The research findings presented in this monograph suggest several relationships across dimensions of experience that would benefit from greater articulation: the relationship between direct and indirect experience, the relationship between het- erogeneous and homogeneous experience and the relationship between experience before doing and experience during doing.
1. Direct and Indirect Experience
The relationship between direct and indirect experience is an issue that would benefit from additional research. Several studies have shown that direct and indirect or vicar- ious experience substitute for each other. Levine and Prietula (2012), for example, argued that investments in learning from direct experience weakened the benefits of learning from indirect experience or knowledge transfer. Similarly, Haas and Hansen (2005) found that experienced teams were hurt more by indirect experience embed- ded in a knowledge management system than teams with less direct experience. Other studies have shown that direct and indirect experience complement or rein- force each other positively (Bresman, 2010). Consistent with research on absorptive capacity (Cohen & Levinthal, 1990), direct experience enabled the teams in the Bresman (2010) study to make better use of indirect experience. Studying individual learning, Walsh and Anderson (2011) found that indirect experience or instruction affected learning as manifest in behavioral change but that direct experience was required to affect neural learning as measured by changes in neural indicators.
These findings have implications for theory as well as suggest directions for future research. What are the conditions under which direct and indirect experience complement each other? What are the conditions under which learning from direct and indirect experience substitute for each other? Are there certain kinds of knowl- edge that can only be learned from direct experience?
2. Heterogeneous and Homogeneous Experience
The second relationship is between heterogeneity and homogeneity. An organization can foster heterogeneity by employing individuals with different backgrounds, pro- ducing a wide range of products or services, using different tools, experimenting with different structures, encouraging different strategies, and so on. Alternatively, an organization can foster homogeneity by hiring similar members, socializing them intensely, producing similar products or delivering similar services, using similar procedures and tools, and discouraging experimentation. Some heterogeneity is needed to generate new combinations and create new knowledge while some homo- geneity is needed to develop common understandings and transfer knowledge throughout the enterprise.
This issue is related to the previously described relationship between group and organizational learning. An organization that fosters learning at the level of the group is more likely to have a diverse experience base than one that fosters learning at the level of the organization. Thus, factors that lead an organization to promote learning at one level over another also lead an organization to favor heterogeneity over homogeneity. For example, organizations performing creative tasks (Jackson, May, & Whitney, 1995) or facing turbulent environments (Moorman & Miner, 1997) are more likely to benefit from heterogeneity than their counterparts who perform less creative tasks in less turbulent environments.
The relationship between heterogeneity and homogeneity highlights the impor- tance of different factors in the creation of knowledge than in the transfer of knowl- edge. Heterogeneity fosters the development of new knowledge. For example, groups composed of diverse members have been found to be better at generating new or emergent knowledge and developing more sophisticated solutions than groups com- posed of similar members (see Chap. 5). By contrast, homogeneity promotes knowl- edge transfer (see the Nummi example, as described by Adler & Cole, 1993). Homogeneity increases the relevance of knowledge acquired in one unit for another.
Organizations can handle the relationship between heterogeneity and homogene- ity to some extent by encouraging units engaged primarily in knowledge creation, such as Research and Development departments, to have a more diverse experience base (e.g., by promoting diversity of members’ backgrounds and encouraging experimentation) than units such as production where knowledge creation is not the primary goal. The approach should be used with appreciation that too much diver- sity can be harmful even for knowledge-creation tasks, such as research, if the diver- sity results in a lack of shared values or goals (Owens & Neale, 1998). Further, some degree of diversity can be beneficial—even for groups such as production not explicitly charged with knowledge creation—because innovation and learning occur in these groups as well (Brown & Duguid, 1991). Indeed, in our studies of produc- tion facilities, we saw many instances of new knowledge being created (see also von Hippel & Tyre, 1995).
Another approach to dealing with the relationship between heterogeneity and homogeneity is to foster heterogeneity during certain phases of the product’s life cycle and homogeneity during others. Diversity is most likely to be beneficial in the development, design, and initial launching of a product while homogeneity is most beneficial during its production. This approach is consistent with the finding that diversity fosters creativity (Jackson et al., 1995) as well as with the finding that too much diversity (as measured by engineering change orders) can harm productivity (Hayes & Clark, 1986). The approach has merit but should be used with apprecia- tion that some diversity might still be beneficial even after an organization goes into the production phase and some homogeneity is needed even in the research and development stage. The approach should also be used with acknowledgement that there is feedback and recycling across the phases of a product’s life cycle and phases can occur concurrently.
The tension between heterogeneity and homogeneity is related to the tension between exploiting old competencies and exploring new possibilities eloquently described by March (1991). According to March, maintaining the right balance between exploitation and exploration is key to organizational survival and pros- perity. Too much emphasis on exploitation can lead an organization to fall into a “competency trap” (Lant & Mezias, 1990; Levitt & March, 1988) whereby it persists in a strategy it perfected that may no longer be optimal. Too much emphasis on exploration can lead to a lack of depth or distinctive competence for a firm.
The right balance between exploration and exploitation is difficult to achieve or maintain. Because the rewards to exploitation are more immediate and cer- tain, exploitation often drives out exploration in organizations (March, 1991). Ford’s production of the Model T Ford is a classic example of the costs incurred by a firm when exploitation drove out exploration (Abernathy & Wayne, 1974). Ford became so proficient at producing the Model T that it neglected exploring other options that turned out to be more attractive to customers (see Leonard- Barton, 1992, for further discussion of “rigidities” that impair organizational performance).
Although the balance between exploration and exploitation is difficult to achieve, some organizations are adept at both. Tushman and Anderson (1986) examined whether competence-enhancing or competence-destroying innovations were most likely to come from incumbent firms or new entrants to an industry. Competence- enhancing innovations build on the existing knowledge base and are therefore related to exploitation. By contrast, competence-destroying innovations depart from the existing knowledge base and involve experimentation and exploration. Results indicated that competence-enhancing innovations generally came from incumbent firms in an industry. Although competence-destroying innovations came primarily from new entrants in the cement industry, they came about equally from both incum- bent firms and new entrants in the computer industry (Tushman & Anderson, 1986). Thus, certain incumbent firms in the computer industry were able to make both competence-enhancing and competence-destroying innovations (Dahlin & Casciaro, 1998). How these organizations exploit existing competencies while exploring new ones is addressed in research on organizational ambidexterity (Raisch, Birkinshaw, Probst, & Tushman, 2009).
3. Timing of Experience
The third dimension of experience is experience timing. What is the relationship between learning before doing and learning by doing? When will planning in advance result in desirable outcomes for the organization (Carillo & Gaimon, 2000)? When is learning by doing the best way to proceed?
A few studies have explicitly addressed whether learning by planning or learning by doing is the preferred mode of knowledge acquisition for firms. Pisano (1994) contrasted how an emphasis on laboratory experimentation (learning before doing or planning) affected process development lead times in traditional chemical-based pharmaceuticals and newer biotechnology-based pharmaceuticals. Results indicated that the extent to which the underlying scientific knowledge base was understood well determined the effectiveness of learning before doing (planning in our termi- nology). For chemical-based pharmaceuticals, where the underlying knowledge base was understood well, an emphasis on laboratory experimentation and learning before doing was associated with more rapid development. In contrast, for biotech- nology-based pharmaceuticals where the underlying knowledge base was not under- stood well, hours invested in laboratory research did not shorten project lead times. Pisano concluded that the nature of a firm’s knowledge base influenced the effec- tiveness of learning strategies. Deep knowledge of the effect of specific variables and their interactions or higher “stages” of knowledge (Bohn, 1994) increased the effectiveness of planning and learning before doing. By contrast, learning by doing is indicated when organizations do not have the underlying knowledge base needed to simulate effects in advance. Under these conditions, it is only through learning by doing that an organization acquires an understanding of how variables interact to affect performance. Von Hippel and Tyre (1995) made a similar argument in their analysis of problem identification on the factory floor.
Similarly, Eisenhardt and Tabrizi (1995) concluded that learning by doing was more effective than planning for launching uncertain new products in the computer industry. Planning slowed the pace of product development while learning by doing (measured by design iterations and testing) accelerated the pace. Eisenhardt and Tabrizi (1995) argued that in uncertain settings improvisational approaches that combine real-time learning and testing are more effective than planning.
The extent to which knowledge is dependent on the context also affects the rela- tive effectiveness of planning and learning by doing. An example of knowledge that is not very dependent on the context is how to calculate net present value or other financial measures. The methods for calculations do not vary as a function of fea- tures of a particular organizational context. When knowledge is not dependent on the specific people, tools, cultures, or structures in place at a firm, planning in advance is very effective.
By contrast, when knowledge is dependent on the context, learning by doing is key. When knowledge is dependent on people, structures, and technologies in place at a firm, it is difficult to predict in advance the effects of these variables and their interactions. Under these conditions, the best way to learn is by doing. An example of such context-dependent knowledge is the best allocation of individuals to tasks. The best allocation is difficult to predict with confidence without experience with the particular people, tools and tasks at an organization. Through actually perform- ing a task, individuals learn about their own skills, the skills of their colleagues, the capabilities of the equipment, and so on. Thus, when knowledge is context- dependent, learning by doing is critical to successful performance.
In short, several factors have been found to affect the relative effectiveness of learning by planning and learning by doing. Planning is effective when the knowl- edge base is understood well, when the knowledge is not dependent on the context, and when the task is certain. Conversely, learning by doing is preferred when the knowledge is uncertain, not understood well, and highly dependent on the organiza- tional context.
Source: Argote Linda (2013), Organizational Learning: Creating, Retaining and Transferring Knowledge, Springer; 2nd ed. 2013 edition.