WHITE PAPER JUNE 2011 Aitamurto-Leiponen-Tee brought to you by www.ideasproject.com This whitepaper is written with the quest to understand idea crowdsourcing from the point of view of current scholarly research. The theme is very topical and a lot of progress is excpected in this area in the next few years – both from practical application and from social science research. Nokia’s idea crowdsourding team is looking forward to the shared learning journey with the audience of this report. Pia Erkinheimo, Head of Crowdsourcing, Nokia. The Promise of Idea Crowdsourcing – Benefits, Contexts, Limitations Tanja Aitamurto, Stanford University Aija Leiponen,1 Imperial College London and Cornell University Richard Tee, Imperial College London Introduction Idea crowdsourcing has become a hype term associated with unrealistic expectations for innovation and unclear understanding of its requirements and challenges. This paper intends to bring together and evaluate the key insights that shed light on the phenomenon conceptually or empirically and help us qualify and attenuate the expectations. What do we know about crowdsourcing as of early 2011? We review the concept of idea crowdsourcing, focusing particularly on findings reported in the academic literature. Going beyond studies on the sources of inputs for innovation, we review research published in journals, conferences, and working papers in a variety of fields and disciplines, including strategic management, innovation, information systems, communication, and marketing. Overall, we find that existing work clearly indicates that crowdsourcing potentially contributes significantly to innovation. When applied in the right circumstances, crowdsourcing can deliver considerable benefits to firms in terms of inputs into innovation. However, success requires careful analysis of goals, problem-solving environment, required expertise, firm‟s strategies, complementary activities and capabilities, and the competitive environment. 1 Contact author, email a.leiponen@imperial.ac.uk. The authors would like to thank Pia Erkinheimo for supporting this research and look forward to continued cooperation with the Nokia Crowdsourcing Team. 1 (30) WHITE PAPER JUNE 2011 Aitamurto-Leiponen-Tee brought to you by www.ideasproject.com We also attempt to evaluate the direct costs and benefits of crowdsourcing with the limited data available from a few well-documented crowdsourcing platforms. Meanwhile, there is little empirical research on the indirect strategic value of crowdsourcing such as brand leverage or innovation real options. We see this as a potential avenue for further research. Future research should also seek to quantify the impact of crowdsourcing compared to and complementing more traditional methods of innovation. Furthermore, research on companies‟ absorptive capacity to utilize crowdsourced ideas and to co-create and refine ideas further with customers would benefit both academia and industry. Our report is inspired by Nokia‟s recent and ongoing experiments in this area. The paper is structured as follows. In section 1, we define the concept of crowdsourcing, position it in relation to other literatures, and discuss it in the wider context of open innovation. Section 2 analyzes how crowdsourcing can be utilized and implemented. In section 3 we investigate what existing evidence can tell us about the benefits of crowdsourcing. Section 4 looks at the context in which crowdsourcing can best be utilized. In section 5 we discuss possible downsides and limitations of crowdsourcing and in section 6 how crowdsourcing might be combined with more traditional methods. Section 7 assesses the crowdsourcing business case. In section 8 we look at the methods used in crowdsourcing research. Finally, in section 9 we assess open research questions and conclude in section 10. 1. What is crowdsourcing? Crowdsourcing is a new phenomenon enabled by evolving information and communication technologies (ICT) that has only very recently begun to be investigated by academic researchers.2 Before we further characterize crowdsourcing, it is useful to position the concept vis-à-vis related phenomena. Crowdsourcing is often based on the framework of collective intelligence (Lévy, 1997), the idea that knowledge is the most accurate when it consists of inputs from a distributed population - „all of us together are smarter than any one of us individually.‟ The opposite of collective intelligence is relying on a single agent, for example, a knowledgeable expert. The concept of collective intelligence has been popularized as the wisdom of crowds (Surowiecki, 2004), and crowdsourcing can be defined as a tool to gather collective intelligence for certain tasks. Related concepts to crowdsourcing are co-creation (Prahalad and Ramaswamy, 2000), open innovation (Chesbrough, 2003) and user innovation (Von Hippel, 1976; 1988; 2003) First, as suggested by Brabham (forthcoming), we might position crowdsourcing as one particular form of participatory social media, of which other examples include open-source production, blogging, and video and photo-sharing sites. Crowdsourcing distinguishes itself by involving an organization-participant relationship, where the organization engages in a top-down, managed process to seek a bottom-up, open input by users in an online community. Brabham (forthcoming) identifies four types of crowdsourcing: 2 Howe is generally credited as coining the term “crowdsourcing” in 2006. 2 (30) WHITE PAPER JUNE 2011 Aitamurto-Leiponen-Tee brought to you by www.ideasproject.com 1) Knowledge discovery and management (e.g. Peer-to-Patent Community Patent Review) 2) Broadcast search (e.g. Innocentive) 3) Peer-vetted creative production (e.g. Threadless), and 4) Distributed human intelligence tasking (e.g. Amazon's Mechanical Turk, Microtask) In a related framework, Howe argues that “crowdsourcing isn‟t a single strategy. It is an umbrella term for a highly varied group of approaches that share one obvious attribute in common: they all depend on some contribution from the crowd” (2009: 280). Howe also presents a taxonomy that focuses attention on (1) crowd wisdom (similar to broadcast search above), (2) crowd creation (similar to peer-vetted creative production), (3) crowd voting, including prediction markets, and (4) crowd funding, including crowd-based microlending institutions. Second, from a conceptual point of view, crowdsourcing also overlaps with related notions of collaborative innovation, such as open innovation (see e.g. Chesbrough, 2003) and user innovation (see e.g. Von Hippel, 2005). Following Schenk and Guittard (2009), we might visualize how crowdsourcing relates to other concepts as shown in Figure 1. Here, crowdsourcing is a subset of user innovation, which in turn is a subset of open innovation. The notion of “co-creation” combines user innovation and crowdsourcing and also is a subset of the open innovation concept (also termed “crowd-creation, see Howe 2008). Schenk and Guittard provide an alternative typology of crowdsourcing, distinguishing between integrative and selective crowdsourcing. For integrative crowdsourcing, the goal is to pool vast amounts of complementary information from a large number of users (e.g. OpenStreetMap,Ushahidi). On the other hand, for selective crowdsourcing, the goal is to identify and select input from competing users. The latter type might take the form of idea contest or other types of open competition. Figure 1. Crowdsourcing, Open Innovation, User Innovation and Open Source Open innovation User innovation Co-Creation Crowdsourcing Source: Modified from Schenk and Guittard, 2009 3 (30) WHITE PAPER JUNE 2011 Aitamurto-Leiponen-Tee brought to you by www.ideasproject.com Generally, crowdsourcing can be defined by the following two elements: 1) an open call and 2) a crowd (Burger, Helmchen and Penin, 2010). The open call refers to the fact that, in crowdsourcing, there is no selection mechanism that identifies upfront who the “supplier” of the required content will be (which can be e.g. an idea, solution, prototype, or intellectual property). Participation is non-discriminatory and in principle anyone can answer the call. Given the usage of an open call, “the crowd” will usually be characterized by several features: a large number of participants; heterogeneity of participants (e.g. in terms of knowledge, geographical background etc.), and voluntary participation.3 The alternative to crowdsourcing, in this sense, is outsourcing a task to a specific agent (cf. Afuah and Tucci, 2011). Within the open innovation paradigm, crowdsourcing can be perceived as a tool to gather ideas, innovations, or information for certain purposes. It can thus be viewed as a method of open innovation. As defined by Henry Chesbrough (2006a), open innovation is the use of purposive inflows and outflows of knowledge to accelerate internal innovation and expand the markets for external use of innovation. The concept of „open innovation‟ refers to a new research and development paradigm in which the there are two key factors that are changing the economics of innovation: the increasing costs of research and development and the shortening product lifecycle (Chesbrough, 2006a). To thrive in this challenging innovation landscape, firms are increasingly attracted to use open innovation as a part of their research and development strategy (see e.g., the Procter and Gamble discussion by Huston and Sakkab, 2006). Open innovation means shifting away from the traditional closed innovation system, where innovation processes mainly happen inside the organization and ideas from outside of the organization are often treated with “not-invented-here” mentality. Open innovation establishes new paths to commercialize the innovations done within the company, both by using informal and formal ties to partners, for example through exploiting the possibilities for revenue streams by using open application programming interfaces. (Chesbrough 2003, Simard and West, 2006; Aitamurto and Lewis, 2011). Chesbrough (2011) applies the open innovation paradigm in the service business and claims that in numerous industries, including telecommunications, service ecosystems are offering a competitive edge to traditional product-based businesses. Products, whether traditional goods or services, rapidly become commoditized, and companies can find competitive advantage by differentiating their products in a co-created process with customers. Furthermore, customers are demanding more customized products, and to meet this need, companies may need to use co-creation to build on the tacit knowledge from customers‟ experiences. Co-creation is another concept related to crowdsourcing, and perceived as a tool in the open innovation toolkit. Co-creation is a joint effort of the producer and the customer to develop new products or services (Prahalad & Ramaswamy, 2000; 2004), involving a two-way 3 What constitutes “large” or “heterogeneous” is relative to the number of participants when the task would be conducted in a traditional manner. 4 (30) WHITE PAPER JUNE 2011 Aitamurto-Leiponen-Tee brought to you by www.ideasproject.com interaction between customers and companies, as well as peer-to-peer communication among customers. In marketing, co-creation is seen as a new branding paradigm (Hatch & Schultz, 2010), which “breaks free from the industrial age paradigm for branding”. In this scheme, the meaning of value and the process of value creation are evolving from company- and productbased value to personal consumer experience (Prahalad and Ramaswamy, 2004). For example, Lego Group has successfully built a brand community, in which Lego users interact about the product and the brand and design new products to Lego group (Hatch & Schultz, 2010). This initiative has resulted in entire school curriculum developed to teach children robotics using Legos, and new products such as the Architecture line, in which Lego builds kits imitating famous architectural designs, often co-created with users (Chesbrough, 2011). In this paradigm, the company sees the value of process-oriented offerings, instead of focusing only on the traditional product-oriented offerings. In the new approach, value is created also in the process, in which customers interact with the company and with each other, the shared object being the company‟s product, or, in the wider perspective, the brand. This approach has been called User-Generated Branding (UGB) in brand management literature. Here, the brand is created in interaction with customers rather than in the outside-in flow of building the brand image. UGB can be defined as a result of voluntarily created and publicly distributed brand messages, from comments to reviews, ratings, and blog posts. However, UGB does not refer to a co-design process in which solutions are sought for defined problems, but to personal brand meaning. In this approach, the users are considered as creative consumers (Burmann, 2010). To operationalize the concepts of open innovation, crowdsourcing, and co-creation, let‟s discuss a real-world example. Procter and Gamble has successfully shifted its R&D strategy to open innovation, according to Huston and Sakkab (2006). P&G started looking for innovations more actively outside the company instead of investing only in internal R&D work. The core of the new "connect & develop" strategy is to find connections to externally sourced ideas to develop highly profitable innovations, through both proprietary networks and open platforms such as NineSigma and InnoCentive, etc. Initially, when launching the new strategy, the goal was to acquire 50% of innovations outside the company. The authors state that after implementing the strategy, 35% of company's new products originate from outside the company, and 45% of the initiatives in the product development have key elements that were discovered externally. R&D productivity has increased by nearly 60%, and R&D investment as a percentage of sales is down from 4.8% in 2000 to 3.4%. The case describes some best practices how to implement the strategy in finding adjacencies to existing products, and using technology game boards to evaluate the consequences of technology acquisition in the product portfolio. Crowdsourcing in the Procter & Gamble case was used as a way to gather solutions to scientific problems the company posted on InnoCentive. The company had a systematic approach to the open innovation strategy, which emphasizes crowdsourcing as one tool in the company‟s open innovation toolkit. The term crowdsourcing itself is being debated. Many prefer the broader term co-creation, because it suggests a two-way interaction. The term crowdsourcing often seems refer to oneway interaction whereby individuals submit information or ideas to a specific task. For well- 5 (30) WHITE PAPER JUNE 2011 Aitamurto-Leiponen-Tee brought to you by www.ideasproject.com defined problem solving (e.g. Ushahidi), this may work, but for more creative ouputs, or seeing value in the interaction process with the customers themselves, crowdsourcing may be a too narrow term. In contrast, co-creation refers to collaboration rather than to harvesting ideas from the public. Moreover, as the nascent market for labor crowdsourcing (e.g. Mechanical Turk, SamaaSource, Crowdflower, Microtask) is growing and the supply of crowdsourcing work intermediaries becoming more plentiful, a political debate of labor exploitation has heated up. In addition to the term co-creation, mass collaboration has gained popularity in referring to processes where large numbers of individuals may participate and contribute to innovative activities. 2. How can crowdsourcing be utilized and implemented? There are several ways through which firms or organizations can deploy crowdsourcing strategies. Broadly speaking, we can distinguish between internal, firm-developed crowdsourcing initiatives to leverage the knowledge within a company, as opposed to crowdsourcing with external collaborators. External collaboration can be implemented on companies‟ virtual communication environments (VCE‟s), such as Lego Group‟s Lugnet, Dell‟s IdeaStorm, Starbucks‟ MyStarbucksIdea, or Nokia‟s ideasproject. Alternatively, collaboration can be outsourced to specialist service vendors, such as InnoCentive or NineSigma, which other firms can also use. These intermediaries offer the co-creation process either for anybody (Jovoto, OpenIdeo), or for a pre-defined group of participants. Furthermore, crowdsourcing efforts can happen in either temporary innovation challenges or more permanent and ongoing VCE settings. Additionally, when companies use crowdsourcing as their fundamental strategic function (e.g., Threadless, the t-shirt company), this can arguably be identified as yet another form of crowdsourcing. Below we illustrate, first, internal idea-crowdsourcing initiatives, specifically IBM‟s Innovation Jam and the SAPiens platform at SAP, and then crowdsourcing initiatives with external collaborators at Dell‟s IdeaStorm-project and Salt Lake City‟s design contest. Based on an in-depth case study, Bjelland and Wood (2008) analyze how IBM leverages its firm-wide intelligence located at geographically dispersed sites through a process called “innovation jams”. The innovation jams serve as tools to promote ideas internally to other units or divisions and to win funding for projects. Overall the process focuses on goal identification, idea generation and commercialization. It consists of several stages: · Planning, which focuses on goal and topic identification, and preparation of infrastructure (e.g. websites, wiki's) · Jam Phase 1 (72 hour period online, to allow collaboration across time zones and geographies) followed by a face to face review session (review "big ideas" by 50 senior exectutives) · Jam Phase 2: refinement of ideas in phase 1, again followed by review stage using eclustering and human review. · Propose new businesses, funding for new business units. 6 (30) WHITE PAPER JUNE 2011 Aitamurto-Leiponen-Tee brought to you by www.ideasproject.com Another example of an idea competition is SAPiens, a web-based idea competition initiated by software producer SAP (Leimeister et al., 2009). The design of SAPiens was informed by earlier idea competitions (see Table 1). The authors of the study focus on incentives (learning, direct compensation, self-marketing, and social motives) and "activation supporting" elements (focused on collecting, relating, creating, and donating ideas) and how these influence motivations of participants. They survey a randomly selected sample of student SAP users who submit ideas to improve SAP's software, and analyze students‟ motivations to participate. They find that appreciation of the organizer scores particularly high. Table 1. Selected Examples of Ideas Competitions Source: Leimeister et al. (2009) Benbya and Van Alstyne (2011; see table 2) studied a broader set of internal knowledge markets focusing on how these facilitate information sharing in large organizations. Based on interview data with over 30 companies the authors provide lessons to design an effective internal knowledge market. They distinguish three phases: launch, development and evolution. For each phase, they highlight both the challenges involved (e.g. lack of critical 7 (30) WHITE PAPER JUNE 2011 Aitamurto-Leiponen-Tee brought to you by www.ideasproject.com mass of content), as well as the response the firm might take (e.g. seed the market with critical knowledge). Table 2 How to design an effective internal knowledge market Source: Benbya and Van Alstyne (2011) To understand when firms should use firm-owned as opposed to intermediary platforms, Terwiesch and Xu (2008) analyze what the appropriate mechanisms are for different innovation types, focusing on costs and the number of problem solvers. Distinguishing between internal development and open innovation via firm-owned or intermediary platforms, they develop a model that classifieds projects according to technical and market uncertainty. For each project they analyze the appropriate reward structure (winner-take-all vs. multiple-prize; performance-contingent vs. fixed-price) in relation to firm profit. From a profit perspective, winner-takes-all rewards outperform multiple-prize structures for both 8 (30) WHITE PAPER JUNE 2011 Aitamurto-Leiponen-Tee brought to you by www.ideasproject.com ideation and trial-and-error projects, whereas performance-contingent structures outperform fixed-price rewards. Dell IdeaStorm is an example of co-creation with customers in a permanent virtual collaboration environment (VCE). Dell IdeaStorm launched four years ago, in February 2007. By February 2011, there has been about 15,400 ideas posted on the platform, 740,000 votes have been cast on the platform, and 91,000 comments have been posted. Through IdeaStorm, end users contribute their business ideas - either about new products or improvements to existing Dell products - to be reviewed, discussed, and voted upon by the user community. A case study by Di Gangi and Wasko (2009) analyses the factors which influence an organization‟s adoption decision when innovations come from outside of organization‟s boundaries, from an external innovation community. The study concludes that the decision to adopt ideas from the innovation community depends on the complexity of the innovation and Dell's ability to reduce the innovation complexity through its absorptive capacity. The study indicates that Dell‟s decisions to whether absorb or reject an idea is affected by the reasoning and pressure performed by the user community, and the amount of data they provide to both reduce the ambiguity of the idea and to bridge the knowledge gap between the users and the company. In order to create and maintain successful innovation communities, the study encourages companies to further develop their capability to understand the behavior of user communities so that the companies could more effectively gain benefit from their VCE efforts. However, a quick browsing of the current ideastorm website suggests that the ideas largely consist of regular customer feedback rather than more elaborate or sophisticated business or innovation ideas. Numerous non-profit organizations, including public and government organizations, are also using open innovation model in their product development processes. In the United States, the Federal Transit Administration (FTA) pilot project focused on bus stop design project called "Next Stop Design" in Salt Lake City, Utah. The goal was to collaborate with a neighborhood's residents to design a better functioning bus stop. 47 designs were submitted to the design challenge over the course of the study. The study concludes that crowdsourcing is effective in transit planning. Open method brings variety to the designs, and the designs include both amateur and professional-level submissions. However, preliminary findings suggest that once the quality of submissions rises to professional level, the amateurparticipants might not see the value in their participation. The case highlights the need to carefully define and communicate the goal of the project, whether it is to get more innovative ideas for new products, to harvest the most ready and refined ideas to be manufactured by the organization, or to establish co-creation as an integral part of value chain. This is a critical issue in planning crowdsourcing initiatives, because the goal determines how the initiative should be executed. Our last example is an entire business built on crowdsourced submissions. The company is called Threadless, a firm that crowdsources T-shirt designs from the public. Anybody can submit a t-shirt design on Threadless‟ open platform, and users can also vote for the designs they like. The most popular designs are manufactured, and usually the designs sell out because people who have voted order the designs they have supported. Threadless has managed to create a lucrative business with high profit margins. Brabham (2010) identifies 9 (30) WHITE PAPER JUNE 2011 Aitamurto-Leiponen-Tee brought to you by www.ideasproject.com four primary motivators for participation at Threadless: the opportunity to receive prize money, the opportunity to develop one‟s creative skills, the potential to take up freelance work, and the love of community at Threadless. An additional source of motivation for collective intelligence participation is “glory”, as suggested by Malone et al. (2010). This is the case when individuals respond to the opportunities to be recognized by peers in a community. Brabham also discusses the role played by addiction that makes the participants to return to the Threadless site. The author suggests that when using crowdsourcing, organizations should develop more deliberate means for the crowd to support problem-solving missions, to contribute to the public good and express their addiction to – or love for – the project, product, or activity. Table 3. Examples of crowdsourcing projects Source: Schenk and Guittard (2009) 10 (30) WHITE PAPER JUNE 2011 Aitamurto-Leiponen-Tee 3. brought to you by www.ideasproject.com Does crowdsourcing work? Even though the potential benefits of crowdsourcing are intuitively compelling, there has been surprisingly little concrete evidence of how crowdsourcing performs compared to established alternatives. There are several important exceptions however, which we discuss below. A study by Poetz and Schreier (2011) convincingly demonstrates the potential value of crowdsourcing in a real-world setting, a firm specializing in baby products. The authors compare the quality of ideas generated by users to those of professionals (e.g. marketers, engineers, designers). The quality of ideas was measured in terms of 1) novelty, 2) customer benefit, and 3) feasibility (technically and economically), and they were evaluated by firm executives in a blind survey, in other words, the raters were not aware of the source of the idea. The study finds that user-generated ideas score significantly higher than those generated by professionals. Interestingly, whereas user ideas scored somewhat lower in feasibility (but not so much that they were seen to constitute a barrier), they were more frequently rated as among the best on novelty and customer benefit. As such, at least in this context, idea crowdsourcing appears to be a very useful complement to internally developed ideas. Future research might shed more light on the extent to which the findings from this particular setting (i.e. technologically relatively simple products) translates to other domains. A defining characteristic of crowdsourcing is an emphasis on “openness”, assuming that heterogeneity of participants will lead to better outcomes. A study by Jeppesen and Lakhani (2010) provides evidence that broadening participation can indeed be beneficial. In their study, the authors analyze how the margina (peripheral) nature of problem solvers relates to effectiveness in problem solving in open R&D calls. The study focuses on a data set of InnoCentive.com, a platform where scientists compete to solve real-world challenges formulated by clients of InnoCentive. The dataset covers 166 problems, coming from R&D labs of 26 firms in 10 countries. The analysis demonstrates that technical and social marginality is positively related to successful problem resolution. Here, technical marginality is measured based on a self-assessed distance of the problem field in terms of technical expertise. Social marginality is measured by being female, a proxy for distance of the establishment in the science community. The findings highlight the importance of broadening participation to address solving problems; as such it demonstrates the value of openness, by disclosing problems and removing barriers to entry to non-obvious participants. 4. When does crowdsourcing work? As explained in section 1, a central feature of crowdsourcing is using an “open call”, where a large section of the public can volunteer to participate. To understand whether crowdsourcing is a useful solution to a business problem, we first need to understand whether the problem is best addressed by an “open” solution. A study by Pisano and Verganti (2008; see figure 2) sheds more light on this topic. Here, the authors provide a framework to understand what collaboration mode is most suitable depending on the situation. They distinguish two key dimensions, participation (open or closed) and governance (hierarchical or flat). Most 11 (30) WHITE PAPER JUNE 2011 Aitamurto-Leiponen-Tee brought to you by www.ideasproject.com relevant to crowdsourcing are the open participation modes: hierarchical ones are labeled "innovation malls" (e.g. InnoCentive), and flat ones "innovation communities" (e.g. Linux). The general advantage of the open mode is to attract a wide range of ideas beyond the focal firm's domain. Challenges include the costs involved screening the ideas, as well as that the best idea generators may prefer closed networks, since their ideas are more likely to be realized that way. Generally, the open mode is the most suitable when proposed solutions can be cheaply evaluated, or when user needs are unclear. The authors also recommend choosing the mode that is best suited to the organization's own capabilities. If an “open” process is clearly the best approach, the next question is how to organize this distributed innovation process. An important question in this respect is whether to make use of collaborative communities or competitive markets, a topic addressed by Boudreau and Lakhani (2009; see table 4). They find that communities are especially useful when an innovation problem is based on cumulative knowledge, i.e. when it continually builds on past advances. On the other hand, markets are more appropriate when an innovation problem is best addressed by broad experimentation. Furthermore, communities are more focused on intrinsic motivations of external innovators (e.g. the desire to be a part of a larger cause). On the other hand, markets usually reward extrinsic motivations (such as financial compensation). The authors integrate these findings with the platforms and business model literature, illustrated with several empirical examples. A large number of participants involved is also a key feature of crowdsourcing. However, as highlighted in the following section, large numbers of ideas are not necessarily beneficial to firms, and might in some cases constitute a major burden. To better understand how quantity of participants affects quality, a study by Boudreau, Lacetera and Lakhani (2011) examine how the size of innovation contests affects outcomes. Their study is based on a dataset of Topcoder.com, a site specialized in software programming contests. It focuses in particular on the effect of competition size (i.e. number of participants) on overall quality of solutions. On one hand, a larger number of participants might diminish incentives because of the lower likelihood of success for any individual. On the other hand, a larger number of participants might increase the likelihood of finding an optimal solution. Both effects are found in the dataset. The nature of the problem (in terms of degree of uncertainty) acts as a moderator - in particular, multi-domain problems (which have a higher degree of uncertainty) benefit more from increased numbers of participants, and also exhibit a lesser negative incentive effect. As such, the study provides insights to managers who design innovation contests: adding more participants is likely to lead to improved contest performance, but only when problems are highly uncertain and require a greater breadth of search for the best approach or path to a solution. 12 (30) WHITE PAPER JUNE 2011 Aitamurto-Leiponen-Tee Figure 2. How to choose the mode of collaboration Source: Pisano and Verganti (2008) brought to you by www.ideasproject.com 13 (30) WHITE PAPER JUNE 2011 Aitamurto-Leiponen-Tee brought to you by www.ideasproject.com Table 4. Examples of alternative platform business models Source: Boudreau and Lakhani (2009) When setting up a crowdsourcing effort, firms also need to decide whether to include experts in the process. Analyzing the costs of two crowdsourced encyclopedia initiatives, Citizendium and Wikipedia, a study by O‟Neil (2010) compares the effect of having credentialed and identified experts in a leading role in Citizendium, versus Wikipedia where the model is based on anonymity and anti-credentialism. Citizendium was founded by a Wikipedia co-founder, Larry Sanger, who has been emphasizing the role of expertise in online mass-collaboration. In Citizendium, the experts (called Editors) self-certify by providing their resumes and relevant links to their work. The experts work with regular, uncertified collaborators, and their role is to write guiding articles, advise contributors, and solve disputes. When comparing these two cases, it appears that Citizendium failed to attract users: In January 2010 Citizendium listed 65 editors and 3,492 authors, whereas the English Wikipedia listed 1,700 administrators and 11,380,900 registered users. Nevertheless, success 14 (30) WHITE PAPER JUNE 2011 Aitamurto-Leiponen-Tee brought to you by www.ideasproject.com doesn‟t come without challenges, and O‟Neil lists three types of direct costs the Wikipedia non-expert approach: uncertainty, lack of perspective, and irresponsibility. The greatest indirect cost is fighting among Wikipedia contributors. To analyze and classify different types of crowdsourcing, Malone et al. (2010) argue that to classify different types of collective intelligence, it is useful to analyze four dimensions: What, why, who, and how. “What” refers to the goals of the activity; “why” refers to the incentives to participate; “who” is the staffing of the activity; and “how” explains what structure and process are followed. Goals may include collective creation or decision making, and process can be independent or interdependent. The intersections of these characteristics generate a 2x2 matrix with the following cells: Collection (including contests; e.g., Flickr, Threadless, InnoCentive), collaboration (e.g., open-source projects), individual decisions (e.g. online markets), and group decisions (e.g., voting; prediction markets). Table 5. Structures for crowdsourcing Independent (Inter)dependent Create Collection Collaboration Decide Individual decisions Group decisions Source: Malone et al. (2010) Applying the above framework shows that examples such as Threadless and InnoCentive include both creation and decision tasks. Both services‟ mode of collection of creative solutions is a contest, and they are associated with monetary motivation (rewards). The decision tasks include choosing among the proposed solutions, and here is where their structures diverge: At InnoCentive, it is the management of client companies who decide the “beauty contest”, whereas, at Threadless, even this part of the process is delegated to the crowd. The latter uses an averaging/voting process that is motivated by “love”, i.e., voluntary contributions based on intrinsic motivation and enthusiasm for the community. Similarly, the process utilized by Wikipedia can be classified according to this framework. Its tasks include both creation and decision (editorial decisions about inclusion of and updates for articles); most tasks are done by the crowd, although the final decision about deleting articles is done by an administrator through an existing insider hierarchy; and all activities are motivated by “love” (intrinsic motivation for knowledge) and “glory” (opportunities to advance in the insider/intellectual hierarchy). According to these authors, the prerequisite for engagement with crowdsourcing is that the resources needed to complete the task are widely distributed or not known. They also suggest that “love” is a powerful motivator but that money and glory may speed up activities (but not always). However, there is not enough systematic research on which motivator to utilize in different circumstances, as money (or even glory) can sometimes alienate intrinsically 15 (30) WHITE PAPER JUNE 2011 Aitamurto-Leiponen-Tee brought to you by www.ideasproject.com motivated and perhaps more creative participants, whereas appealing to intrinsic motivation may backfire if the firm is perceived as trying to get people to work for them for free. Table 6. When are different forms of collective intelligence useful? Source: Malone, Laubacher and Dellarochas (2010). According to the Malone et al. framework, the “collection” mode of creation works best when the problem can be partitioned such that individuals can provide meaningful solutions. Those solutions can be obtained through a “contest” when only one or few are needed. In contrast, “collaboration” modes of creation are appropriate when the task cannot be partitioned in this way but it is possible to modularize the project sufficiently to allow people to work separately but contribute to the shared goal. Finally, group decision-making can be categorized into voting, averaging, and prediction markets. Whereas the voting mechanism is well understood based on its offline applications, averaging can be useful if the decision to be made can be summarized as one number and when the crowd is unlikely to be systematically biased. If there is reason to suspect biases, prediction markets may be more accurate. Here, participants buy and sell “shares” of predictions about future events. They may be rewarded with real money or other prizes. For example, prediction markets can be used to predict the delay or success of projects. Malone et al. (p. 9) provide an example from Microsoft predicting the delay of a major project. Another example of a successful prediction was the Intrade market that correctly predicted the 2006 16 (30) WHITE PAPER JUNE 2011 Aitamurto-Leiponen-Tee brought to you by www.ideasproject.com elections in the US, where, somewhat surprisingly, Democrats gained both houses of Congress (see Bonabeau, 2009). Reviewing the large literature on innovative search, Afuah and Tucci (2011) emphasize the unique ability of crowdsourcing to convert “distant” search (search for information, ideas or solutions in a field not very familiar to an individual or a group) to “local” search (search in domains where the underlying knowledge is well known and can be built on). Crowdsourcing should thus be utilized in problems where relevant knowledge is likely to be dispersed and particularly distant from the innovating organization. Afuah and Tucci also suggest that the nature of the relevant problem-solving expertise, the nature of the problem itself, and the characteristics of potential problem-solving agents determine whether crowdsourcing should be used for a specific task. In summary, these authors propose a number of factors that increase the likelihood that crowdsourcing is feasible (see table 7). Table 7. Factors increasing the probability that crowdsourcing is a feasible approach Firm‟s own expertise is low Likely availability of expertise in the crowd is high Firm‟s own expertise is distant from the likely solution to the problem The problem to be solved is NOT tacit, immobile, unique, or complex Relevant expertise is dispersed The problem is modular Relevant expertise to solve the problem is tacit, immobile, unique, or complex IP associated with the problem (not necessarily with the solution) is protected Firm‟s relevant complementary assets are valuable or inimitable Firm has a monopoly/monopsony position Source: Afuah and Tucci (2011) In summary, these factors describe the relevant characteristics of the firm itself, the crowd, the problem, the required expertise, associated IP conditions, and the firm‟s competitive position. Of course, many of these factors are not necessarily explicitly known, and the firm needs to form assumptions or make educated guesses about them up front. Nevertheless, these feasibility factors may be useful to consider in deciding about engagement in crowdsourcing. 17 (30) WHITE PAPER JUNE 2011 Aitamurto-Leiponen-Tee 5. brought to you by www.ideasproject.com When should crowdsourcing NOT be used? In the previous sections, it has become clear that there are several potential benefits associated with crowdsourcing. We have also discussed situations in which crowdsourcing might best be utilized. It is also important to understand the limitations of crowdsourcing and its potential downsides. First, Burger, Helmchen and Penin (2010) provide a conceptual discussion to understand the limits of crowdsourcing, focusing on crowdsourcing inventive activities (abbreviated as "CIA"). Drawing on transaction cost economics and evolutionary theories of the firm, they suggest that CIA (as opposed to crowdsourcing routine work or content) is only feasible when a problem is clearly defined (i.e. highly codified) and can be legally protected (e.g. via patents). These conditions are strongly present in e.g. in science-based industries such as chemistry and pharmaceuticals, as well as software development. Conversely, the authors predict that in sectors where clear knowledge codification is more challenging and protection mechanisms absent, crowdsourcing inventive activity will not occur. Second, Alexy, Salter and Criscuolo (2010) highlight the potential problems firms may face in crowdsourcing. The authors focus on the specific challenges of incorporating unsolicited ideas generated by end-users to commercial firms. The authors use exploratory interviews and a web-based analysis of management practices to understand the costs involved, including managerial attention (e.g. identifying internal expertise) and legal issues (e.g. IP ownership). Contrary to existing work that has mostly emphasized the upsides to utilizing external ideas, this study highlights the challenges involved in this process. The authors uncover three distinct processes, labeled signaling, structuring, and selecting, that firms can use to improve the efficiency and efficacy of incorporating unsolicited ideas (see figure 3). In the study based on interviews with 48 interviews in 31 European industrial firms, Lichtenthaler (2010) analyzes the potential risks of open innovation, more generally, and associated managerial countermeasures. For example, external technology exploration (including crowdsourcing) could lead to problems such as deficiencies in internal development of critical technological knowledge; reduced possibilities to develop or renew core competencies based on such knowledge; problems in identifying external sources of knowledge; and negative feedback effects from internal R&D employees because of their potentially lower level of motivation. The author concludes that managerial countermeasures are needed for firms to fully profit from open innovation activities. In the area of technological exploration these might include continued and sufficient investments in internal R&D and explicit efforts to find strategic complementarities between internal and external exploration activities. 18 (30) WHITE PAPER JUNE 2011 Aitamurto-Leiponen-Tee brought to you by www.ideasproject.com Figure 3. The unsolicited ideas processes, issues, and potential solutions Source: Alexy, Salter and Criscuolo (2010) Finally, we have found little information on legal issues directly related to crowdsourcing. A partial exception is a study by Alexy et al. (2010), who highlight three implications related to acquiring and integrating these unsolicited ideas: exclusion, ownership and usage, and misappropriation. First, the possibility of excludability is a key issue if gaining intellectual property rights is relevant to the firm. This means that the idea has to be sufficiently novel, for instance, it has not been publicly revealed earlier. Second, getting ownership of the idea so that the firm is able to use it is a non-trivial issue. In some cases, users may freely forfeit their ownership rights. In other cases, the idea submitter might want to sell the idea as intellectual property, in which case complex negotiation processes might need to take place. Third, misappropriation can be an issue when a firm is working on an idea independently of a similar one that has been externally submitted. Establishing this proof can be a costly process, in terms of having a well-documented “paper trail of novelty” (Alexy et al. 2010). However, in areas where innovations are moving away from high-level system designs (e.g. computer manufacturers, phone manufacturers) and their proprietary components (Pisano, 2006), to new ecosystems e.g. based on services, the question of intellectual property rights in the traditional sense can become less relevant. Either way, firms should strategically maximize the value of IP rather than maximizing the control of IP. Relatedly, Afuah and Tucci (2011) suggest that the more the IP related to the firm‟s problem is protected, the more likely they will outsource (crowdsource) the problem. Furthermore, if 19 (30) WHITE PAPER JUNE 2011 Aitamurto-Leiponen-Tee brought to you by www.ideasproject.com the firm has complementary assets (capabilities that facilitate commercialization of the underlying business idea) for which it can protect the IP or that are otherwise difficult to imitate, the more feasible it is to use crowdsourcing for problem-solving or ideation. Similarly, a monopolistic position in an industry segment might facilitate crowdsourcing strategies because there is relatively little threat of imitation. As a more practical example, InnoCentive has very clear guidelines and services related to IP transfer from problem solvers to solution seekers. The types of challenges at InnoCentive include reduction to practice, where solvers provide original research data or experimental results; and paper/theory, where solvers provide theoretical papers and research proposals. More novel categories include ideation, where solvers would provide only a brief submission and it is guaranteed that at least one solver would receive the prize. IP-wise this form of challenge is unique, however, as seekers receive nonexclusive licenses to use any of the submitted ideas, independent of their winning or not. Finally, electronic requests for proposals involve seekers posting product or service requirements and solvers bidding to provide the most cost-effective solution. Here, contractual terms are negotiated directly between seekers and solvers after the bidding. InnoCentive challenges impose a strong form of anonymity, whereby seekers and solvers do not know anything about each other ex ante. During the challenge, solvers provide a temporary license for seekers to evaluate the solution according to the InnoCentive rules. For selected solutions IPRs are then transferred to the seeker, but for all other solutions, seekers cannot use any part of the solution in the future, and InnoCentive has a right to audit this. Around 200 solvers typically express interest in any given challenge, and on average ten solutions are received per challenge. As Lakhani (2008) suggests, seekers are thus presented with ten parallel experiments in problem solving. In terms of solvers‟ benefits, although the monetary benefits are important, intrinsic motivation is also stimulated by the problemsolving experience. 6. How can crowdsourcing be combined with other forms of innovation ideation? In addition to the potential benefits and downsides associated with crowdsourcing, it is important to emphasize that crowdsourcing can also be incorporated within existing (more traditional) innovation practices. Several authors have illustrated how crowdsourcing can complement existing firm practices. Magnusson (2009) analyzes the role of ordinary users in the ideation process of technologybased services, focusing on mobile telephony services. The study employs a comparative quasi-experimental design to simulate three scenarios of user involvement: (1) no user involvement: ideation by professional service developers from TeliaSonera; (2) guided users: technically skilled and ordinary users instructed about the underlying technology; and (3) pioneering users: ordinary users without training or with creativity training. The analysis focuses on the quality of ideas in terms of user value, “producibility” (i.e. feasibility), and 20 (30) WHITE PAPER JUNE 2011 Aitamurto-Leiponen-Tee brought to you by www.ideasproject.com originality (as well as combinations of these). The main finding is that professional developers‟ ideas score the highest on producibility; guided users‟ ideas score on user-value and producibility; and pioneering users‟ ideas score the highest on originality and user-value. These results are thus quite highly aligned with those reviewed earlier by Poetz and Schreier (2011). As such, users can fulfill different roles in ideation, and they can be utilized in different ways depending on the goal of the ideation process. This emphasizes the need to design the potential crowdsourcing activity with a view to the rest of the innovation process. Piller and Walcher (2006) discuss methods to integrate users in new product development, focusing on "toolkits," i.e. specific processes for idea competitions. Their exploratory analysis is based on a pilot study conducted at sporting goods firm Adidas Salomon in Germany. The authors map idea competitions in terms of two dimensions: task specificity (low for open tasks, high for specific tasks) and degree of elaboration of user input (low for simple ideas, high for solutions). The Adidas case scores low on both dimensions, i.e. users generate ideas (as opposed to solutions) in open (as opposed to specific) tasks. The pilot involved 136 customers, whose output (creativity of the idea) was measured based on a standardized creativity (CAT) score, measuring originality, usefulness, and level of elaboration. Among the highest scoring ideas, two reached the implementation stage. Overall, the authors suggest that a toolkit for idea competitions involving users can serve as a useful complement to traditional New Product Development (NPD). 7. Crowdsourcing business case As the available literature does not contain much about the actual costs and quantifiable benefits of in-house crowdsourcing (as opposed to that conducted using service intermediaries), we decided to triangulate the cost of running such a service through analysis of existing services. InnoCentive being the most well known and studied crowdsourcing service, it makes sense to explore their financials and economic impact. The HBS InnoCentive case study by Lakhani (2008) shows that the company has run about 930 challenges between 2000 and 2008; this is approximately 100 challenges per year. 45% of these have been solved and the total cash prizes awarded amount to $4.7 million, which implies about $9000 per solved contest. In return for the prize, in most cases, the seeker firm receives IP rights to the solution. Each seeker pays a posting fee of $15,000 up front and receives help defining the problem and support during the solution evaluation process. In successful challenges, InnoCentive also retains a commission, the size of which is not revealed in the HBS case, although a case study by Forrester Consulting (2009) calculates an average commission/”finder‟s fee” of $6500 per challenge. These data combined with additional information in the case suggest that in 2008-2009, the revenue side of the InnoCentive business model generated about $3 million per year from posting fees and unknown amounts (but probably less than the posting fee revenue; assuming mean commission of $6500, they would probably generate in the order of 90*$6500=$585,000) from commissions. With a 30 strong workforce, the calculation suggests that the firm may still be making a loss, or else the average numbers don‟t accurately reflect the latest years. In 2008 the company received additional venture capital 21 (30) WHITE PAPER JUNE 2011 Aitamurto-Leiponen-Tee brought to you by www.ideasproject.com funding of $6.5 million. This is quite late in the startup process, but was probably justified by the rapid growth of posting revenue (most recently 100% annual growth rate) and the potential of a user base of 140,000 problem solvers, which needs to be efficiently monetized. The InnoCentive website also posts (not surprisingly) highly positive client case studies by Forrester Consulting that indicate the total ROI from InnoCentive challenges for a large consumer products organization has been about 74% (quantifiable benefits). In addition to cost savings and resource savings in R&D, the client company achieved access to a more diverse network of experts; fostered a more innovative culture; improved its ability to frame R&D problems; and achieved a smoother IP transfer process. In summary, total quantifiable benefits amounted to $1.3 million and total costs to $745,000, mostly consisting of challenge posting fees, but also including internal costs of setting up challenges (usually 40h of work per challenge). In another case study using the same methodology, Forrester found that the ROI was 182% for a large multinational agricultural company. These cases suggest very positive outcomes are possible although the returns are probably quite highly variable. However, no negative ROI cases were published on the InnoCentive.com website, for obvious reasons. Another well-known and thoroughly-analyzed crowdsourcing case is Threadless.com that has a user community of 700,000 individuals who contribute t-shirt designs, give each other feedback, vote on the best designs, and eventually buy those that the company decides to produce (based on votes and staff opinions). The company gives cash prizes of $2500 to the winners of weekly contests and does not retain IP rights to the produced designs, so designers are free to use their creations in other applications. Although the company says it pays around $1.5 million to designers on an annual basis, with over 130,000 designs submitted every year, the probability of winning the prize is less than 1%. It is thus clear that the extrinsic reward cannot be driving the enthusiasm and dedicated participation of users, many of whom are professionally trained in arts and design. Meanwhile, the business model has turned out to be very profitable. Harvard Business School multimedia case on this company (Lakhani and Kanji, 2008) suggests that in 2007, with $23 million in sales, the company raked in about $7 million, or 30% in profits. The cost structure estimated by the case writers indicates that 30 full-time employees, most of whom work at the warehouse, run the operations. The IT infrastructure costs around $750,000; warehouse operations $750,000; and shirts themselves cost $7 each (1.5 million shirts sold in 2007). Additionally, numerous co-creation and ideation platforms such as Jovoto offer services to set up and run innovation challenges, the cost ranging from $40,000 to $60,000. The cost includes usually the whole “challenge package” from community organization and management to running the challenge. Although these numbers do not provide a clear answer to the question of how beneficial and costly might an in-house crowdsourcing facility and activity be, they suggest a few possible interpretations. First, the InnoCentive-type crowdsourcing service can be very helpful for certain types of problems, but with the cost vs. impact suggested in the Forrester research, it might not be worthwhile to organize such activity internally. For one thing, the consulting and training provided by InnoCentive appears to be a very important part of the service, and for any other company, developing this expertise internally might take quite long because an 22 (30) WHITE PAPER JUNE 2011 Aitamurto-Leiponen-Tee brought to you by www.ideasproject.com in-house team would probably not be exposed to hundreds of challenges in a short amount of time. Second, the sunk investment in InnoCentive systems has most likely been substantial, whereas the payoff has taken years to emerge. For example, building a user community of 140,000 experts, of which 40% have PhD degrees would not be feasible for an individual company seeking to engage an external community in its own innovation activities. On the other hand, a very different type of a community at Threadless is meaningfully contributing to a very profitable continuous t-shirt design contest. This product is very simple and the design space clearly constrained, which facilitates broad and also non-expert participation in creation and voting. Nevertheless, the enthusiasm of the community is exceptional and the company takes great care to balance business imperatives with legitimacy and trust of the community. To summarize, crowdsourcing can be both economically and intellectually (providing longer-term unquantifiable benefits) fruitful activity, but firms may need to be realistic about what types of problems and users they can feasibly engage, and what capabilities they have or need to manage the community and its expectations. 8. Methods and reliability of crowdsourcing research There is a variety of research methods used in research on crowdsourcing: - Conceptual papers that develop definitions and categorizations for the crowdsourcing concept (e.g. Brabham, forthcoming) or that relate the notion of crowdsourcing to other literatures (e.g. Schenk and Guittard, 2009; Burger et al., 2010). - Conceptual papers based on mathematical economic models (e.g. Terwiesch and Xu, 2008). - Empirical studies based on econometric analyses of external firm-based datasets such as Innocentive.com and TopCoder.com (see e.g. Jeppesen and Lakhani, 2010; Boudreau et al., 2011) - Empirical research using comparative experiments and survey data derived from users and other constituents (Leimeister et al., 2009; Poetz and Schreier, 2011; Magnusson, 2009; Piller and Walcher, 2006). - Empirical studies based on one or multiple case studies, using interviews and other archival data sources (Alexy et al., 2010; Pisano and Verganti, 2008; Benbya and Van Alstyne, 2011; Bjelland Wood, 2008; Boudreau and Lakhani, 2009). Naturally, each of these methods used have value in their own right, and are typically chosen depending on the overall research question. However, as highlighted earlier, we have relatively little direct evidence of the value of crowdsourcing, in particular in technologyintensive sectors. To gain more insight into the actual impact of crowdsourcing, we think comparative field experiments are a particularly useful method, especially if complemented by interview data from relevant stakeholders. Further, a longitudinal perspective (e.g. tracing 23 (30) WHITE PAPER JUNE 2011 Aitamurto-Leiponen-Tee brought to you by www.ideasproject.com the entire trajectory from idea to product and its subsequent reception in the market place) would also deliver valuable insights into the quantitative and qualitative value of crowdsourcing, and could generate ways to improve utilization, and our understanding, of the potential of crowdsourcing as a supplementary tool of open innovation. Overall, the rapidly expanding literature demonstrates the great interest this topic currently receives. However, the literature on crowdsourcing is in its early stages and still suffers from vague definitions and lack of clarity and comparability regarding the relevant dimensions of the phenomenon, and hence regarding the relevant directions of research. 9. Open research questions As highlighted in previous sections, existing research has clearly highlighted the potential of crowdsourcing, emphasizing the benefits of openness and diversity as a source of value. There is also empirical evidence (as described in section 3) that utilizing the “crowd” might have strong benefits. At the same time, it is important to emphasize that many of these studies (with the exception of Magnusson, 2009) have not directly focused on mobile communications. For instance, the study by Poetz and Schreier (2011) examined baby products; Jeppesen and Lakhani (2010) looked at scientific problem solving contests, and Brabham et al (2009) highlighted an innovation challenge for designing bus stops, organized by a public organization Obviously, these settings differ from mobile communications in important ways, such as the amount of technical user knowledge required (in the case of baby products). Therefore, the extent to which these findings apply to other settings is still an open question. It seems that it would be important to more thoroughly investigate the potential benefits of crowdsourcing to generate ideas for different types of innovations. For example, new service ideation, services being less tangible than physical goods, might benefit from a different type of a process than product ideation. Research designs that allow generalization or at a minimum comparison across technology fields would thus be very valuable. Furthermore, we have found limited research regarding the significance of the organization‟s capability to absorb innovations from external contributors, apart from di Gangi‟s and Wasko‟s study on Dell IdeaStorm. This is an important factor, as the ideas offered by the crowd could have a lot of potential, but the company might not have the capability to utilize and execute them in the best possible manner due to the organization‟s internal product development system. However, if the open innovation approach, including crowdsourcing, is embedded systematically in a company‟s long-term strategy and becomes an integral part of the R&D process, the company can profit significantly, as Huston and Sakkab (2006) argue. It is also important to emphasize that the overall benefits of crowdsourcing might extend to areas that are not immediately obvious to the particular crowdsourcing project. Sometimes these benefits are difficult to quantify (e.g. what is the exact value of an “enthusiastic” user or developer community?), but may be very important when strengthening the brand, and creating more personalized product-and brand-related experiences for the users, and to access the tacit knowledge of users. 24 (30) WHITE PAPER JUNE 2011 Aitamurto-Leiponen-Tee brought to you by www.ideasproject.com As mentioned in section 9, we think there is great value and much promise in examining the benefits and limitations of crowdsourcing, e.g. via an experimental study comparing different user groups. This would allow researchers and managers to further investigate the impact of crowdsourcing in the mobile communications setting and to shed light on the factors that influence how to use crowdsourcing in a company such as Nokia. This type of a study can give more solid and reliable direct empirical evidence regarding the value of crowdsourcing practices in a high-tech sector such as mobile communications. 10. Conclusion This survey of the various literatures – innovation strategy, New Product Development management, Intellectual Property Rights management, information systems, and communication – has uncovered a number of aspects that should be considered when implementing crowdsourcing practices to enhance external innovation sourcing. The first key consideration appears to be the decision about the goals of crowdsourcing activity. Internal or external crowdsourcing can contribute to a number of different aspects of a business, but the design of the process should be driven by the goals. For example, if brand enhancement or market research are the primary goals, broad external engagement with a large number of non-expert users with attractive demographic characteristics would be desirable, whereas technical problem solving would necessitate capturing the attention of highly-educated and intrinsically-motivated experts in a diverse set of fields and locations, but the demographics and buying power of these individuals would be irrelevant. After the goal has been set, the information and problem-solving environment needs to be analyzed in order to determine what processes, participants, and system features generate the most appropriate environment for user inputs. The literature has thus far discussed a number of characteristics of the information environment (see table 8). Characteristics of the targeted ideas or solutions can be mapped in terms of the targeted audience (experts vs non-experts; fields of expertise), the appropriate governance structure for the process (hierarchical vs. flat; open vs. closed), and the degree of competitive vs. community processes (including incentive structures and numbers of participants). For example, uncertainty about the problem or the knowledge base required to solve it suggests an open governance approach with a large number of participants to mitigate the effects of uncertainty. On the other hand, when knowledge needed for the solution is cumulative, a community approach is better than anonymous competition. An open governance approach is also likely to be beneficial here. Finally, there may be a tradeoff between expert and non-expert contributors, as the former are likely to generate more feasible, but the latter more creative solutions. Balancing the needs for originality and feasibility is thus an important consideration. Alternatively, these needs may need to be targeted in different challenges or designs. 25 (30) WHITE PAPER JUNE 2011 Aitamurto-Leiponen-Tee Table 8. brought to you by www.ideasproject.com Characteristics of targeted ideas or solutions Dimension Implications for crowdsourcing design Potential structure Uncertainty about the problem/solution Increase number of participants when problem uncertainty is higher, e.g. for multi-domain problems (Boudreau et al. 2011) Use experts and/or non-experts (i.e. no exclusion ex-ante). Governance: open Cumulativeness of associated knowledge When an innovation is based on cumulative knowledge, use communitybased solutions (as opposed to marketbased), via e.g. integrator, product or twosided platforms (Boudreau and Lakhani, 2009) Use experts and/or non-experts. Governance: open Community process Complexity of problem/knowledge/ innovation Firm can reduce complexity through its absorptive capacity; whether ideas get absorbed further depends on user community pressure/arguments (Di Gangi and Wasko 2009) Originality/creativity/ variety Focus on ideation by end-users as opposed to professionals (Piller and Walcher, 2006; Magnusson, 2009; Poetz and Schreier, 2011) Use non-experts Feasibility Focus on ideation by professionals if feasibility is key priority, though end-users can also submit feasible ideas (Magnusson, 2009; Poetz and Schreier, 2011) Use experts (possibly complemented by nonexperts) Usefulness/user value Focus on ideation by end-users (Magnusson, 2009) Use non-experts Task specificity Task specificity is usually low in crowdsourcing initiatives, to stimulate heterogeneity of potential ideas (Piller and Walcher, 2006) Use experts and/or non-experts Governance: open To conclude, the academic and practitioner literatures on crowdsourcing have begun to generate useful conceptual material to support thinking about these practices and to facilitate their adoption and design in corporate contexts. However, at this early stage, there is no comprehensive understanding and coherence about the relative importance of the characteristics mentioned in the table above, their interactions with each other, or their causal 26 (30) WHITE PAPER JUNE 2011 Aitamurto-Leiponen-Tee brought to you by www.ideasproject.com implications for the design of a crowdsourcing system. In particular, only a handful of highquality empirical studies have been completed and results published, and each study, by necessity, can only address a limited set of research questions. Hence, there is ample room for more empirical research that utilizes state-of-the-art research methods and builds on and extends insights from innovation, decision sciences, and strategic management. For example, although motivation to participate is an ancient topic in information systems and management more generally, there is no clear consensus about why and when intrinsic or extrinsic motivations dominate and how they interact. The safest conclusion is to say that both matter, but it seems that extrinsic motivators rarely work in isolation – intrinsic ones are essential and can be enhanced by extrinsic ones. For example, Google found in its internal prediction markets that, although the $1000 cash prize was appreciated, people really liked the “cool” tshirts that enabled them to show they had been top-performing traders in the market (Coles, Lakhani, McAfee, 2007). Exploring the mediating role of participants‟ motivation in online real-time experiments would seem to be a way forward. We propose to study crowdsourcing of innovation inputs with Nokia in a systematic way, attempting to identify the aforementioned dimensions (table 8) and their implications in the company‟s ongoing crowdsourcing initiatives, and participate in designing future crowdsourcing activities such that enlightened and selective experiments can be set up to understand the true drivers of successful, high-impact, and effective crowdsourcing. Preliminarily, the following research questions have surfaced that would seem to be of broad interest in strategic innovation research: 1. How to design crowdsourcing processes that match the relevant characteristics of the firm, the problem, required expertise, and the competitive environment to calls? 2. How to define the calls in a way that enhances user group targeting, user appeal/motivation, and quality or appropriateness of solutions/inputs? 3. How to integrate these processes with internal innovation systems, and with other open innovation practices? We feel there is already much data accumulating within Nokia to shed light on these questions in a preliminary study, and going forward, we could work with the Nokia crowdsourcing team to enhance the design of new challenges in a variety of experimental contexts to really highlight the causal relationships between the conceptual dimensions. 27 (30) WHITE PAPER JUNE 2011 Aitamurto-Leiponen-Tee brought to you by www.ideasproject.com References Afuah, A. and C. Tucci (2011): Crowdsourcing as solution to distant search. Unpublished manuscript, EPFL and University of Michigan. Aitamurto, T. (2011) The impact of crowdfunding on journalism. Case study of Spot.Us, a platform for community-funded reporting. Journalism Practice. Routledge. Vol 5 (3) Aitamurto, T. and S. Lewis (2011) Public APIs and News Organizations: A Study of Open Innovation in Online Journalisms. Conference paper. Online Journalism Symposium, Austin, Texas. Alexy, O., A. Salter and P. Criscuolo (2010). No Soliciting: Strategies for Managing Unsolicited Innovative Ideas (August 16, 2010). Available at SSRN: http://ssrn.com/abstract=1659749 Benbya, H. and M. Van Alstyne (2011). How to Find Answers Within Your Company. MIT Sloan Management Review, Winter. Bjelland, O. M. and R.C. Wood (2008) An Inside View of IBM‟s Innovation Jam, MIT Sloan Management Review, Fall. Boudreau, K, N. Lacetera and K. Lakhani (2011). The Effects of Increasing Competition and Uncertainty on Incentives and Extreme-Value Outcomes in Innovation Contests. Management Science (forthcoming). Bonabeau, E. (2009). Decisions 2.0: the Power of Collective Intelligence. Sloan Management Review 50(2): 45-52. Boudreau, K.N. and K. Lakhani (2009). How to Manage Outside Innovation. MIT Sloan Management Review 50(4): 69-75. Brabham, D. C. (forthcoming). Crowdsourcing: A model for leveraging online communities. In A. Delwiche & J. Henderson (Eds.), The Routledge Handbook of Participatory Culture. Brabham, D. (2010). Moving the crowd at Threadless. Information, Communication, and Society 13(8): 1122-1145. Brabham, Sanchez and Bartholomew (2009). Crowdsourcing public participation in transit planning: Preliminary results from the next stop design case. Proceedings paper Transportation Research Board Annual Meeting 2010. Burger-Helmchen, T. and J. Penin (2010). The limits of crowdsourcing inventive activities: What do transaction cost theory and the evolutionary theories of the firm teach us? Conference paper, Workshop on Open Source Innovation, Strasbourg, France. Burmann, C. (2010). A call for User-Generated Branding. Journal of Brand Management 18, (1) 1–4. Chesbrough, H. W. (2003). Open Innovation: The New Imperative for Creating and Profiting from Technology. Boston, MA: Harvard Business School Press. 28 (30) WHITE PAPER JUNE 2011 Aitamurto-Leiponen-Tee brought to you by www.ideasproject.com Chesbrough, H. W. (2006a). Open Business Models. How to Thrive in the Changing Innovation Landscape. Cambridge, MA: Harvard Business School Publishing. Chesbrough, H. W. (2006b). Open Innovation: A New Paradigm for Understanding Industrial Innovation. In H. W. Chesbrough, W. Vanhaverbeke, & J. West (Eds.), Open Innovation: Researching a New Paradigm (pp. 1-27) Chesbrough, H. W. (2011). Open services innovation: Rethinking your business to grow and compete in a new era. Jossey-Bass, San Francisco. Coles, P.A., K.R. Lakhani and A.P. McAfee (2007). Prediction Markets at Google. HBS case study 9-607-088. di Gangi, P. and M. Wasko (2009). Steal my idea! Organizational adoption of user innovations from a user innovation community: A case study of Dell IdeaStorm. Decision Support Systems 48, 303–312. Forrester Consulting (2009): The Total Economic Impact ™ of InnoCentive Challenges. Single Company Case Study. Project Director: Michelle Bishop. Hatch and Schultz (2010). Toward a theory of brand co-creation with implications for brand governance. Journal of Brand Management 17 (8), 590–604. Howe, J. (2006). The Rise of Crowdsourcing. Wired 14(6). Howe, J. (2009). Crowdsourcing: Why the Power of the Crowd is Driving the Future of Business. Revised edition, Crown Publishing, New York. Huston, L. and N. Sakkab (2006): Connect and Develop. Inside Procter & Gamble's new model for innovation. Harvard Business Review. Jeppesen, L. B. and K. R. Lakhani (2010). Marginality and Problem-Solving Effectiveness in Broadcast Search. Organization Science, 21: 1016-1033. Lakhani, K.R. (2008): InnoCentive.com. HBS case study 9-608-170. Revised October 28, 2009. Lakhani, K. R., and Z. Kanji (2008). Threadless: The Business of Community. Harvard Business School Multimedia/Video Case 608-707. Leimeister, J. M., M. Huber, U Bretschneider and H. Krcmar (2009). Leveraging Crowdsourcing - Activation-Supporting Components for IT-based Idea Competitions. Journal of Management Information Systems, 26, no. 1. Levy, P. (1997) Collective Intelligence: Mankind's Emerging World in Cyberspace. Perseus Books, Cambridge, MA Lichtenthaler, U. (2010). Proceedings of the R&D Management Conference, Manchester, Great Britain. Magnusson, P.R. (2009). Exploring the contributions of involving ordinary users in ideation of technology-based services. Journal of Product Innovation Management, 26(5), 578-593. 29 (30) WHITE PAPER JUNE 2011 Aitamurto-Leiponen-Tee brought to you by www.ideasproject.com Malone, T.W., R. Laubacher, and C. Dellarocas (2010): Harnessing Crowds: Mapping the Genome of Collective Intelligence. MIT Sloan School Working Paper 4732-09. O‟Neil, M. (2010) Shirky and Sanger, or the cost of crowdsourcing. Journal of Science Communication 09(01). Piller, F.T. and D. Walcher (2006). Toolkits for idea competitions: A novel method to integrate users in new product development. R&D Management, 36(3), 307-318. Pisano, G. (2006). Profiting from innovation and the intellectual property revolution. Research Policy 35 1122–1130. Elsevier. Pisano, G.P. and R. Verganti (2008). Which kind of collaboration is right for you? Harvard Business Review, December. Poetz, M.K. and M. Schreier (2011). The value of crowdsourcing: Can users really compete with professionals in generating new product ideas? Journal of Product Innovation Management (forthcoming). Prahalad, CK and V. Ramaswamy (2000). Co-Opting Customer Competence. Harvard Business Review January 2000. Prahalad, CK and V. Ramaswamy (2004). Co-Creation Experiences: The next practice in value creation. Journal of Interactive Marketing. 18 (3) 5-14. Wiley Periodicals. Schenk E. and C. Guittard (2009). Crowdsourcing: What can be Outsourced to the Crowd, and Why? Workshop on Open Source Innovation, Strasbourg, France. Surowiecki, J. (2004). The Wisdom of Crowds: Why the Many Are Smarter Than the Few and How Collective Wisdom Shapes Business, Economies, Societies and Nations. Doubleday, New York. Terwiesch, C. And Y. Xu (2008). Innovation contests, open innovation, and multi-agent problem solving. Management Science, 54(9), 1529-1543. von Hippel, E. (2005). Democratizing innovation. Cambridge, MA: MIT Press. von Hippel, E. (1976). The Dominant Role of Users in the Scientific Instrument Innovation Process. Research Policy 5, no. 3 (July):212-39. von Hippel, E. (1982). Appropriability of Innovation Benefit as a Predictor of the Source of Innovation. Research Policy 11, no. 2 (April): 95-115. von Hippel, E. (1988). The Sources of Innovation. New York: Oxford University Press. Von Hippel, E. and R. Katz (2002). Shifting Innovation to Users Via Toolkits, Management Science, Vol 48, No. 7 (July) pp 821-833. 30 (30)