, and Lockett, 2004)? To what extent do ven-tures with portfolios of simultaneous business modelexperiments affect the overall performance of VCinvestment portfolios? VCs usually spread their riskby investing in several focused ventures with differ-ent business models. Simultaneous experimentation,though, spreads risk within the venture. As businessmodels are in constant flux, the act of balancing riskboth within and between ventures introduces inter-esting research questions on optimizing investmentportfolio management and outcomes. We hope thisstudy contributes to and inspires future research inthis area.

CONCLUSIONS

While existing literature suggests that venturesshould experiment to develop and improve theirbusiness models, it is unclear how they should orga-nize these experimentation processes in an effectivemanner. This study extends organizational learningtheory by analyzing (1) whether ventures use differ-ent experimentation and learning approaches underuncertainty and, if so, (2) what the rationale andimplications of these approaches are.

Based on six longitudinal case studies, we iden-tified two distinctive approaches to business devel-

opment under uncertainty: (1) focused commitmentversus (2) simultaneous experimentation. While theformer is conceptually related to organizationallearning approaches identified in the literature, theidentification of ventures developing a portfolio ofdiverging related searches is less straightforward.These ventures spend from two to five years devel-oping their portfolios of business models throughrelated business model search, which they graduallynarrow down until a viable business model remains.Our research reveals that while focused commit-ment is initially instrumental in acquiring dedicatedresources, this dedication hinders the financing andexecution of subsequent searches. As a result,focused commitment limits the variety of businessmodel experiments and hampers long-term survi-val. Simultaneous experimentation with carefullyselected business models, executed in a disciplinedmanner, however, increases the chance of identify-ing a viable business model and, hence, of survivingin the long run.

Our research extends organizational learningtheory by demonstrating that enacting a variety ofbusiness model experiments does not intrinsicallyrequire distant search. Simultaneous experimenta-tion also allows ventures to explore a broad varietyof business models. In addition, it is far more effi-cient in this respect than focused commitment. Assimultaneous experimentation implies a combina-tion of means-driven experimentation with theconscious planning, execution, and selection ofexperiments, our study is one of the first to illustratethe translation of effectual logic into effectual behav-ior and also reconciles the apparent juxtapositionbetween ‘action’ and planning.’

ACKNOWLEDGEMENTS

The authors would like to thank Dr. Cathy Lecocq forhelp with data collection in the initial phase of thisresearch project. Our thanks also go to Mike Wright andtwo anonymous reviewers for their pertinent commentsand suggestions. Part of this research was made possiblethrough a research grant from C.I.M. Also, the supportof the BNP Paribas Fortis Chair ‘Herman Daems’ isgratefully acknowledged.

REFERENCES

Afuah AN. 2003. Business Models: A Strategic Manage-ment Approach. McGraw-Hill: New York.



Ahuja G, Katila R. 2004. Where do resources comefrom? The role of idiosyncratic situations. StrategicManagement Journal 25(8/9): 887–907.

Alvarez SA, Barney JB. 2007. Discovery and creation:alternative theories of entrepreneurial action. StrategicEntrepreneurship Journal 1(1/2): 11–26.

Ambos TC, Birkinshaw J. 2010. How do new venturesevolve? An inductive study of archetype changes inscience-based ventures. Organization Science 21(6):1125–1140.

Amit R, Zott C. 2001. Value creation in e-business.Strategic Management Journal 22(6/7): 493–520.

Anderson P, Tushman M. 1990. Technological discontinui-ties and dominant designs: a cyclical model of techno-logical change. Administrative Science Quarterly 35(4):604–633.

Andries P, Debackere K. 2007. Adaptation and performancein new businesses: understanding the moderating effectsof independence and industry. Small Business Economics29: 81–99.

Bhide A. 1992. Bootstrap finance: the art of start-ups.Harvard Business Review 70(6): 109–117.

Brown SL, Eisenhardt KM. 1997. The art of continuouschange: linking complexity theory and time-paced evolu-tion in relentlessly shifting organizations. AdministrativeScience Quarterly 42: 1–34.

Clarysse B, Bruneel J, Wright M. 2011. Explain-ing growth paths of young technology-based firms:structuring resource portfolios in different competitiveenvironments. Strategic Entrepreneurship Journal 5(2):137–157.

Cyert R, March J. 1963. A Behavioral Theory of the Firm.Prentice Hall: Englewood Cliffs, NJ.

De Carolis DM, Yang Y, Deeds DL, Nelling E. 2009.Weathering the storm: the benefits of resources tohigh-technology ventures navigating adverse effects.Strategic Entrepreneurship Journal 2(3): 147–160.

Demil B, Lecocq X. 2010. Business model evolution: insearch of dynamic consistency. Long Range Planning 43:227–246.

de Weerd-Nederhof PC, Visscher K, Altena J, FisscherOAM. 2008. Operational effectiveness and strategicflexibility: scales for performance assessment of newproduct development systems. International Journal ofTechnology Management 44(3/4): 354–372.

Dougherty D. 1995. Managing your core incompetenciesfor innovation. Entrepreneurship Theory and Practice 19:113–135.

Dowell G, Swaminathan A. 2006. Entry timing, explora-tion, and firm survival in the early U.S. bicycleindustry. Strategic Management Journal 27(12): 1159–1182.

Dutta DK, Crossan MM. 2005. The nature of entrepreneur-ial opportunities: understanding the process using the4I organizational learning framework. EntrepreneurshipTheory and Practice 29(4): 425–449.

Eisenhardt KM. 1989. Building theories from case studyresearch. Academy of Management Review 14(4): 532–550.

Eisenhardt KM, Graebner ME. 2007. Theory buildingfrom cases: opportunities and challenges. Academy ofManagement Journal 50(1): 25–32.

Fiegenbaum A, Karnani A. 1991. Output flexibility: a com-petitive flexibility for small firms. Strategic ManagementJournal 12(2): 101–114.

Fisher G. 2012. Effectuation, causation, and bricolage: abehavioral comparison of emerging theories in entrepre-neurship research. Entrepreneurship Theory and Practice36(5): 1019–1051.

Gibbert M, Ruigrok W, Wicki B. 2008. What passes asa rigorous case study? Strategic Management Journal29(13): 1465–1474.

Gruber M, MacMillan IC, Thompson JD. 2008. Look beforeyou leap: market opportunity identification in emergingtechnology firms. Management Science 54(9): 1652–1665.

Gunther McGrath R. 1999. Falling forward: real optionsreasoning and entrepreneurial failure. Academy ofManagement Review 24(1): 13–30.

Hamel G. 2000. Leading the Revolution. Harvard BusinessSchool Press: Boston, MA.

Knockaert M, Wright M, Clarysse B, Lockett A. 2010.Agency and similarity effects and the VC’s attitudetowards academic spin-out investing. Journal of Technol-ogy Transfer 35(6): 567–584.

Levinthal DA. 1997. Adaptation on rugged landscapes.Management Science 43(7): 934–950.

Liao J, Gartner WB. 2006. The effects of pre-venture plantiming and perceived environmental uncertainty on thepersistence of emerging firms. Small Business Economics27(1): 23–40.

March J, Simon H. 1958. Organizations. Blackwell:Cambridge, MA.

Mauer R. 2009. Learning from technology ventures: theimpact of causal and effectual behavior on uncertainty. InOfficial Paper Proceedings Academy of Management(AOM) Annual Meeting, Chicago, IL.

Miller D, Friesen PH. 1984. Organizations: A QuantumView. Prentice Hall: Englewood Cliffs, NJ.

Minniti M, Bygrave W. 2001. A dynamic model ofentrepreneurial learning. Entrepreneurship Theory andPractice 25(3): 5–16.

Morris M, Schindehutte M, Allen J. 2005. The entrepre-neur’s business model: toward a unified perspective.Journal of Business Research 58: 726–735.

Nicholls-Nixon CL, Cooper AC, Woo CY. 2000. Strategicexperimentation: understanding change and performancein new ventures. Journal of Business Venturing 15(5):493–521.

Pentland B. 1999. Building process theory with narrative:from description to explanation. Academy of Manage-ment Review 24: 711–724.



Raynor M. 2007. The Strategy Paradox. CurrencyDoubleday: New York.

Read S, Dew N, Sarasvathy SD, Song M, Wiltbank R. 2009.Marketing under uncertainty: the logic of an effectualapproach. Journal of Marketing 73: 1–18.

Sarasvathy SD. 2001. Causation and effectuation: toward atheoretical shift from economic inevitability to entrepre-neurial contingency. Academy of Management Review26(2): 243–263.

Sarasvathy SD. 2008. Effectuation: Elements of Entrepre-neurial Expertise. Edward Elgar: Cheltenham, U.K.

Suddaby R. 2006. From the editors: what grounded theory isnot. Academy of Management Journal 49(4): 633–642.

Vohora A, Wright M, Lockett A. 2004. Critical juncturesin the development of university high-tech spinoutcompanies. Research Policy 33: 147–175.

Westhead P, Wright M. 1998. Novice, portfolio andserial founders: are they different? Journal of BusinessVenturing 13(3): 173–204.

Winborg J, Landström H. 2001. Financial bootstrapping insmall businesses: examining small business managers’resource acquisition behaviors. Journal of BusinessVenturing 16: 235–254.

Wright M, Vohora A, Lockett A. 2004. The formation ofhigh-tech university spinouts: the role of joint venturesand venture capital investors. Journal of TechnologyTransfer 29(3/4): 287–310.

Yin RK. 1994. Case Study Research: Design and Methods(2nd edn). SAGE Publications: Thousand Oaks, CA.

Zott C, Massa L, Amit R. 2011. The business model:recent developments and future research. Journal ofManagement 37(4): 1019–1042.



simultaneous experimentation as a … experimentation as a learning strategy: business model...

Documents