Why Do Category Labels Stick? Unpacking the Innovation Paradox

Why do Category Labels Stick? Unpacking the Innovation Paradox

Stine Grodal

Fernando Suarez

Diego Zunino

Boston University Questrom School of Business

ABSTRACT

Adding to the traditional studies of technology adoption by scholars in economics and management, scholars with socio-cognitive and institutional perspectives have recently proposed an apparent paradox in the adoption of new technologies: The need to balance the perceived familiarity and novelty of a new technology. This idea has become well accepted, but scant empirical research has been doneto confirm or deepen our theorizing. By examining an important element of how innovations are understood, their related category labels, we show how adoption is affected by the degree to which a new product label is perceived as familiar or creative.Drawing on a large data set from the smartphone industry from 1998 to 2011, we track firms’ adoption of category labels over time. We find support for an inverted U-shape relationship between the firms’decision of which label to adopt and the degree to which labels are perceived as familiaror creative. We confirm these results with an online experimentdesigned to address potential endogeneity concerns. Our paper expandsour understanding of the innovation paradox and further builds theory around why some category labels stick while others are left in the dustbin of history.

INTRODUCTION

The question of how new technologies get adopted has been a key concern of scholars in technology management and economics for decades (Hannan and McDowell, 1984; Rogers, 2010; Majumdar and Venkataraman, 1998). A more recent literature has begun to explore the role that socio-cognitive and institutional factors play in the stakeholders’ framing and understanding of new technologies, and their consequent influence on how each technology fares in the market(Kennedy and Fiss, 2013). This literature has highlighted an apparent paradox that every new technology needs to overcome. On the one hand, new technologies have to differentiate themselves from existing technological offerings; in other words, theymust convey ideas of novelty, creativity, and originality. However, at the same time, new technologies have to appear familiar enough to stakeholders and invoke existing understandings, so as to minimize the natural reluctance to something new and unproven (Hargadon and Douglas, 2001; Bingham and Kahl, 2013).

A growing body of research has examined the socio-cognitive aspects surrounding new technologies and markets (Rosa et al., 1999; Pontikes, 2012; Suarez, Grodal and Gotsopoulos, 2014). This research has stressed the role of categories and their associated labels in shaping stakeholders’ understanding and perception of new technologies and products (e.g. Navis and Glynn, 2010). Categories are defined as socially constructed partitions or taxonomies that divide the social space into groupings of objects perceived to be similar (Bowker and Star, 2000). Category labels --words or phonemes that are used to reference objects that are perceived to be distinct or belong together-- are the first instantiation of categories. As such, they are one of the earliest factors that shape stakeholders’ understandings about a new product.

When introducing an innovation to the market, producers often lack terms for how to reference their new product (Kaplan and Tripsas, 2008). In order to communicate the meaning of their product to stakeholders they often invent category labels, like such as “pocket PC”, “camera phone” or “smartphone”. When confronted with a newly created category label, stakeholders have no prior experience with the kind of products that such label may refer to; consequently, at the beginning,the meaning of most new category labels tends to be shallow. Extant research has suggested that stakeholders begin to make sense of new labels by linking the label’s constituent parts via a relation to existing objects, or by projecting properties of a constituent (Wisniewski, 1996; Smith, Osherson, Rips and Keane, 1988). Through an iterative and negotiated process by which labels are adopted and used by different stakeholders, labels become infused with value and began to be associated with pre-existing categories, and thus gradually become meaningful categories themselves (Peirce, 1931; Grodal et al., 2014). As the meaning of the new categories develop, they begin to form “rules” or “boundaries” that dictate which objects can claim membership to a given category (Hannan, Polos, and Carroll, 2007; Navis and Glynn, 2010). Category labels that do not get traction fail to get infused with meaning and are ultimately abandoned. As this process continues, emerging industries often coalesce in around a dominant category, or “the conceptual schema that most stakeholders adhere to when referring to products that address similar needs and compete for the same market space” (Suarez et al., 2013).

Despite the importance that category labels have in shaping stakeholders’ understanding about new technologies, we know relatively little about the process by which some of these labels are selected over others and how they “emerge and fall out of use” (Kennedy and Fiss, 2013, p. 1). This is an important gap in our theory, since the performance of entrants into a new industry can depend on the categorical labels they choose when positioning their products (Rosa et al., 1999; Pontikes, 2012). In this paper, we tackle this gap by presenting an empirical study of the creation and selection of categorical labels in what became to be known as the “smartphone” industry. By collecting data from the official product launches of each smartphone in the period 1998 to 2011 in the U.S., we trace the categorical labels used by smartphone producers over time. Borrowing from linguistics and extant research on categories, we test and expand existing theory about why some labels tend to stick while others are abandoned.

In particular, we revisit and test the idea that, to secure adoption, producers have to find ways to convey both notions of change and stability when introducing new products (Hargadon and Douglas, 2001). We posit that one of the key ways in which producers convey these notions to consumers and other stakeholders is through their choice of category labels. The literature, however, has not been consistent or specific enough as to what kind of change andstability is needed to maximize adoption. Some authors seem to imply some type of trade-off between these dimensions (e.g. Bingham and Kahl, 2013), while other authors seem to suggest that both can be achieved at once (Hargadon and Douglas, 2001). Studies of adoption in other contexts that also use this change-stability distinction are equally unspecific. For instance, Uzzi et al. (2012), in one of the few large scale empirical studies besides ours, use the opposite sides of a single measure to operationalize both constructs (in their terms, novelty and conventionality) in their study of how atypical combination of knowledge affect the adoption (cites) of scientific papers, which would suggest some kind of trade-off. However, then they argue that most-cited papers score high in both novelty and conventionality. Mckinley et al. (1999), studying adoption of schools of thought in organizational theory,recognize the existence of a tension between continuity and novelty, but then they go on to argue that adoption requires an “adequate” level of each (p.637).

In this paper, we provide a fresh empirical analysis from which our existing theory and understanding can be improved. By focusing on why some category labels are adopted more than others, we provide both granularity and specificity to the discussion of these important issues. In particular, we hypothesize a non-linear relationship between label adoption and a construct denoting stability and conventionality, which hereon we refer to as label familiarity; we also hypothesize a non-linear relationship between label adoptionand a construct denoting change and novelty, which hereon we will refer to as label creativity. Our results largely support our hypotheses, thus representingnot only one of the first large empirical studies of category label adoption, but arguably the most detailed data-based work to date on the change-stability paradox. Moreover, we complement our econometric analysis by carrying out and reporting the result of an online experiment specially designed to provide further confirmation of our findings, and to address possible concerns about endogeneity in our regressions. The results of the experiment strongly confirm our empirical findings, making our paper also, to our knowledge, the first multiple-methods work on this topic.

THEORY AND HYPOTHESES

Categories are arguably the most widely-researched construct among several that have been proposed to capture the socio-cognitive dimension of technology emergence and adoption, such as “field frames” (Lounsbury 2001; Lounsbury, Ventresca and Hirsch, 2003); “schemas” (Bingham and Kahl, 2013), and “technological frames” (Kaplan and Tripsas, 2008; Orlikowski and Gash, 1994). Categories help group similar objects and can indeed determine the set of characteristics that the objects belonging to a given category are expected to posses, and the elements that differentiates them (or not) from members of other categories (Vygotsky, 1986).

As the first instantiation of categories, category labels help make sense of new productsby allowing stakeholders to develop semantic links to other categories and their associated labels, relating a label to other objects or concepts and borrowing from the inherited properties of the label components (Wisniewski, 1996). It is through this process that the “deepening” of meaning takes place (Grodal et al, 2014)through whichsome of these labels eventually become established categories (Peirce, 1931; Bingham and Kahl, 2013). When stakeholders observe a category label, they construct the group of objects that they perceive as being associated with it (Yamauchi and Markman, 1998). While producers are the main creators of labels in the process of introducing their products to an emergent industry, but labels are socially constructed and thus other stakeholders such as users, industry analysts or observers, bloggers, etc, can also create them. For instance, the “mountain bike” label was created by users who tweaked their standard bikes for racing down hills, the “robot” label was created by a writer of a science fiction book, and the label “impressionism” was created by art critics (originally with a negative connotation) to refer to the unconventional work of Monet and other painters of the time.

Extant research has sought to understand the extent to which firms spanning categorical boundaries face performance penalties (Zuckerman, 1999; Hsu et al. 2009), and the conditions under which spanning categorical boundaries might be permissible (Ruef and Patterson, 2009; Fleischer, 2009; Pontikes, 2012; Granqvist et al. 2013). Despite notable progress in our understanding of categories, scant research, theoretical or empirical, has been conducted to unveil the dynamics of competing categories and their associated labels; i.e. the contestation process that leads to the adoption of some labels and the abandonment of others. In order to tackle this challenge, in this study we test and extend the theoretical argument that, in order to drive higher adoption, category labels have to resolve the paradox of being simultaneously (1) creative enough to convey notions of novelty andchangeso as to attract the attention of stakeholders, and (2) familiar enough that they convey notions of continuity and stability, so as to be readily comprehensible. Similar versions of this broad argument have been proposed by a handful of authors. For instance, Hargadon and Douglas (2001) suggest that, “Without invoking existing understandings, innovations may never be understood and adopted in the first place. Yet, by hewing closely to existing institutions, innovators risk losing the valued details, representing the innovation’s true novelty, that ultimately change those institutions. Success, then, requires entrepreneurs to locate their ideas within the set of understandings and patterns of action …in order to gain initial acceptance, yet somehow retain the inherent differences” (p. 478).

While the general argument of a paradox is appealing, it is by no means widely accepted, particularly when it comes to the form and limits of the tension between the two constructs. Moreover, the argument has, to our knowledge,never been empirically tested. The existing literature has been ratherunspecific and even in apparent disagreement with respect to the existence of a familiarity-creativitytension, and the role that it may take during technology adoption. The quote from Hargadon and Douglas above would suggest that the authors believe that familiarity and creativity can be achieved at the same time, and that having more of each is always beneficial for adoption. However, no conclusion can be drawn from their theorizing as to how much of each dimension maximizes adoption; for instance, should a technology sponsoralwaystry to convey more familiarity and more creativityin their label to secure stronger adoption? Bingham and Kahl (2013), in contrast, perceive a trade-off.They claim that their study resolves, “a conundrum related to the process of emergence –how to manage the simultaneous existence of two inconsistent states, familiarity and novelty” (p. 15). They propose that these “inconsistent states” can be overcome by focusing on one first (familiarity) and then the other (novelty). Studies of adoption that have used the familiarity-creativity dilemma in contexts other than technology are equally unspecific. McKinley et al. (1999) state that, “Our central thesis concerning novelty and continuity is that although there exist tension between them, adequate level of each are necessary [for adoption].” Not surprisingly, they give no clues to determine what an “adequate” level is.

Wecontribute to existing literature by providing a fresh perspective to study these open issues in the literature, based on a careful empirical study complemented byan online experimentthat allows us to add precision to our theorizing. We focus on the structure and content of category labels in order to develop a more nuanced account of the familiarity and creativity constructs.In our view, these constructs are best understood through the prism of recombination. An established literature has shown that new creations are formed through the recombination of existing elements (Schumpeter, 1939; Wisniewski, 1996), and that the level of creativity in these recombinations can influence both the success of the innovation (Fleming, 2001; Fleming, Mingo, and Chen, 2007) and how widely they diffuse (Grodal and Thoma, 2014).

A category label can be creative while still using very familiar words of phonemes. For example, when John Burton Carpenter in the 1970s created the compound label “snowboard” it was creative, because it combined two words “snow” and “board” that had seldom been used together, but at the same time it was familiar, because even stakeholders who were exposed to the compound for the first time were able to associate it with elements they knew and that helped them make sense of the new product. Indeed, compounds, defined as “the simple concatenation of any two or more nouns [or other words] functioning as a third nominal” (Downing, 1977: 810), are an important way for stakeholders to create category labels that: (a) build links to existing categories and meanings, thus invoking continuity and familiarity but, at the same time, (b) allow for novelty and change by supporting creativerecombinations (Lieber, 1983; Wry, Lounsbury and Jennings, forthcoming).

A central tenet of our theorizing in this paper is that, when it comes to adoption, there are first increasing and later decreasing returns to both creativity and familiarity. In other words, the more familiar a category level, the higher the adoption, but only up to a point. The more creative a category label is, the higher the adoption, but also up to a point. Both of econometric analysis and online experiment below are geared to test these hypotheses.

Labels become familiar by establishing links to existing categories and their associated labels. Category labels that fail to make these connections will most likely be abandoned because stakeholders will be confused about the labels’ meaning. As we noted before, the label “snowboard” became dominant in that industry and was widely adopted, as it represented a creative recombination of words that conveyed both change and familiarity. However, snowboard was not the label used by the first such product introduced to the market. The first product was introduced using a different label, the “snurfer,” which failed to make enough connections to existing labels and categories and therefore was later was abandoned in favor of snowboard. While the snurferlabel drew from the words “snow” and “surfer,” these associations were disguised, making it hard for stakeholders encountering the label for the first time to make sense of it. While being unfamiliar might be problematic, being too familiar has its disadvantages as well, because when category labels are perceived to be too familiar they become taken-for-granted and thereby fail to elicit interest and scrutiny by stakeholders (Hsu and Grodal, 2014). Too much familiarity may render the label obvious or uninteresting and thus fail to capture the attention of stakeholders. For instance, one of the early labels in the smartphone industry was “camera phone,” directly derived from the fact that one of the first technological features that the new type of phones added was a digital camera. The label was an obvious combination of two very well known labels that had existed for a long time, and as such did not capture major attention from stakeholders.