r Academy of Management Discoveries 2015, Vol. 1, No. 1, 31–60. Online only http://dx.doi.org/10.5465/amd.2013.0016 WHY THE INTERNET MAKES BUYING A CAR LESS LOATHSOME: HOW TECHNOLOGIES CHANGE ROLE RELATIONS

STEPHEN R. BARLEY Stanford University

Drawing on ethnographic data collected over a two-year period in two car dealerships, this paper employs role theory and a dramaturgical analysis of sales encounters to show how the internet has changed the relationship between car salesmen and their customers. The paper explores why Goffman’s dramaturgical approach to analyzing encounters provides a way of analyzing technologically occasioned changes in the in- teraction order of a work system that allows students to grapple more holistically yet systematically with the social and material aspects of such change.

In a world enthralled with tweeting, texting, so- for merchants in cities (Fischer, 1992). The internal cial networking and shopping online, to proclaim combustion engine’s implications for livery drivers that the internet and technologies built on it are (who became taxi drivers) were substantially differ- changing the way we live is hardly news. Only a ent than they were for blacksmiths (who went out catatonic could miss the almost daily pronounce- of business). Accordingly, there is no reason to ex- ments by high-tech evangelists that some new in- pect the internet to alter the work of academics or ternet tool is ushering in another brave new world. doctors in ways even remotely akin to how it shapes Academic literatures in medicine (Demaerschalk what purchasing agents or car salesmen do, how- et al., 2012; Ross, Sepper, & Pohjonen, 2010), business ever signifi cant those changes might be for each (Turban, Lee, King, Peng, & Turban, 2009; VanHoose, line of work. 2011), engineering (Kühnle, 2010; Smite, Moe, & Faced with such heterogeneity, social scientists Agerfalk, 2010), the sciences (National Research have three options. The fi rst is to move up levels of Council, 1999) and the social sciences (Castells, analysis until they can elide differences in context. 1996; Wellman & Haythornthwaite, 2002) testify to This is precisely what most research on the internet the same story, albeit less passionately. But even does. Rather than ask how the internet has affected though the internet is rapidly becoming as infra- the daily work of car salesmen, travel agents, or aca- structural as electricity, there is surprisingly little demics, scholars ask how the internet has infl uenced research on how it is altering work systems or the patterns of automobile sales (Ghose, Mukhopadhyay, work that people do. Studies of email’s use (Barley, & Rajan, 2007; Kuruzovich, Viswanathan, Agarwal, Meyerson, & Grodal, 2011; Dabbish & Kraut, 2006; Gosain, & Weitzman, 2008; Viswanathan, Kuruzovich, Sproull & Kiesler, 1991) and distributed teams Gosain, & Agarwal, 2007; Zettelmeyer, Morton, (Hinds & Bailey, 2003; Hinds & Kiesler, 2002; & Silva-Risso, 2006), the structure of the travel O’Leary & Mortensen, 2010) are exceptions, and industry (Buhalis & Zoge, 2007; Lang, 2000), or even this research pays more attention the experi- the breadth of a researcher’s scholarship (Evans, ences and performance of individuals and groups 2009). In fact, much research on the internet even than to work systems or the way work is done. seeks to span industries by speaking broadly Perhaps it is unsurprising that we know so little about e-commerce, long tails, and mediated com- about how the internet affects work aside from the munication. Second, researchers can opt for pro- diffi culties people have coping with fl ooded in- ducing a corpus of situated studies of how the boxes and collaborating with people they never internet has altered work in this or that workplace. meet. Email overload and distributed teams are With enough studies, we might fi nd general pat- among the internet’s best known consequences be- terns. But given the current emphasis on making cause so many people experience them regardless theoretical if not generalizable contributions in every of employer, industry, or occupation. More impor- paper, aside from the efforts of an occasional eth- tantly, changes occasioned by infrastructural tech- nographer, any widespread move toward empirical nologies like the internet accrue over the long haul, nominalism of a concerted sort is unlikely. Finally, are maddeningly diverse, and are often occupation- researchers can seek to develop perspectives that ally specifi c. The telephone, for example, had very focus less on the effects of a specifi c technology, different ramifi cations for rural farmers than it did than on the processes by which technologies

31

Copyright of the Academy of Management, all rights reserved. Contents may not be copied, emailed, posted to a listserv, or otherwise transmitted without the copyright holder’s express written permission. Users may print, download, or email articles for individual use only. 32 Academy of Management Discoveries March

(including those of the internet) bring about change structures faded into the background. If the contin- in a line of work or the organization of a production gency theorists had cast their eye at too high a level system. The agenda would be less to reveal what of analysis by focusing on the alignment of types of the internet does to any particular occupation than environments with production systems, the con- to help researchers decide how, when, and where structionists ratcheted their resolution down to to look for changes that might occur. where it was diffi cult to see larger patterns of or- This paper plays the third approach off the sec- ganizing. Ideally, one would like an analytic ap- ond. Drawing on role theory and Erving Goffman’s proach that preserves the constructionists’ concern dramaturgical analysis of encounters, it proposes with the concrete, while linking situated action to an approach to studying how technologies change meso-level, if not more molar, changes in organi- interaction orders or systems of role relations and zations and occupations. role playing and, by extension, work systems. The Put differently, analysts need a way of leverag- narrative fi rst sketches a framework for studying ing action and structure simultaneously. Giddens technologically occasioned change in role relations theory of structuration has offered one such lever in a line of work or a workplace. It then illustrates (Barley, 1986; Orlikowski, 1992; Orlikowski & Robey, the framework’s utility with ethnographic data on 1991; Poole & DeSanctis, 1990). Role theory offers how the internet has been altering the work of car another. Roles (or more precisely, the patterned salesmen and the tenor of their encounters with cus- ways in which people play them) allow us to pivot tomers. The paper ends by exploring the implications from an ongoing stream of action to a more ordered of a dramaturgical approach to studying technolog- world of social structure. As the Chicago school so- ically occasioned change in a work system. ciologists understood, role is a Janus faced concept: it simultaneously looks in one direction toward the individual and in the other toward the social order A ROLE-BASED PERSPECTIVE ON HOW (Becker & Strauss, 1956; Hughes, 1937). Unlike struc- TECHNOLOGIES CHANGE WORK AND turation theory, which at best provides a broad on- ORGANIZATIONS tological perspective, role theory, especially as As organizations began to adopt computers and developed by interactionists like Goffman (1959, other computational technologies, organizational 1961b) and Turner (1968, 2006), tells us what to scholars gradually became dissatisfi ed with existing look for empirically. In other words, role theory approaches to studying technology. Contingency provides analysts with a set of concepts with which theory (Gerwin, 1979; Harvey, 1968; Perrow, 1967; one can structure observations. Woodward, 1958) and even socio-technical sys- tems theory (Emery & Marek, 1962; Rice, 1963; Trist ROLES, TECHNOLOGIES AND INTERACTION & Bamforth, 1951), once it fi xated on autonomous ORDERS teams, seemed unable to account for variations in organizing that were sometimes occasioned by iden- Barley (1990) was the fi rst to propose a role-based tical machines. Accordingly, by the 1990’s students theory of how technologies change work systems of technology began to jettison deterministic theo- and organizational structures. To devise a strategy ries of technological change in favor of a philosophy for documenting when and how new technologies of social construction and to study how specifi c change organizations, Barley linked Nadel’s (1957) technologies were construed and used in situ to un- theory of relational and non-relational roles to net- derstand how they might trigger diverse outcomes work theory’s dictum that every organization’s (Barley, 1986; Fulk, Steinfi eld, & Schmitz, 1987; structure is a network inscribed by the relations Orlikowski, 1992, 1996; Poole & DeSanctis, 1990). among its members. Specifi cally, he argued that The shift had two major consequences: In addition technologies trigger change by altering workers’ to encouraging an ontological shift, construc- non-relational roles: the tasks they perform and tionists advocated studying concrete, situated prac- how they perform them. Such changes may then tices that emerged as people used specifi c artifacts lead to changes in the nature of the interactions and software (Boczkowski & Orlikowski, 2004; workers have with members of their role set (the Leonardi & Barley, 2010; Orlikowski & Scott, 2008). people with whom they interact while doing their Emphasizing the concrete, however, had an unan- work) as well as who comprises the role set. If role ticipated side-effect. Studying practices led schol- relations change in either way, then the social net- ars away from asking how technologies alter work work may change and, if it does, one can say the systems and the structures of organizations in technology has altered the work system. Changes in which they were deployed. Agency and interpreta- role relations are, therefore, key to changes in work tion came to the fore; work systems and social systems. 2015 Barley 33

Using a combination of ethnographic and social of whether studying changes in roles requires fi eld- network data, Barley illustrated his analytical ap- work, role analysts also face diffi cult epistemic proach for linking changes in patterns of practice problems. First, to the degree that role-based stud- and interaction to changes in structure by showing ies focus on patterns of relationships defi ned largely how computerized imaging modalities (like ultra- in network terms, one risks losing sight of how the sound and CT scanners) challenged professional technologies alter the tenor of the actual interac- dominance in radiology departments. Historically, tions that constitute the ties among members of a the production and interpretation of x-rays was role set. We learn who interacts with whom and po- bifurcated: technologists produced fi lms and radi- tentially about what, but not how they enact their ologists interpreted them. With the arrival of com- relationships. To study how people play their roles puterized modalities, the separation of production we need a way of documenting repetitive patterns and interpretation broke down. Technologists who of typical encounters. operated computerized modalities could not pro- Second, role-based studies have at least implic- duce medically viable studies without understand- itly treated new technologies as exogenous shocks ing something about how computers functioned that break a prior social order. In fact, analysts— and, more importantly, without being able to inter- even those who hope to avoid technological deter- pret images as they produced them. Because many minism and who appreciate that the actions of radiologists initially knew little about the new im- people and the particulars of a context will shape ages or the technology, role relations morphed. Re- the chain’s unfolding—often construe technology lations between radiologists and technologists who as a trigger in a causal chain. Barley’s model, in par- operated computerized modalities became more ticular, is sequentially causal. He proposed a se- collaborative and technologists acquired more au- quence leading from changes in technology, to tonomy. As these differences solidifi ed, radiology changes in tasks, to changes in role relations, to departments split into two work systems: a hierar- changes in network structures. If the chain is bro- chically organized main department run by admin- ken, the work system is unlikely to change signifi - istrators and the computerized departments run by cantly. Such a model may be useful for studying the radiologists where relations were more collegial implications of discrete, physical machines like CT and reciprocal. scanners or even techniques like surgery whose oc- Subsequent studies have occasionally offered currences can be identifi ed and counted, but it is role-based accounts of how new technologies alter less viable for studying technologies that lack tangi- work systems. Edmondson, Bohmer, and Pisano (2001) bility or even ostension. This is precisely the prob- examined the deployment of minimally invasive lem posed by the internet. cardiac surgery at 16 hospitals. They showed that The internet is a cover term for a complex, pliable, successful implementations occurred when rela- changing and every-expanding portfolio of tools, tions among surgeons, nurses, technicians, and an- information, and media that alters the grounds on esthesiologists became less hierarchical and more which we act in situations where we previously collaborative. In one of the few studies of how the would have acted differently. There is nothing mys- internet alters role relations and, by extension, tical about the internet’s capacity to reground action; work systems, Schultze and Orlikowski (2004) stud- the phenomenon is entirely practical. It involves ied an online quoting system deployed by WebGA, our ability to do things that, in similar situations, a fi rm that assisted independent insurance agents we could not have done so easily before the inter- in determining which insurance companies offered net. For example, before the internet it was impos- the best policies and prices for the clients the agents sible to communicate instantaneously as well as served. To WebGA’s chagrin, the tool allowed agents asynchronously across time and space. Before the to generate their own quotes, thereby bypassing internet, it was diffi cult to access vast bodies of in- WebGA’s representatives. As a result, the technol- formation without leaving one’s chair to visit a re- ogy decreased the frequency and altered the tenor pository. Before the internet (and smartphones) one of interactions between service representatives and could not resolve disagreements over facts during a agents. Ironically, this change undercut the benefi t dinner conversation with one’s spouse in a restau- that agents once saw in working with WebGA’s ser- rant. (Of course, this may still be ill-advised.) With vice representatives. the internet, people now have easy access to infor- Nevertheless, role-based studies of how technol- mation (e.g., what a particular car cost an automo- ogies alter work systems remain relatively rare. One bile dealer in Tehachapi, or knowledge that there is might argue that their rarity refl ects the fact that a specifi c person in Tonapah who wants to start a such studies usually involve ethnography, which is relationship with a person more or less like you) too time-consuming for most scholars. Irrespective that they previously could not have found. 34 Academy of Management Discoveries March

To analyze such regroundings, we need concepts The key, for Goffman, did not involve understand- for grappling with how technologies are enrolled in ing individual-level sense-making as occurs when the day-to-day fabric of social action such that they people attempt to comprehend the unexpected, but reconfi gure the web of constraints and opportuni- rather sense-making and behavior that is contingent ties that shape our behavior in specifi c contexts. In on a “defi nition of the situation” or what he later other words, we need a way to map what Orlikowski called a “frame” (Goffman, 1959, 1974).2 (2007) has called “constitutive entanglement:” how Although defi nitions of a situation are sometimes the social and material entwine. Actor-network the- negotiated on the spot (as when we explicitly ask orists (Eriksson & Goldkuhl, 2013; Latour, 2007; each other, “What’s going on here?”), they are not Ribes, Jackson, Geiger, Burton, & Thomas, 2013) usually constructed de novo and they are not un- and students of sociomateriality (Leonardi, 2013; constrained. Constraints arise from how activity is Orlikowski, 2007; Orlikowski & Scott, 2008) have embedded in and arises from a context. Once we attempted to grapple with such entwining. Actor- classify some goings-on, we generally know how to network theorists do so by reducing technologies and act precisely because we treat it as a typical case of humans to analytical equivalents, known as “act- some social scene. Goffman argued that frames are ants,” and by studying their joint contribution to a layered contexts that confi gure the lines of action system of action. In the process, however, actor- that participants can take in a particular type of network theory sacrifi ces our everyday understand- encounter. Importantly, these layers include the ing of the differences between technology and “physical framework,” by which Goffman meant people in ways that sometimes border on the anthro- artifacts and natural phenomena, and the “social pomorphic. Furthermore, actor-network theorists do framework,” by which he meant the other actors not focus on roles and situated interactions. Advo- who are present as well as the institutional setting cates of socio-materiality, on the other hand, offer a (such as a theater, a church, a lecture) that provides useful ontology, but little guidance about how to de- specifi c rules for acting (Goffman, 1974: 21–25). ploy the ontology empirically. Their analyses tend to People cannot ignore the physical framework any gravitate to the micro-practices of how people use more than they can ignore the social. As Pinch technologies rather than to people’s interactions (2010) recently pointed out, Goffman understood with each other, which role theory takes as the build- that people take into account a setting’s material ar- ing blocks of all social organization. What would be tifacts as they enact encounters. In fact, technolo- useful are concepts and schemas that allow us to un- gies assist people in defi ning a situation and in ravel how new technologies become entangled with playing their roles accordingly. Once a situation is the social such that they reconfi gure what Goffman defi ned, people engage in “guided doings” (Goffman, (1983) called an “interaction order”: the situated, 1974: 23); they construct a line of action within the patterned, and recurrent ways of behaving and inter- confi nes of the possibilities set by physical and so- acting associated with a particular context. Ways of cial frameworks. It is doubtful, for example, that behaving toward others, which constitute the rela- one would skateboard down the aisle of a church, tional aspects of roles, are precisely what are lost in especially while a priest is delivering a sermon. The most previous research on technological change, frame, “attending church,” largely precludes such sometimes even that which attends to roles. activity. One could conceivably skateboard down a Dramaturgical analysis. On this score, Goffman’s sidewalk while listening to the same sermon on an (1959, 1961a, 1967, 1974, 1981) essays on encoun- iPod. Re-groundings involve changes in frame. ters, interaction rituals and forms of talk are helpful Like other dramaturgical sociologists (Messinger, because they provide a schema for analyzing the Sampson, & Towne, 1962; Turner & Edgley, 1976; ways people play roles vis-à-vis each other and how Wilshire, 1982), Goffman argued that encounters, settings and artifacts (including technologies) sup- interaction rituals, and other “guided doings,” once port and limit role-playing. Over his career, Goffman sought to understand why social life is so well- ordered and to reveal why encounters of particular 2 One should not confuse Goffman’s use of “frame” types (a lecture, a game of checkers, and so on) un- with its use by contemporary institutional theorists who fold so similarly regardless of who is involved.1 often use the term as a synonym for a logic. Among insti- tutionalists, frame has the connotation of a rhetoric, con- 1 Although some scholars claim that Goffman was pri- cept or ideology. In Goffman’s terminology, frames defi ne marily interested in the construction of selves and iden- classes of situations which exist at a much more situated tities and the negotiation and maintenance of face, his level of analysis. Goffman used frame in the sense of a agenda was ultimately to explain the social structure of picture frame, a set of cues or parameters that differenti- everyday life (Collins 1988, 2004; Giddens 1987). I draw ates the situation from its surroundings and that defi nes on Goffman’s structural agenda in this essay. acceptable lines of action within the context. 2015 Barley 35 framed, can be usefully viewed as theatrical perfor- asking them to seat themselves and bus their own mances.3 The dramaturgical eye draws the analyst’s tables, turns a fi ne restaurant into a fast food attention to specifi c elements of the encounter and establishment. its setting that support and structure the unfolding Scripts unfold through moves, the actions that a line of action. Scripts are the pivotal element of any second party takes in response to a fi rst party’s ac- dramaturgical account. A script organizes and typi- tions, that in turn serve as clues and cues for the fi es interaction by defi ning how parties to an en- next actions the fi rst party should take (see Collins, counter should play their roles. The vast majority 2004). Moves are strategic in intent; through them of encounters have scripts even if they are as mini- actors hope to achieve objectives. Encounters also mal as the two-turn exchange of greetings among vary by their footing, the stance that actors assume strangers acknowledging each other’s presence on a toward each other as they make their moves. Foot- sidewalk. One can think of a script as the plot of a ing sets the tone of the encounter. When you hand recurrent activity that defi nes the essentials of the an immigration offi cial your passport, you engage parts that participants must play. Consider going to in roughly the same behavior as when you purchase a restaurant (Schank & Abelson, 1977). When we go a movie ticket: in both cases you hand a legal docu- to a restaurant, the stream of action unfolds in a ment to someone in a glass booth who then offers relatively predictable order: the maître d’ greets us; admission beyond. In the case of immigration, we ask for a table; the maître d’ seats us; the waiter however, you present yourself as a supplicant for greets us at the table and hands us a menu; the admission which the offi cial can question and waiter asks if we’d like something to drink; and so deny. In the case of going to a movie, you present on through dessert and paying the bill. As Garfi nkel yourself as a customer and you do not expect to be (1967) demonstrated, when scripts are breached— denied admission unless the show is sold out. To as would occur if the waiter immediately asked if appreciate the difference in footing one need only we would like dessert—we become disoriented, even consider the relative propriety and the conse- anxious or angry. quences of expressing irritation at the ticket taker Scripts are both cognitive and behavioral phe- versus an immigration offi cial. nomena. Cognitively, scripts are expectations about As mentioned above, other dramaturgical elements how things ought to go, given a defi nition of the of an encounter support and shape the scripted line situation. When our expectations are fulfi lled we of action and the moves by which actors play their have the sense that “nothing-out-of-the-ordinary-is- parts. Like plays, encounters always occur on some going-on-here,” although we are not likely to be stage often in front of some audience, which might aware that we are experiencing a sense of the usual. be present or imagined. The stage is bounded in Behaviorally, as the dramaturgists emphasize, time and space, demarking the region within which scripts are loosely prescribed sequences of behav- an encounter of a particular type can and should iors and interactions associated with types of en- occur. Usually, settings have front stages that are counters. People enact or play their respective roles available to the audience and back stages that are by making key moves at appropriate points in an not. Moreover, stages usually contain props that interaction’s unfolding (see Pentland, 1992). A buttress the line of action and the actors’ perfor- script’s enactment is supported by, and may often mance of their roles. Stages and props are the mate- require, a specifi c stage (a restaurant in the evening), rial artifacts (some of which may be considered a set of props (tables, plates, candles, a corkscrew), technologies) that actors employ as they play out supporting actors (bartenders, bus boys) and some- encounters of different types. One of Goffman’s key times an audience (other diners). When an en- points is that actors cannot sustain an encounter of counter’s dramaturgical elements change, scripts a given type without the right stage and the right may change and when scripts change, by defi ni- props. If you change the props or the stage, you risk tion, the interaction order has changed. For exam- changing the defi nition of the situation. For in- ple, asking customers to pay before they eat food stance, one cannot play chess without a chessboard that was cooked before they ordered it and then and chess pieces, however contrived or even imagi- nary they might be. Similarly, to dine in a restau- rant requires the presence of tables, chairs, plates, forks, knives, and spoons. In restaurants people do 3 To argue that encounters are rituals or theatrical per- formances is not to say that actors are somehow disingen- not stand or eat with their hands. At picnics they uous or that their actions and interactions are contrived, may. The point is that technologies and other arti- although the scripted aspects of encounters do make pos- facts structure our defi nition of the situation, the sible the kind of dissembling that drew Goffman’s atten- scripts that guide the encounter, and the way we tion to confi dence artists. play our roles. 36 Academy of Management Discoveries March

It is important to note that scripts, roles, moves, turns to the research that produced the data on props, stages, and footing implicate each other. which the analysis rests before moving on to com- They operate jointly to defi ne the situation and, pare how sales encounters unfold when customers hence, the line of action that the situation warrants. purchase a vehicle over the internet instead of in To see how this works, consider again encounters person. with immigration offi cers and people who sell tick- ets at movies. Even if you assume the footing of a METHODS supplicant, if you make the mistake of handing the immigration offi cer offi cial documents known as Background: How Americans Buy and Sell Cars money, you may fi nd that there are worse possibili- Americans buy new and most used cars from a ties than being denied admission to the offi cer’s dealer, a retailer franchised to sell one or more homeland. Of course, in some countries, presenting manufacturers’ vehicles. Dealers purchase new ve- the offi cer with money might speed entry, as long as hicles from the manufacturer at a “dealer’s invoice you make the move properly. In contrast, handing a price” plus a “destination charge” (the cost of ship- ticket seller your passport is likely to invoke little ping the vehicle to the dealer). Once the vehicles more than puzzlement and under no circumstance arrive, the dealer displays them on a “lot” sur- would it land you a better seat in the theater. rounding the dealer’s buildings. Dealers can offer In sum, a dramaturgical analysis of technological cars to buyers at any price, but most offer new cars change differs from other approaches to studying at the “Manufacturer’s Suggested Retail Price” technology. Rather than make the technology the (MSRP, also known as the “sticker price,” which center of attention it shifts attention to the system can vary widely by the “options” on a vehicle) plus of actors, actions, and interactions in which the a destination charge and other incidental costs. technology is embedded. We are no longer simply In the U.S., purchasing a new car is a nearly interested in how a technology changes the way we unique shopping experience. Whereas Americans do some task, a pattern of communication, how readily pay the asking price for most goods, they people conceptualize the technology, or even the expect to negotiate when purchasing a car. As one practices that emerge around a technology’s use. In- website oriented to educating would-be car buyers stead, dramaturgical analysis attends to the entire puts it: “Buying a new car is more like haggling for milieu that fl ows from and sustains a defi nition of a donkey in Marrakech than buying a refrigerator at the situation and plausible lines of action. If any- Sears.”4 In general, buyers can expect to pay some- thing, it is the script rather than the technology that where between the invoice price and MSRP, al- lies at the core of the analysis. The analyst’s job is though dealers may sell overstocked cars at less to determine whether the presence of a technology than invoice and models in high demand at more has somehow reconfi gured the scripts, the stage, than MSRP.5 Thus, the car buying game has histori- the props, the moves that actors make, and the en- cally been one in which the car salesman attempts counter’s footing in ways that sustain a different to extract a price as close as possible to the MSRP line of action. Because scripts, by defi nition, en- (because commissions are tied to the size of the dif- code role relations, dramaturgical analysis there- ference between the selling and the invoice price), fore brings the playing of roles to center stage. while the customer attempts to purchase the car Furthermore, from this perspective internet tech- closer to the invoice price. nologies are analytically no different than any other Because haggling is foreign to Americans, many technology; to have social consequences, using the fi nd purchasing a car to be stressful. Homes are the internet must reconfi gure some combination of the only other purchase about which Americans expect scripts, moves, actors, footing, props, and stage that structure an interaction ritual to transform the way a type of encounter unfolds. To appreciate how a dramaturgical approach 4 “What Dealers Won’t Tell You About Car Prices” might assist us in studying how technologies alter (www.leaseguide.com/Articles/carprices.htm. Accessed role relations and the work systems they inscribe, February 5, 2011). 5 let us explore how the internet is reshaping the Even when dealers sell at or below invoice price they usually make money on holdbacks. Holdbacks are a per- way Americans buy and sell cars, and more spe- centage of the invoice price (usually 2 to 3 percent) that cifi cally, the structure and tenor of encounters be- the manufacturer returns to the dealer after the car has been tween car salesmen and customers. After briefl y sold. Because dealers borrow money based on invoice describing the status of car salesmen in American price and destination fees, holdbacks allow dealers to culture and how the internet occasioned an alter- borrow more money. Holdbacks also allow dealers to pay native way of buying and selling, the narrative higher commissions to salesmen and guarantee a profi t. 2015 Barley 37 to negotiate and, in this case, most buyers and sell- these leads would be less expensive than the ers hire agents to do the haggling for them. Al- cost of generating customers through tradi- though being a salesman of any type carries stigma tional advertising, as well as resulting in a in the U.S., few salesmen are perceived to be less reduction in labor costs. Because advertising prestigious or trustworthy than a car salesman. and labor account for about 60 percent of most Nakao and Treas (1994) found that a nationally dealers’ operating expenses, these savings were representative sample of Americans gave car sales- large enough that a signifi cant discount could men a score of 25 on an occupational prestige scale be passed on to the buyer.8 ranging from 1 to 100; lower than the ranking of all In 1997 General Motors became the fi rst auto other sales occupations except pushcart vendors maker to announce it would sell cars online. Costco (22), telephone solicitors (21) and newspaper ped- entered the online market for autos in 1998 as an dlers (19). A search of JSTOR for articles in aca- “affi nity group” that negotiated prices with dealers demic business journals that contain either “car on behalf of its members. By 2007, 67 percent of all salesman” or “automobile salesman” in their text car buyers had used the internet to research their provides further evidence of the occupation’s per- purchase, 27 percent said that the internet had in- ceived moral stature. Of the 72 papers that met the fl uenced their purchase, and 12 percent bought from criterion as of 2011, 53 percent (38) were published a dealer recommended to them by an internet site in the Business Ethics Quarterly or the Journal of (J.D. Power and Associates, 2007). By the same year, Business Ethics or had words like “trust,” “truth,” 94 percent of all dealers in the U.S. had an internet “ethical,” “credibility,” “lies,” and “puffery” in site (J.D. Power and Associates, 2007). Morton et al. their titles. In a 2009 Gallup Poll, only Congress- (2001) reported that buyers who used independent men were rated as having lower ethical standards car buying sites paid, on average, 2 percent less for than car salesmen.6 Americans believe that car their car than did customers who bought from a salesmen routinely take advantage of customers, a showroom. The same researchers (Morton et al., belief not without warrant (Browne, 1973; Jacobs, 2003) also reported that the internet eliminates the 2001; Lawson, 2000). well-known tendency for dealers to charge minor- In the mid-1990s, websites began to appear offer- ity customers more than Caucasians in face-to-face ing car buyers an alternative to negotiating face- encounters (Wise, 1974). Although these and other to-face with a salesman. In 1995 Autobytel became studies (Ghose et al., 2007; Viswanathan et al., 2007; the fi rst website for researching and purchasing Zettelmeyer et al., 2006) imply that the internet automobiles online.7 Autobytel’s business model must somehow alter the social dynamics of the sales was: encounter, they do not tell us in what way or how. . . . to empower the car buyer while providing dealers with a more effi cient sales process. The company would offer the buyer, at no charge, Data Collection pertinent information, such as the dealer’s true Two primary research designs allow analysts to cost, and an online request form detailing the identify how a new technology changes work prac- car and desired options. A local participating tices and work systems. The strongest design is to car dealer, which would pay a fee to Autobytel, study the same work system longitudinally, before would then contact the buyer with a no-haggle and after the technology’s adoption. The second ap- price, and offer an opportunity for a test drive. proach is comparative: to study a work system that Dealers would benefi t because Autobytel sales employs the technology and one that does not, ide- were essentially found business. Moreover, ally within the same organization. Even stronger would be to replicate the comparison in two or 6 “Honesty and Ethics Poll Find Congress’ Image more similar organizations to avoid confusing site- Tarnished” (December 9, 2009, www.gallup.com/poll/ specifi c dynamics with the effects of the technology. 124625/honesty-ethics-poll-finds-congress-image- The current study adopted the latter approach. tarnished.aspx). The data were collected by participant-observation 7 It is worth noting that Autobytel was also one of in two car dealerships in a metropolitan area in the earliest internet businesses. In 1995, the World Wide Northern California over an 18-month period Web was still relatively undeveloped. According to the Smithsonian Institution there were only 100,000 active websites of any kind in 1995. Two years earlier there were but 623. (National Museum of American History “Birth 8 FundingUniverse 2011 (http://www.fundinguniverse. of the Internet Timeline” http://smithsonian.yahoo.com/ com/company-histories/Autobytel-Inc-Company-History. timeline.html.) html). 38 Academy of Management Discoveries March between February 2006 and October 2007. One accompanied the salesman from the start of the dealer sold Chevrolets and the other Toyotas, al- sales encounter until the customer left. When not though both also sold related brands (GMC trucks shadowing salesmen, the researchers talked with and Scions, respectively). Each was ranked by its the sales staff about their work, their careers, their manufacturer among the top 10 dealers in its region experiences selling cars, how the dealer and the by sales volume. Both had separate staff for fl oor manufacturer worked together, and the procedures and internet sales. The Chevrolet dealer employed the dealer used for ordering automobiles, fi nancing six fl oor salesmen and two internet salesmen, while sales, and tracking information. In addition, the re- the Toyota dealer employed eight salesmen of each searchers attended weekly sales meetings and in- type.9 Internet sales constituted a larger portion of terviewed owners, sales managers, and fi nance the sales at the Toyota dealership. managers. Although observation focused primarily Floor salesmen at both sites worked the showroom on fl oor sales, the researchers fortuitously docu- and the lot, the stages on which they encountered mented several internet sales. customers. They and their managers had offi ces on After spending several months reading, coding, the edges of the showroom, where the dealers dis- and analyzing the fi eldnotes from the fi rst phase of played several of the manufacturer’s most recent the study, we concluded that there were no impor- models. Both dealers displayed the remainder of tant differences in how Toyotas and Chevrolets their inventory on parking lots adjacent to their were sold. Unexpectedly, however, there did seem showroom and, at the Toyota dealership, on the roof to be salient differences between fl oor sales and the of the building. At both sites internet sales occurred internet sales that we happened upon. Moreover, backstage out of sight of the showroom.10 At the Toy- the differences seemed consistent across the dealer- ota dealership, internet sales occurred on the second ships. We were stuck by our sense that encounters fl oor adjacent to the business offi ces. At the Chevro- between customers and internet salesmen evinced let dealership, they occurred in offi ces on a back little of the tension and animosity that often marked hallway on the fi rst fl oor. Although the internet sales fl oor encounters. We also noted that price seemed staff at both locations occasionally worked the fl oor, to be the fi rst topic of conversation in internet sales fl oor salesmen never handled internet sales. while it was usually the last topic of conversation We did not set out to compare fl oor and internet during a fl oor encounter. At the time, however, we sales; in fact, we began the study unaware of how had not observed enough internet sales to confi rm strongly the internet shaped sales encounters. Our that the differences occurred with regularity or initial agenda was to compare how Toyota and what they implied for relations between custom- Chevrolet dealers operated, believing that we might ers and salesmen. All we knew was that the differ- fi nd differences analogous to those that others have ences between fl oor and internet sales could not found in how the two manufacturers design and be explained by whether customers had consulted make vehicles (MacDuffi e, 1995; Womack, Jones, & the internet before contacting the dealer. Many Roos, 1990). Our intent was to write about how customers who visited the dealers’ lots had previ- sales cultures differed by manufacturer. Accord- ously spent time on the internet researching cars ingly, during the fi rst nine months of the study we and arrived fairly well informed. Accordingly, we focused almost exclusively on fl oor sales. During shifted our attention to comparing internet and this phase, three doctoral students conducted fi eld fl oor sales. work at both sites. The researchers usually situated To rectify the scarcity of data on internet sales, themselves in the dealers’ showrooms, where sales- we again took to the fi eld during the summer of men waited for an “up”: a chance to serve a cus- 2007. During the second period of data collection tomer.11 When a customer arrived, the researchers we focused explicitly on comparing the dynamics of fl oor and internet sales encounters. The three re- searchers involved in the second phase of the study 9 We encountered only two female salespersons in the (one of whom had participated in the fi rst) concen- course of our study, which is consistent with the gendered trated on documenting sales encounters by shad- nature of auto sales (Lawson 2000). This paper uses the owing both fl oor and internet salesmen from an masculine term, “salesmen,” to emphasize the demogra- encounter’s start to fi nish. When agreeable to both phy of the occupation and to simplify the syntax. customer and salesman, we tape recorded the en- 10 Internet sales were part of direct sales, which also included “fl eet sales,” that is sales to companies that counters. We subsequently transcribed the tapes bought vehicles for their business. We observed only one and integrated the transcripts into the observers’ fl eet sale in the course of our study. fi eldnotes to produce a narrative that included a 11 Throughout the paper words in quotations signify verbatim record of what customers and salesmen slang terms employed by auto salesmen. did and said. 2015 Barley 39

Data Analysis how the encounter would unfold. We also coded passages that described points in the encounter at The analysis draws heavily, but not exclusively, on which the interaction became awkward or seemed the data collected during the second phase of the in danger of breaking down; for example, long un- study. Data from the fi rst phase inform the general broken silences and moments when a customer or a discussion. We documented 61 fl oor encounters from salesman contradicted what the other had said or start to fi nish, half at the Chevrolet dealership (31) challenged the other’s integrity or truthfulness as and half at the Toyota dealership (30). Because the when one of the parties accused the other of lying. Chevrolet dealer was less heavily committed to inter- Awkward moments disrupted the smooth fl ow of net sales, our data on internet encounters are skewed turn taking or punctured the aura of respect that toward the Toyota dealer. We observed and recorded people in interaction normally grant each other. 23 internet encounters: 19 at the Toyota dealership Having fl agged such passages, we met to talk about and four at the Chevrolet dealership.12 Because we the various passages we had fl agged and to develop could fi nd no signifi cant differences between the a preliminary system for classifying sales and cus- fl oor or internet sales across the two sites, aside from tomer moves as well as different types of awkward the infrequency of the latter at the Chevrolet dealer- moments. On the next pass through the data, we cat- ship, the paper combines the data from the two sites. egorized moves and awkward moments. The catego- To prepare for coding we read the fi eldnotes mul- ries of sales moves, customer moves, and awkward tiple times. We then met to devise an initial set of moments that are that pertinent to this paper are codes for chunking the text. Although we used listed in the three panels of Appendix A along with Goffman’s dramaturgical imagery as a general guide defi nitions and examples from our fi eldnotes.13 for coding lines of action, we had no preconceived notions, for example, of what moves we would 13 Note that our methods emphasize repeated behav- fi nd. In this fi rst pass we fl agged all passages that iors, which can be counted. It has become common pertained to a particular fl oor or internet encounter among contemporary ethnographers to equate ethnog- so that it would be easy to separate sales encounters raphy with documenting what informants say and to from the remainder of the fi eldnotes. For each en- elaborate upon informants’ interpretations of the phe- counter, we then coded the major steps in the en- nomenon under study. Although this is a viable ap- counter’s fl ow: for example, greeting the customer, proach to ethnographic research, as Kenneth Pike (1967) taking a test drive, making an initial phone call, ne- noted, ethnographies can be both emic and etic. Ethnog- gotiating, leave taking and so on. These codes al- raphies, particularly those based on participant observa- lowed us to identify the typical stages in a fl oor or tion as opposed to those based primarily on interviews, internet encounter’s unfolding (its script). often emphasize behaviors and social structures, usu- Next, for every encounter, we fl agged as a “sales ally together with but sometimes without giving atten- tion to emic interpretations. This was particularly true move” any part of an interaction in which a sales- of Chicago School ethnographies in sociology, the work man said or did something to infl uence the custom- of ecological psychologists such as Roger Barker (1963) er’s perception of the vehicle, learn more about the and also in anthropology, at least before the interpre- customer, persuade the customer to purchase the tive turn that Clifford Geertz (1973) stimulated, and that car, and so on. At the same time we fl agged as a was later elaborated by anthropologists such as George “customer move” any part of an interaction in Marcus, who founded the journal Cultural Anthropol- which the customer said or did something to infl u- ogy. In fact, there have been ethnographers who provide ence the salesman, end the encounter, or shape counts (see Dougherty, 1985) and some, such as Per Hage and Frank Harary (1991), who have made extensive use 12 Desiring a more balanced proportion of internet of mathematics and social network analysis, going well encounters across the two sites, we repeatedly visited beyond counting. There is an entire journal devoted to the Chevrolet dealer at various hours of the day and on mathematical anthropology: Mathematical Anthropology various days of the week explicitly to observe internet and Cultural Theory. In short, ethnography and quanti- sales. However, the area was rarely staffed when we were fi cation are orthogonal to each other (Barley, 2004). It is on site. If we were examining statistical or interpretive not whether you use numbers, but how you use them that rather than structural differences across the two sites, matters. Ethnographers employ counts and provide per- the smaller amount of data on internet sales at the Chevy centages not to test hypotheses, but to show the reader dealer would have posed a signifi cant problem. However, the empirical warrant for inferences about the behavioral because we focus on the scripts of sales encounters and patterns detected in observational data collected through because internet sales at both sites followed identical participant observation. The defi ning characteristic of an scripts, the analytical problem is less serious. Neverthe- ethnography—which stands above the interpretive fray— less, the small number of observations of internet en- has always been long-term immersion by the researcher counters that we were able to observe at the Chevy dealer in a culture or subculture to understand and subsequent- represents a potential limitation of this study. ly describe the ways of a people. 40 Academy of Management Discoveries March

The subsequent ethnographic descriptions of approached the customer even if he suspected the fl oor and internet sales weave key customer and customer was a poor prospect. sales moves, stages, props, and supporting actors As Figure 1 depicts, fl oor encounters usually un- into narratives of the scripts by which the two types folded according to a script parsed into three scenes of sales encounters typically unfolded. The analy- known to salesmen as: “landing the customer on a sis then substantiates that the tenor of fl oor and in- car,” “taking a test drive,” and “doing the paper- ternet encounters differed by examining how work.” Each scene usually occurred on a different customers reacted to each and by examining the stage: on the lot, in a moving car, and in the show- relative frequency of moves and awkward moments room, respectively. In each scene salesmen and that we observed. The analysis then draws on the customers engaged in a set of scene appropriate data to offer a succinct dramaturgical account of moves. Although customers could end encounters why the internet altered interactions between sales- at any time, in practice, most fl oor sales were termi- men and customers, before outlining how research nated at one of three transitions: just before the test on technology and organizing might benefi t from drive, just after the test drive, or during the paper- dramaturgical analyses of role relations. work when it became clear that the customer would not buy or that the dealer and customer could not agree on a price. Salesmen almost never ended an SALES ENCOUNTERS encounter once it began, but they could accept Floor Scripts without protest the customer’s defi nition of the en- counter as over. Alternately, they could attempt to Waiting for an up. Except on particularly busy extend the encounter, if they thought the odds of a days, fl oor sales were a waiting game. Each deal- sale were high enough. The decision to do one or er maintained a rotation schedule that listed the the other was always determined in situ. names of the salesmen on duty in repeating se- Landing the customer on a car. Encounters quence. Whoever’s name appeared next on the list began when a salesman greeted the customer, in- was “up,” with the right as well as the duty to take troduced himself, and asked an opening question the next customer. Once a salesman had taken his designed to reveal the customer’s preferences for or her “up,” the top name was crossed off the list a model or a specifi c vehicle. Although salesmen regardless of the encounter’s outcome, and he or sometimes opened encounters with small talk she waited until the name again fl oated to the top (“Great weather we’re having today!”), they quickly before taking another customer. moved to questions that established the encounter’s Aside from assuring that every salesman had an footing as one between a salesman and customer: opportunity to sell and reducing confl ict over who “Welcome to (Dealership’s name). How may I help would serve a customer, the system of “ups” had you?” or more directly “What are you looking for to- several notable consequences. The most obvious to day?” Such questions set and signaled the encoun- a casual observer was that salesmen spent consider- ter’s footing: the salesman was the gatekeeper to any able time loitering around their stages, the lot and car a customer might want to view, drive, or buy. As the showroom, waiting for an encounter to occur. gatekeeper, the salesman became the dominant par- Although salesmen sometimes fi lled their time by ticipant in the sales encounter. doing miscellaneous tasks that the dealer needed Salesmen examined customers’ responses to done, studying the manufacturer’s literature, or re- opening questions for clues about their readiness to viewing the dealer’s current inventory, more often buy, what they might be able to afford (“What are than not they passed the time by watching TV, you driving now?”), and how the encounter was smoking, and talking to each other. Second, loiter- likely to unfold. Occasionally a customer voiced an ing in twos or threes contributed to some custom- interest in a particular vehicle in a specifi c color ers’ perceptions that salesmen lay in wait, ready to with particular options and even at a specifi c price, pounce on them the moment they arrived. Finally, especially if they had seen a vehicle advertised in a because salesmen earned most of their money newspaper or had made use of the internet. More through commissions, they detested “wasting an commonly, customers simply announced that they up” on a “looker” or “drooler,” a window-shopper wanted to look at a particular model: a Suburban, a who had no intention of buying. Floor salesmen, Camry, an Impala, and so on. The remainder did therefore, often delayed initiating encounters, be- not know exactly what they wanted and told the cause they thought that if they waited, “droolers” salesman that they were simply “browsing,” “look- might reveal themselves as such and leave the lot ing at cars,” or “doing research.” A few even at- before the salesmen wasted the “up.” When it was tempted to avoid the encounter with such moves as clear that the salesman could no longer delay, he turning their backs and walking away from the 2015 Barley 41

FIGURE 1 The Floor Script

salesman, continuing to look at cars without ac- vehicle’s performance. In other words, the sales- knowledging the salesman’s presence, speaking to man used vehicles as props to support moves that each other in another language without indicating spurred the interaction forward. At the same time, that they knew English, and in one case, running the salesman attempted to establish “rapport” by quickly back to their car and driving away. learning more about the customer and by high- Assuming the customer did not immediately end lighting similarities between the customer’s and the encounter, the salesman soon began guiding the salesman’s lives and preferences. The particular customers through the lot (occasionally the show- moves in the conversation’s unfolding depended room) to a vehicle (or vehicles) similar to the one in on the salesman, the customer, and the vehicle be- which they had expressed interest. Although deal- ing shown. For example, if there was a child seat ers sometimes had the precise vehicle a customer in the customer’s car, a salesman might engage in wanted in stock (the right model in the right color conversation about children. Salesmen often asked with the right options), in most cases they did not. customers where they lived or how they planned Frequently, customers did not know what color or on using the vehicle. Then, they might express fa- options they preferred, which gave the salesman miliarity with the location or talk about how they leeway. In any case, the salesman’s objective at this used their own vehicle similarly, for instance, point was to “land the customer on a car”: to focus driving an SUV to ski resorts. The following ex- the customer on a vehicle and persuade him or her cerpt from our fi eldnotes provides the fl avor of to “take a test drive.” such conversations. Landing a customer on a car unfolded around two entwined lines of action. The salesman would A couple, in their late 20’s, walks into the show- “show” customers one or more vehicles by draw- room. Robert (the salesman) greets them. Mark, ing their attention to the vehicle’s features and by the husband, tells Robert they want to look at answering questions about those features or the Highlanders and 4Runners. They mainly want 42 Academy of Management Discoveries March

to sit in the vehicles so that he and his wife, Be- better.” She gets out of the car and we walk to the atrice, can narrow their choices. Robert asks if back of the SUV where Robert shows them how they need something they can take off-road, to to fold down the back seats. Beatrice goes back which Mark replies that Beatrice is “a SUV kind and sits in the driver’s seat again and Mark sits in of gal.” Robert opines, “That’s because she has a the passenger’s and they begin to talk in French. sense of safety.” Robert asks if this is their fi rst time at the dealership. Mark says, “Yes.” Mark starts talking about test-driving and says that the best thing would be to drive the two Mark asks Robert if Beatrice can see the High- models (this and 4Runner) back to back. Mark lander, so Robert leads the couple to the eleva- asks about the changes between the ’07 and ’08 tor and we head to the roof. In the elevator, Highlander. Robert explains that the ’07 was Robert says that his goal is to save the couple built on a Camry chassis and the ’08 is built time and money. As soon as we’re outside, Bea- on the Avalon chassis, which is “wider, longer, trice sees a 4Runner, gasps, and says, “I like and more comfortable.” Beatrice looks at the the dark grey.” She asks, “Can I look inside?” two models and says yeah that there’s a “huge Robert opens the 4Runner and Beatrice sits in difference. I like the new model.” Robert adds the driver’s seat. Beatrice tells Robert that she that “the body style is a bit more European” and drives a Jetta, so Robert says the 4Runner must asks if they have 10 minutes to drive this one. be a “bit of a shock.” Beatrice agrees and says Mark asks if it is possible to drive it back to back the front is very long and high and that there with the 4Runner. Robert says it is and gets their isn’t as much visibility. Robert then shows her driver’s licenses to copy. . . . (Toyota Fieldnotes) the height adjustment lever on the seat and Taking a test drive. Landing a customer on a car they fi ddle with that for a bit. Robert tells Mark always ended with the salesman asking customers that he would be “happy to arrange for you to (or vice versa) if they would like to take a vehicle drive any one of these models.” for a test drive. The vehicle to be driven was always a version of the model in which the customer was Beatrice gets out of the car and we walk to the interested, but it might be a different color or have back door of the 4Runner which Robert has different options. If the customer assented, which opened. Sounding skeptical, Beatrice remarks happened in a third of the fl oor encounters we ob- that she needs space for two dogs. Robert asks served (20 of 61), the salesman would ask the cus- if they are going to tow anything and Mark says tomer for his or her driver’s license, photocopy it in no, he only needs a bike rack. Robert says they the showroom, and retrieve the keys to the vehicle might want to look at the Highlander then, and from a locked key box. Depending on whether the tells the couple, “We have a lot of them down- customer was alone or with friends or family, on stairs (on the main lot).” Mark tells Robert that the preferences of the latter, and on the size of the they would like heated seats but he knows that vehicle, the salesman might sit in the passenger’s those usually come with higher end models seat beside the driver or in the backseat. During most that also come with leather. Mark asks if he can test drives, the salesman continued to build rap- get higher-end features without the leather, to port, answer the customers’ questions, and call the which Robert replies that they need to sit down customer’s attention to additional features, usually and look at their options. He adds that it can the controls on the dash including the sound and be confusing researching cars online because navigation systems. Conversations during test drives sometimes a website says something’s avail- also focused on how the customer experienced the able but it turns out that it might be available interior and the quality of the ride. Most test drives in Nebraska but not California. lasted no more than 15 minutes. The following pas- sage documents Mark and Beatrice’s fi rst test drive: Walking back to the elevator, Robert asks about their dogs and Beatrice says they have Labra- As soon as we get in the car, Beatrice says, dors. Once we’re in the showroom, Robert tells “I love that new car smell!” Mark tells her the couple that the Highlander on the fl oor is the that she shouldn’t think of that when decid- 2008 model but they have very few of these. ing which car to get. Beatrice then looks at the Robert leads the customers to a red Highlander touch-screen navigation display and asks what on the lot and opens the door for Beatrice to it is. Robert says it’s the navigation system and climb in. Robert says that the Highlander is really Beatrice responds, “Oh my God! I want that! I easy to drive. Beatrice says that she felt so tiny really like that!” Robert says that it only comes in the other car and that “this one is defi nitely in the Limited and then says that he think 2015 Barley 43

someone’s making an offer on the dark grey ’08 tense with the customer and salesman evincing model from earlier. He says they may get some increasingly less patience with each other. more in tonight but you can never tell what Paper and pens were critical props when haggling. color they’re going to be. Salesmen wrote down numbers for the asking price and interest rates on loans. They often wrote in large Robert (who drove the car off the lot) pulls into letters, underlined numbers to emphasize points, and the parking lot of a tennis court and Beatrice gets crossed off numbers as the negotiation proceeded. behind the wheel. Robert tells Beatrice, “You Some salesmen used the “4-square method.” They have a lot of seat adjustments” and proceeds drew lines to divide the paper into four quadrants. In to show her all the adjustments. Beatrice says the fi rst they wrote the sticker price, in another the it’s “too many buttons.” Once we start driving, value of the customer’s trade-in, in the third the down Robert says, “These have a power rear door.” He payment the customer was offering, and in the fourth adds that there is a button to the left of the steer- the monthly payment. Paper served not only as a ing wheel to open it and that Beatrice can open record, it provided salesmen with a tool or prop for the door, call her dogs, have them hop in, close negotiating and making points as well as observable the door, and go. Robert tells her she’s also got a evidence of where the negotiation stood. backup camera, but Mark says that doesn’t come During negotiations salesmen employed a number on the one they want so she shouldn’t consider of moves that they believed increased the odds that it . . . (After returning to the dealership, Robert a customer would agree to a price. For example, takes the customers on a test drive in a 4Run- they might simply ignore a customer’s comment if ner. At the end of the 2nd test drive, Mark has it threatened to sidetrack or derail the negotiation. decided that he likes the larger 4Runner and If the customer indicated that the initial price was Beatrice thinks she likes the Highlander better) too high and made an unreasonable counteroffer, . . . Beatrice turns to Mark and asks, “Do you the salesman would inevitably inform the customer want to think about it?” Mark tells Robert that that he would have to consult the sales manager. At maybe they just need to test drive the vehicles a this point, the salesman would leave the customer few more times. They make a plan with Robert and retreat into the recesses of the showroom, the to come back tomorrow to look at the cars some backstage, where the sales managers had offi ces. more. (Toyota Fieldnotes) Although the salesman would inform the manager of the customer’s counteroffer and the manager Doing the paperwork. If given the chance, after would suggest a price somewhere between the a test drive salesmen asked customers if they would initial quote and the customer’s offer, salesmen like to go to the offi ce “to do the paperwork,” a eu- often purposely delayed returning to the cus- phemism for “would you like to negotiate price, talk tomer for fi ve to ten minutes. It was common for about fi nancing, and buy this car?” More often than sales men to leave customers alone repeatedly dur- not, customers precluded the salesman’s asking by ing negotiations. The salesman would sometimes indicating that they were not buying today and by announce why he was leaving and at other times he thanking the salesman for his time. Of the 20 fl oor might not. Regardless, leave taking confused encounters that involved a test drive, 35 percent (7) customers and sometimes made them uncomfortable, proceeded further. Thus, roughly 1 out of 10 fl oor as illustrated by the following snippet from a encounters (7/61) made it to the point of price negotiation with an Indian couple who were negotiation. interested in purchasing a Corolla: Negotiations always occurred on the periphery of the showroom at the salesman’s desk. The After a few minutes Jeff (the salesman) returned salesman’s fi rst act, once seated, was almost always from checking on whether the dealer had a Co- to enter the customer’s demographic information rolla with a sunroof (which Jeff already knew into the computer, which allowed the dealer to they didn’t have) and asked the customer if he check the customer’s credit score. The salesman had to have a Corolla with a sunroof, because also reviewed the features the customer wanted, “right now we don’t have one with a sunroof.” quoted the customer a price, asked if the customer Once the husband said that a sunroof was not wanted fi nancing, and if so, told the customer the necessary and asked again for the monthly pay- terms on which the dealer would fi nance the ment, Jeff told the customer the annual interest vehicle (an interest rate that depended on the length for loans of various lengths. But he did not pro- of the loan). At this point, haggling usually ensued, vide a monthly estimate. So the customer again with the customer and salesman making offers asked, “How much is going to be the monthly and counteroffers. Haggling could become quite payment?” Jeff confi rmed that the customer 44 Academy of Management Discoveries March

wanted a 5-year loan and explained that he had to When customers made lower than acceptable run the data through a computer to get a monthly offers, salesmen often implied that the customer’s estimate. He again left the customers alone in his offer was insulting because it was unreasonable to cubicle without taking leave. The wife asked expect a dealer to sell below cost. A common move her husband, “Where did he go?” The husband was to tell the customer that the manager would not responded, “I don’t know.” (Toyota Fieldnotes) like the offer or that the dealer made little money selling cars: Another common move, used when a negotiation seemed stymied or an encounter seemed to be John (salesman): (The customer is negotiat- unraveling, was to enlist the aid of a supporting ing for a Camry. He has just told the salesman actor, another salesman. Sometimes, the two that he wants a price of $20,000 “out the double-teamed the customer. More frequently, the door.”) So you want a car for $20,000 out the original salesman “turned” the customer; the fi rst door, right? Well, we can’t do anything about salesman handed the customer over to the second, tax and licensing. You have to pay that no who often specialized as a “turnover guy.” Because matter what. the reason for “turning the customer” was never explained, the change often bewildered customers. Customer: Exactly. The new salesman hoped to use the bewilderment to redefi ne the encounter’s tone and tempo. If the John: You know I’m not the President. I can’t encounter had become strained, the second do anything about your taxes. If you want a salesman might try to establish a less contentious $20,000 car, then you should be looking at an relationship. In most cases, however, the second $18,000 car. You add all this up and that’d ba- salesman’s job was to “turn up the heat” which was sically be what you’re looking at. diffi cult for the fi rst salesman to do if he had established rapport. Customer: Yeah. Throughout the negotiation, salesmen employed a variety of verbal moves aimed at pressuring the John: This car retails for $21,000 alone. So you customer into accepting the dealer’s price. The basically want the out-the-door cost to be less most common was to create a sense of urgency or than what it retails for. I don’t know whether scarcity by telling the customer that this was the you’ve bought Toyotas before, but they don’t only vehicle with the options they wanted, by have too much profi t, especially a base model. implying that if the customer took more time to $20,000, I just don’t see that happening. Where think about the purchase, the dealer might sell the are you getting this idea that you can get it for car, or by telling the customer that the salesman $20,000? (Toyota Fieldnotes) could not guarantee as good a deal in the future. The Customers rarely endured such moves passively. following excerpt from a fl oor encounter illustrates Most expected salesmen to pressure them to buy a how salesmen created urgency and scarcity: car at a price higher than they thought they should Andy (salesman): I will do this for you. If you pay. Many came ready to do battle or at least protect want to buy right now, I’ll throw a fi gure on themselves from too easily falling prey. A signifi cant you and this is it. (He showed Yinyu a piece of number came armed with data from the internet, paper with a price written on it.) which salesmen routinely dismissed as inaccurate. It was also not uncommon for customers to be Yinyu (who was interested in a Malibu): Not accompanied by a spouse, a parent, a sibling, or a now, I cannot. Wait until Saturday. friend. These others often served as a skeptical voice for their companion who was negotiating the Andy: Listen, this business is like a stock mar- deal. In almost every case, companions who played ket! You came in today, if you want something the skeptic did so by depreciating, parrying, or you should buy today, because tomorrow I contradicting a salesman’s statements or suggestions. don’t know what is going to happen. If tomor- For instance, when a salesman asked a customer row is going good, it might not be like today. interested in a Tahoe if he “needed any other We haven’t sold a car yet today. We need to features like Bose or leather,” his wife immediately make a profi t, so we’re willing to sacrifi ce, to quipped, “Do you mean need or want?”14 open up our till. Saturday is a very busy day and we don’t drop down a lot on prices. Today, 14 Bose is the maker of a higher quality sound sys- during the week, the fi rst sale of the day, we’ll tem than the one with which the vehicle was normally work with you. (Toyota Fieldnotes) equipped. 2015 Barley 45

Another move customers employed during businesses whose services were free to everyone, negotiations was to mention another dealer’s quote for example, Autobytel.com, Autotrader.com, and to imply that they could get a better deal elsewhere. Edmunds.com. These sites allowed customers to On occasion, customers threatened to take their specify the vehicle they wished to purchase, the business elsewhere. Customers also sought an upper options they wanted, and their price range. The hand in sales encounters by disparaging dealers, sites provided invoice costs and price estimates manufacturers, and car salesmen, although they and referred customers who wished to be referred rarely disparaged the dealer they were currently to dealers in their area. Dealers paid a fee to the visiting or the salesman to whom they were online broker for each referral. Other brokers, talking. Disparaging comments occurred in a such as the American Automobile Association, quarter of the fl oor encounters we observed. Their Costco, and credit unions, offered services only gist seemed to be, “We are warranted in not trusting to members. These affi nity groups typically nego- people who sell or make cars.” The following passage tiated standard discounts with specifi c dealers or provides a sense of the kinds of disparaging manufacturers. In either case, the broker sent the comments that customers made during sales customer’s request for a price quote to a customer encounters: relationship management (CRM) system (a data- base) to which dealers subscribed. Customers who Paul (the customer) told Andy (the salesman), did not use a broker typically sent emails directly “I don’t want any black interiors. I don’t want to the dealer expressing interest in a vehicle. Oth- any white ones either.” Andy responded, “OK,” ers occasionally contacted the dealer by phone. The and after a few moments said, “They are really internet sales force logged email and phone inqui- good trucks. Chevy really did a good job this ries into the same CRM system that received requests year.” Paul shot back immediately, “It’s about from brokers. Regardless of the channel by which an time!” Paul is examining the manufacturer’s inquiry arrived, the internet sales force called such brochure on the Silverado and says, “Long-life information “leads.”15 rear axle fl uid. Could I get short-life rear axle Upon a lead’s arrival, the internet salesman fl uid?” He snickers, “Why would you even put contacted the customer, always by phone and often that down? Who would want anything else?” by email as well. The salesman’s objective was to (Chevy Fieldnotes) engage the customer in conversation and deter- If a customer and salesman agreed on a price, a mine in what vehicle he or she was interested. If deal was struck. Of the 61 fl oor encounters we customers could not be reached by phone, salesmen observed, 5 (8 percent) resulted in a deal. Only if left voice mail messages, recorded the attempt in the deal was closed did the salesman and the cus- the CRM database, and subsequently called again. tomer actually do the paperwork, which included a The following excerpt from our fi eldnotes illus- bill of sale, registration forms for the division of trates the kind of voicemail messages that salesmen motor vehicles, applications for dealer sponsored left: loans, and other incidentals. Hi, this is Ed from California Toyota. I sent you out an email quote and I wanted to make Internet Scripts sure that you received it for the Camry CE. If you’ve already bought your Camry, then As Figure 2 illustrates, internet sales followed a congratulations! I’d appreciate a courtesy call radically different script. Although some custom- either way, because then I can give you the ers bought a car over the phone before seeing it, appropriate attention. If you are still in the most successful internet sales involved two market, I’d like to answer any questions you encounters: The fi rst on the phone, the second have about the Camry and the different packages face-to-face. During the fi rst encounter internet available and even check on availability of salesmen could make no use of the stages on which colors. My phone number is 000-000-0000. fl oor encounters occurred and could not use the Thank you. (Toyota Fieldnotes) vehicles and other props (such as displays and brochures) on which fl oor salesmen relied. More- over, the two encounters typically occurred on different days. 15 Although the term, “leads,” is commonly used in The phone encounter. Internet encounters began many types of sales, it was never used by fl oor salesmen, when customers contacted a salesman via one of three because they rarely had prior knowledge of who might be channels. Most internet customers were referred interested in purchasing a vehicle and when they did, the to dealers by brokers. Some brokers were internet customers were called “referrals.” 46 Academy of Management Discoveries March

FIGURE 2 The Internet Script

Internet salesmen considered theirs to be a listed all vehicles in the inventories of nearby dealers “volume” business: to be successful one had to that he was free to quote, acquire, and sell. reply to (and repeatedly follow-up on) a large Second, once the customer showed interest in a number of inquiries. Volume was important because vehicle, the salesman told the customer the invoice most requests for information or a quote did not price (in contrast to the MSRP) and then quoted a evolve into a sale. Internet salesmen attempted to fi nal price by indicating how many dollars above contact each lead until they connected with the (and sometimes below) invoice he or she was customer, if only to learn that the customer was no willing to sell the car. In other words, unlike fl oor longer interested. Internet salesmen at the Toyota salesmen who began with the MSRP and worked dealership estimated that they processed ten times down, internet salesmen began with the invoice more customers than the fl oor salesmen. price and worked up. Unlike fl oor salesmen, the If a salesman managed to connect by phone and internet salesmen had to reel in their customers. if the customer was still interested, the subsequent Moreover, it was pointless to quote MSRP because conversation covered four topics. First, the salesman most internet customers already knew MSRP as confi rmed the model, color, and options that the well as the invoice price. Accordingly, they began customer preferred. As in fl oor sales, the dealer often at the breakeven price and worked up, knowing did not have the vehicle that the customer wanted that few customers would deny the dealer the right in stock. Accordingly, the salesman attempted to to make any profi t. Also unlike fl oor salesmen, interest the customer in vehicle close to what internet salesmen set prices using databases of customer requested or to “switch the customer up costs to guide their judgment. Only on rare or down” to another model. If the salesman was occasions did internet salesmen consult with a unsuccessful in getting the customer to consider a manager before quoting the customer a price. vehicle in the dealer’s inventory, he, unlike fl oor Third, after discussing price, internet salesmen salesmen, often turned to an online database that broached the topic of rebates and fi nancing, giving 2015 Barley 47 the customer a choice between one or the other. Mike tells the customer: “The Sequoia is a pret- Finally, if the conversation was progressing well ty well loaded vehicle, so that’s not a bad price! and the salesman thought the customer was likely The sticker’s at 36. With rebate it’s 31 and with to buy, he or she attempted to convince the customer the low interest rate, you defi nitely save a lot.” to make an appointment to come in and look at the car. The following except illustrates how phone The customer expresses interest in a Sienna LE encounters unfolded:16 (the customer hinted the price of a Sequoia is a little high) to which Mike responds, “A Sienna (“pause” indicates the customer is speaking.) might be a great alternative. We have aggressive The phone rings and Mike picks up the receiver: marketing on the Sienna. Five to six thousand “This is Mike Glenn from California Toyota Direct Sales. How can I help you? (pause) Yes, off.” (Looking at the price guide again) “We sell this is direct sales. (pause) You’re interested in Siennas for $200 under invoice.” a Sequoia 2-wheel drive SR? (pause) Who am I speaking with, please? (pause) Have you been “Let’s see if we can schedule a time?” Mike gives here before? (pause) Are you just starting the the customer his number and asks if he would process? (pause) So you’re looking to get this like to come in Saturday, Monday, or Tuesday. pretty quickly?” (pause) “That’s cool! Just give me a shout and let me know.” Mike again tells the customer his Mike begins to thumb through the price name and number. (Toyota Fieldnotes) guide (a white binder that contained a sheet of The face-to-face encounter. Customers who information on each vehicle in inventory). At agreed to an appointment arrived at the showroom the same time, he searches for the vehicle in like other car buyers. But when a fl oor salesman Advent Connect (an online database that indi- greeted them, they asked to see the internet sales- cates prices, monthly payments given various man with whom they had spoken (or his designated terms of fi nancing, and the like). Mike fi nds the proxy) by name. In response, the fl oor salesman or vehicle and reads the customer the informa- manager paged the salesman on the intercom an- tion, “MSRP on the vehicle is $37,769. Invoice nouncing that the customer was waiting in the is $33,511.” Mike also quotes the delivery showroom. On being contacted, the internet sales- cost. The customer indicates he has found a dif- man went to the showroom, greeted the customer, ferent price online. In response, Mike explains and inquired whether the customer wanted, as one that websites like Edmunds.com and KBB give salesman always put it, “to see the car fi rst or talk an invoice price but also say that dealers may about the numbers?” incur additional costs. He then goes back to Most customers chose to confi rm the terms of the the price guide and tells the customer, “We deal before examining the vehicle, in which case sell it for $300 over invoice. So, we sell it at the salesmen led the customer backstage to their $33,811.” He also tells the customer that there cubicle. With the customers settled into chairs and is a $2,000 rebate on the Sequoia, so the fi nal the salesman in front of his terminal, the salesman cost, $31,811, is “about a $6,000 discount off confi rmed the vehicle in which the customer was the vehicle” (meaning MSRP). interested and went through the pricing, just as he had done on the phone. Often they turned the “What would you put down? (pause) $4,000. monitor so the customer could see it, suggesting (pause) It depends exclusively on whether that they were hiding nothing from the customer. your credit would be an issue. (pause) 720? Once customers indicated that they were satisfi ed (the customer’s credit rating) That’s fi ne!” After with the price, salesmen asked if they would like to explaining the cost of fees, registration, and examine the vehicle. Most did. The salesman then sales tax, Mike quotes the customer a total cost took the customers directly to the lot where they of $36,649. (pause) “Let’s look at this a different examined the vehicle together as the salesman way,” Mike says, “Your monthly payment would called the customer’s attention to the vehicle’s be $957.” Mike realizes that this doesn’t sound features and answered the customer’s questions. right and immediately corrects his quote to The salesman then invited the customer for a test $728.52 per month with 48 months’ fi nancing. drive. Regardless of whether the customer decided to see the car or talk numbers fi rst, once the customer 16 In the case of phone calls, observers had access to agreed to the purchase, the encounter moved to only the salesman’s side of the phone conversation and the fi nal stage: completing the paperwork at the the salesman’s account of what the customer had said. salesman’s desk. 48 Academy of Management Discoveries March

In sharp contrast to fl oor sales, internet sales In sum, internet encounters inverted the script of were marked by a conspicuous absence of haggling. a typical fl oor encounter. On the fl oor, salesmen As a consequence, internet salesmen almost never showed customers vehicles and took them on test used the moves and ploys observed on the fl oor. drives before negotiating a price. In an internet In fact, the only moves that internet salesmen encounter, customers and salesmen settled the employed with any frequency were those targeted price at the start of the encounter, usually on the at neutralizing customers’ attempts to use quotes phone before the customer even came to the dealer. from other dealers as bargaining chips. When Internet sales were, therefore, more transparent customers attempted to use another dealer’s quote, for customers in the sense that they had a reason- internet salesmen sometimes offered to match the able idea of what they would pay from the begin- lower price, if they believed the quote was accurate— ning, thereby eliminating much of the uncertainty a behavior we never observed on the fl oor. More and anxiety that plagued fl oor encounters. In frequently, however, they explained why the other Goffman’s terms, the internet shifted the footing of the quote was not believable. Occasionally they sug- sales encounter: rather than casting salesmen and gested the customer should take advantage of the customers as antagonistic negotiators, internet sales- competitor’s price even though they were sure the men assumed the role of a giver of price and infor- quote was inaccurate: mation while customers became price takers. In short, the footing of the sales encounter resembled that of (Rob (the salesman) and the customers (a man the selling and buying in most other retail outlets. and his wife) are sitting at Rob’s desk.) Rob tells the customer, “The invoice price for a Camry LE is $19,890.” He turns his screen so the customer THE DIFFERING TENOR OF FLOOR AND can see the inventory of LE’s on the lot as well INTERNET ENCOUNTERS as their invoice prices. He points to the screen and says, “I’ll sell you any of these at cost.” Because of these differences in footing, props, “Is that better than the Costco price?” asks the staging, and scripting, the tenor of the interactions customer. Rob explains, “The Costco price is between salesmen and customers were noticeably a preset percentage above the invoice price. different in fl oor and internet encounters. The Since I’m willing to sell at invoice, the Costco difference was captured by the relative frequency at price is bound to be higher.” The customers talk which salesmen and customers employed moves together in their native language. The husband designed to gain advantage over the other and by then asks, “How about a quote on the SE?” the relative frequency of awkward moments during Rob looks back to his computer and reads off the course of an encounter. Table 1 displays the the screen, “I have 4 SEs left. Those I’ll sell for number and percentage of fl oor and internet $300 above invoice.” The customers talk quietly encounters during which we witnessed various in their native language. The husband asks, moves made by salesmen, moves made by customers, 17 “What’s your best price on the LE?” Rob answers and the occurrence of awkward moments. As the without hesitation, “That was my best price.” fi rst panel of Table 1 indicates, fl oor salesmen were The customer responds, “At another dealership, more prone than internet salesmen to employ moves 18 I got a quote for $18,850. I’ll even tell you where. designed to pressure customers. Floor salesmen It was Other-Side-of-the-Bay Toyota.” Rob looks more frequently: surprised and says, “Really? I’ll be right back.” 1. Told customers they could not offer a lower He leaves to consult with the internet sales price because they needed to make a profi t manager. A few minutes later, he comes back (21 percent vs. 0 percent); and sits down next to the computer, “What was 2. Attempted to create a sense of urgency and that price again?” The customer looks up the scarcity (23 percent vs. 9 percent); email quote on his Blackberry and tells him the 3. Deprecated other manufacturers’ vehicles same number. Rob says, “If that’s true, I suggest in which customers might be interested you drive over there right now and get that deal. (16 percent vs. 4 percent); But, I’ve been selling cars for a long time, and I bet they won’t sell you that car for $18,850.” 17 The defi nition of each move can be found in the sec- The customer looks back to his Blackberry and ond panel of Appendix A. rereads the email. A moment later, he says, “Oh, 18 The precentages in the tables and the text are offered wait, they quoted me a CE (a package with fewer as evidence for the warrant of the claims being made. options).” Rob smiles and replies, “Ah, you gotta’ They should be interpreted as a descriptive accounting. compare apples to apples.” (Toyota Fieldnotes) Hence, no statistical tests are offered or are appropriate. 2015 Barley 49

TABLE 1 Moves and Awkward Interactions in Floor and Internet Sales Encounters

Floor Sales (N = 61) Internet Sales (N = 23)

Number of Occurrences Percentage with Number of Occurrences Percentage with

Sales Moves All dealers tell you the same 4 7% 1 4% Want to buy or not? 4 7% 0 0% Manager’s not going to like that 1 2% 0 0% Need to make profi t/ commission 13 21% 0 0% Create urgency 14 23% 2 9% Deprecate alternatives 10 16% 1 4% Ignore customer comment 14 23% 0 0% Double team 7 11% 0 0% Turn the customer 10 16% 0 0% Neutralize internet data 5 8% 7 30% Any sales move 37 61% 10 43%

Customer Moves Disparage dealer or maker 15 25% 0 0% Ignore salesman’s comment 4 7% 0 0% Negative comment about car 5 8% 2 9% Obstinate about price 5 8% 0 0% Threaten to go to another dealer 6 10% 1 4% Use internet to check truthfulness 2 3% 3 13% Use other quote to bargain 10 16% 7 30% Deal is unacceptable 2 3% 0 0% Any customer move 28 46% 9 39%

Awkward Moments Customer confrontational 7 11% 1 4% Customer contradicts salesman 3 5% 0 0% Customer hangs up the phone 0 0% 1 4% Customer implies salesman lies 1 2% 0 0% Frame break 1 2% 0 0% Interrupting 5 8% 2 9% Rapport building fails 7 11% 0 0% Salesman is rude or belittling 7 11% 1 4% Salesman confrontational 3 5% 1 4% Salesman contradicts customer 7 11% 2 9% Salesman implies customer lies 1 2% 0 0% Silence 11 18% 0 0% Any awkward moment 16 43% 5 22%

4. Dismissed or ignored customer’s comments Differences in customers’ moves across fl oor and (23 percent vs. 0 percent); internet encounters were not as pronounced as the 5. Double teamed customers (11 percent vs. differences in the salesmen’s behavior. In other 0 percent); and words, the effect of selling over the internet had a 6. Turned customers over to another salesman stronger effect on how the salesmen played their (16 percent vs. 0 percent). roles than on how the customers played theirs. This is what one would expect given that the salesmen The only move employed more frequently by set the footing of the encounter as the gatekeeper to internet salesmen was neutralizing data that cus- the vehicle. Floor salesmen had most of the set- tomers had acquired while researching vehicles on ting’s resources at their disposal. Customers had the web (30 percent vs. 8 percent). Overall, one or few, except for information and their ability to walk more of the 10 sales moves in the fi rst panel of Table away. Nevertheless, we observed two important 1 occurred in 61 percent of the fl oor encounters, but distinctions that are documented in the second in only 43 percent of the internet encounters. If one panel of Table 1. During fl oor encounters, customers dis regards neutralizing internet data, the fl oor and were far more likely to disparage automobile deal- internet percentages would fall to 52 percent and ers and manufacturers (25 percent vs. 0 percent). In 13 percent, respectively. contrast, customers who bought through the internet 50 Academy of Management Discoveries March attempted more often to use another dealer’s quote they felt more in control and less intimidated than to bargain (30 percent vs. 16 percent) and to use in- they thought they would: ternet data to test a salesman’s truthfulness (13 per- Kevin looked online at Autotrader.com. He went cent vs. 3 percent). In general, making disparaging to other dealers and dropped by (the Toyota remarks about automobile dealers and manufactur- dealer) last night to check the condition of ers was a more vituperative act than using informa- the vehicle listed by internet sales. He called tion to negotiate a better deal. ahead to make sure the car was still here. He The relative frequency of awkward moments that test drove it last night, but since the car was a occurred during the fl ow of interaction offers addi- little more ($2,000) than his budget ($20,000), tional evidence on how the tenor of fl oor and inter- he had to go home and think about it. He went net encounters differed. As mentioned earlier, back online last night after he visited the dealer awkward moments called into question, however and looked at how “comparable” the other cars fl eetingly, the civility of the encounter. The third were. “These guys weren’t pushy,” and Kevin panel of Table 1 displays the incidence of twelve appreciated that. He added, “Some people are kinds of rupture that we observed in sales encoun- real pushy.” (Toyota Fieldnotes) ters. With the exception of long and uncomfortable silences (18 percent vs. 0 percent), no single type of One upshot of the inverted sales script and the awkwardness differentiated fl oor and internet en- less contentious tenor of the sales encounter was counters as strongly as did the moves employed by that internet salesmen were responsible for a sig- salesmen and customers. This is to be expected. Sales nifi cant percentage of the dealerships’ sales. At the and customer moves were legitimate within an en- Toyota dealership internet sales accounted for counter culturally defi ned as a negotiation, whereas 50 percent of the sales. At the Chevrolet dealership, awkward moments were more personal, had the where internet sales were not as heavily emphasized, feel of a faux pas and, in their most egregious form, internet sales still accounted for 25 percent of all threatened to transform a negotiation into a dis- sales. Internet salesmen estimated that 80 percent pute. Nevertheless when summed across encoun- of the customers who arranged a face-to-face encoun- ters, awkward moments of one kind or another ter eventually bought. Of the 10 face-to-face encoun- were twice as common in fl oor sales (43 percent vs. ters we observed, 4 or 40 percent resulted in a sale, 22 percent). and several more ended on the hopeful note that Because we were primarily interested in how the customer would buy within a few days to a customers behaved during sales encounters, we did week. Regardless of whether the rate was 80 percent not systematically interview customers about their or 40 percent, the yield for encounters that had be- reactions to an encounter. Nevertheless a number gun with the internet was much higher than the 8 per- volunteered assessments of their experience. In cent we observed for fl oor sales. general, their comments were in line with the rela- Differences in sales scripts and the roles that tive differences in moves and awkward moments. salesmen played in the two types of encounters Customers perceived fl oor encounters to be more were also refl ected the organizational structure of stressful, contentious, and threatening. Although the dealerships. Both dealers bifurcated their or- such comments rarely focused on the salesman ganizations: they treated fl oor and internet sales as with whom they were interacting, the comments separate departments and, as previously men- we heard on the fl oor were always critical: tioned, located internet sales at a signifi cant dis- tance from and, hence, out of sight of salesmen and Joan began by saying that [the car buying pro- customers on the fl oor. Although there was histori- cess] is horrible and she absolutely hated it. cal precedent for the separation (fl eet sales were “When I go to a dealership and they see a sin- also located backstage), there were strategic and gle woman walking in, they won’t talk straight. dramaturgical reasons for doing so. The dealers did They won’t answer my questions.” Her hus- not want internet encounters to contaminate fl oor band nodded and he added, “It’d be good if the encounters. Floor salesmen would have had diffi - buying process were streamlined and the peo- culty successfully playing out a fl oor script with ple straightforward. Just come out and provide their customers, if the customers, struggling to reasonable options for the customer to choose whittle away at MSRP, could observe other buyers from.” (Chevy Fieldnotes) being offered a set markup over invoice by sales- Many customers who bought through internet men who did not use pressure tactics. Furthermore, sales also anticipated an unpleasant experience, it would have been diffi cult for the dealers to keep but they noted that it had turned out better than their fl oor salesmen motivated if they could ob- they expected. In particular, customers told us that serve internet salesmen closing more deals than 2015 Barley 51 they did. In short, without segregation dealers could raise or lower seats, activate navigational software, not have maintained, as Goffman (1961a: 26) put it, or even employ a new car’s smell. Because they “a world of meaning exclusive” to the fl oor. could not show cars, take customers on a test drive, As typically occurs when organizations differ- provide brochures, peruse the automobiles that entiate, bifurcation sustained two different sales customers currently drove, or even see their custom- cultures with their unique rules and roles, and ers, internet salesmen were disadvantaged in creat- nurtured latent tension between members of the ing rapport, excitement, and a sense of urgency, two cultures, especially on the part of fl oor sales- which signifi cantly altered the relations between men. Floor salesmen disliked the internet because salesmen and customers. Because the customer they believed it made selling more diffi cult and re- was somewhere else, the salesman could not enlist duced their ability make money. As one Chevrolet the help of the supporting actors who played cru- salesman put it, fl oor salesmen viewed the internet cial parts in fl oor sales: the sales manager and other salesmen as the “give away guys.” Other fl oor sales- salesmen to whom they could turn the customer. men told us that the internet had taken “the fun and Even the minimal obligations of civil demeanor in challenge” out of selling because customers knew a face-to-face interaction were weak. Customers, in too much about the price that dealers paid for vehi- particular, were free of the obligations of leave tak- cles. Finally, because each dealer displayed the ing that attend face-to-face encounters. They did names of top salesmen prominently for all employ- not need to make an excuse for taking leave (“I need ees to see, fl oor salesmen were constantly con- to get home”) or even say “goodbye,” they could fronted by the fact that the top earner of the week simply say “I need to think about it” and hang up was almost always an internet salesman. the phone without repercussion. In short, internet salesmen operated without the physical and social frameworks that grounded the fl oor script and that DISCUSSION enabled fl oor salesmen to play their traditional role. The foregoing analysis of fl oor and internet sales Without a stage, props, or supporting actors and as scripted encounters helps us see with greater facing a new set of constraints and affordances (dis- precision how and why the internet transformed tanced interaction, a phone, email, and electronic the work system in which cars are sold. The inter- databases) internet salesmen had little choice but to net did not directly change the salesmen’s work play their role following a different script. Given practices in the way CT scanners changed how ra- the ease with which the customer could exit an en- diologists and radiological technologists produced counter and the impossibility of “landing a customer medical images of ruptured disks or kidney stones. on a car,” settling on a price early in the encounter In fact, neither the salesmen nor their customers became necessary, lest the salesman lose the cus- used the internet during their encounters. If any- tomer to a competitor. What the internet salesmen thing, it was the telephone not the internet that had available for support were databases that pro- became central to the internet salesman’s situated vided information on the features of each vehicle in practice. Instead, the internet shifted the ground on inventory, what each vehicle cost the dealer, the cost which salesmen and customers met to sell and buy of fi nancing the vehicle, and other dealers’ invento- vehicles. Specifi cally, the internet unanchored the ries. Customers also came to the interaction armed sales encounter from its historical mooring in face- with information, including fairly accurate data on to-face interaction, thereby demanding a reconfi gura- the dealer’s costs. Because both parties were con- tion of scripted interaction during a sales encoun ter. fi ned to constructing the deal entirely around infor- The reconfi guration changed the defi nition of the mation and because both parties had reasonably situation in ways that required salesmen and al- accurate data, internet salesmen could no longer lowed customers to play their roles differently. defi ne the sales encounter as a negotiation, so they Eliminating initial face-to-face interaction meant approached it as a matter of offering a reasonable, that salesmen and customers no longer shared a competitive price. The salesmen assumed the stance common stage. Consequently, internet salesmen of a price giver and played their role in a way that could make no use of the props around which the encouraged customers to become price takers. fl oor salesmen built their performances. Compared The stance the salesmen assumed and the script to the lots and showrooms on which fl oor salesman they enacted during the phone encounter carried performed, the internet salesman’s stage was invis- over into subsequent face-to-face encounters with ible to the customer and impoverished: a cubicle customers. With price settled, salesmen had no with a desk, a phone, and a computer that allowed need to establish rapport. They no longer needed access to databases and email. Over the phone, in- to pressure customers, because customers usually ternet salesmen could not turn on sound systems, arrived on the lot committed to buying as long as 52 Academy of Management Discoveries March the price remained constant, they qualifi ed for fi - Although Goffman was not a student of techno- nancing, and the vehicle performed as expected. As logical change, the preceding analysis has illus- a result, internet salesmen had no need for the trated how the dramaturgical scaffolding that armory of moves that fl oor salesmen have devised Goffman pioneered can be repurposed for studying over the years to persuade customers to buy. With technology and organizing in ways that enable the price set, customers no longer arrived at the researchers to skirt problems that have hounded showroom anticipating the need to negotiate; of previous research. First, dramaturgy’s units of course, some tried, only to fi nd the salesman un- analysis—encounters—fi x our attention fi rmly on willing. Because the footing of the face-to-face recurrent, meso-level social patterns, or interaction encounter had shifted to one of price giving and orders, that emerge from and guide situated action. taking, interactions unfolded more smoothly and Dramaturgy, therefore, shares with structuration less contentiously than in fl oor encounters, with theory an appreciation of the duality of structure fewer awkward moments that threatened the suc- and action, but unlike structuration theory it em- cess of the encounter. In short, the internet shifted ploys concepts that are easier to map onto a so- the grounds on which cars were sold and this, in cial setting: roles, scripts, moves, stages, props, turn, altered the constraints and opportunities wo- supporting actors, frames, and footing. Second, in ven into the encounter’s dramaturgical context. lieu of searching for causes, be they material or so- cial, Goffman encourages analysts to identify con- fi gurations of dramaturgical elements to characterize Implications for Studying Technological Change both stability and change in an interaction order. This paper began by drawing attention to prob- The image is of one of interactions channeled by lems that scholars have faced in attempting to un- constraints and opportunities, rather than actions derstand when, how and why technologies alter driven by causes. Finally, because dramaturgy’s work systems. Historically students of technology primitives encompass both the material (props, and organizing have vacillated between operating stages) and the social (moves, scripts, footing, sup- at too high or too low a level of analysis. When the porting actors) and because the elements are con- level of analysis is too high, technologies tend to crete and observable, the approach enables students disappear, as they did with contingency theory. When of technology to embrace the entwining of the ma- the level of analysis is too low, researchers risk los- terial and the social without privileging either and ing sight of work systems as often occurs in con- without becoming lost in ontological abstraction. A structionist studies of a technology’s use (Leonardi prop is no more nor less important than a support- & Barley, 2010). Students of technology also regu- ing actor. Accordingly, it is not unreasonable to larly fall victim to telling stories that are neatly claim that dramaturgical analysis enables research- causal even when they try to avoid technological ers to shift their attention off of (but not lose track determinism. In the latter case, social mechanisms of) technologies and onto relationships to docu- often replace material mechanisms (Orlikowski, ment a socio-technical or work system. Ultimately, 2007). As Orlikowski has argued, avoiding just-so for students of technology and organizing (in con- stories that privilege the material over the social, or trast to students of technology per se), changes in vice versa, requires grappling with how the social interactional patterns and role relations are what and the material are “constitutively entangled.” But matters. such grappling is no easy task. It is one thing to To take advantage of what a Goffmanesque analy- propose that the material and the social are onto- sis offers, researchers may need to alter their ap- logically mutually constitutive; it is another to de- proach to studying technological change. First and velop methods for systematically investigating that foremost, a dramaturgical analysis requires behav- entanglement. This is precisely what the socio- ioral data if the analyst is to specify the scripts that technical systems theorists set out to do (Emery & lie at the core of an interaction order. It is telling Marek, 1962; Rice, 1963; Trist & Bamforth, 1951) that in his last published essay, Goffman (1983) over half a century ago. One can argue that they ul- highlighted the work of Roger Barker (1963), the timately failed because they sought to retain a dual- now nearly forgotten social psychologist who made istic distinction between the technical and the a career of meticulously documenting the minute- social while looking for optimal ways to match the by-minute fl ow of people’s behavior to characterize two (Orlikowski, 1992). More likely their efforts the structure of what he called “behavioral settings.” waned because they became committed to spread- The study of behavior, particularly repetitive be- ing autonomous work groups and, thus, never de- havior, is rare today, even among ethnographers veloped a systematic approach for unraveling how who now focus more on interpretations and mean- the social and the material are entwined. ings than on behavior. This is not to say that the 2015 Barley 53 study of behavior is more important than the study relations, it casts its net beyond the technology and of meaning, but only that one cannot reliably map those who use it to include those with whom users scripts or identify moves without documenting pat- interact regardless of whether they also use the tech- terned behaviors in situ. It is worth noting that nology. Rather than ask how a technology alters when Goffman (1981) wrote of “forms of talk,” he work practices, the dramaturgist asks a more holis- showed less interest in what was said or meant tic question: has the technology shaped role rela- than in how the saying was done and how it was tions within the work system in which it resides. situated such that we know that we are hearing a Two caveats are important. First, not all technol- lecture rather than sermon, or a radio announcer ogies reground encounters and the interactions that rather than an interviewer. comprise them. Those that do not may change the The second implication for research on technol- procedures people employ to accomplish their ogy and organizing fl ows from the fi rst: the data for work, but are unlikely to change scripts and roles. a dramaturgical analysis are best gathered through One need only consider the work of professors and observation rather than interviews. As ethnogra- administrative assistants to see that regrounding phers have long understood, participants in a social does not always occur even when technological scene are unlikely to be aware of the scripts they change is signifi cant. As recently as the 1980’s ad- follow or the moves they make. Participants’ ac- ministrative assistants answered phones, interacted counts of their behavior often differ substantially with students, kept records of accounts, fi led doc- from what they do (Bernard, Killworth, Kronenfeld, uments, and typed letters, memos, and manuscripts & Sailor,1985; Jerolmack & Khan, forthcoming). for faculty (who often wrote the fi rst drafts by Moreover, identifying scripts and moves requires hand). Today, administrative assistants continue to observations of multiple instances of a type of en- answer phones and interact with students, but few counter. Otherwise, it is impossible to separate re- type documents for faculty. Professors now use current commonalities that span instances from word processing programs to create and revise their idiosyncratic variation. In fact, whereas most social own documents. As in the past, administrative as- scientists seek variation in order to partial it, dram- sistants continue to manage accounts, but they do aturgical analysts seek patterns beneath the vari- so using computerized spreadsheets and forms. ance. In linguistic terms, one might say they are As a consequence, fi ling no longer requires rooms more interested in grammar than in speech. fi lled with fi ling cabinets. Because of the increased This is not to say, however, that dramaturgical effi ciency of producing and storing documents and analysts are uninterested in variation, but rather because faculty have assumed the tasks of produc- that variation surfaces primarily by comparing dif- ing documents and answering their own phones, ferent types of encounters or interaction orders. universities now employ fewer administrative as- Thus, the third implication of using a dramaturgi- sistants and those who remain have acquired new cal approach is that research must be explicitly skills and tasks, such as maintaining websites. Yet, comparative. Ultimately, the student of technology administrative assistants continue to have roughly and organizing is interested in the shape of the in- the same relations with faculty as they had in the teraction orders that arise when technologies are or past. In short, what administrative assistants do are not a part of the fabric of encounters. One can and how they do it has changed considerably, but use dramaturgical concepts to deconstruct any situ- role relations and the interaction order have not. ation in which a technology is employed, but one A second caveat is perhaps more troubling for cannot identify differences in scripts, moves, props, students of technology and work. Although techno- stages, supporting actors, and footing without logical change may alter scripts and channel an in- behavior data drawn from relevant comparisons. teraction order in new directions, the resulting Thus, to employ a dramaturgical approach, stu- work system need not be tied to the technology’s dents of technology must either conduct before and presence. During the course of the study we docu- after studies, or study settings in which technolo- mented three fl oor sales involving internet sales- gies are and are not used. Studies of a single tech- men, two at the Chevrolet dealership and one at the nology’s use in a single situation are inadequate. Toyota dealership. These salesmen told their cus- Finally, and perhaps most importantly, to exe- tomers the invoice price of the car and the markup cute a dramaturgical analysis students of technol- at which they would sell the car at the start of each ogy and organizing cannot center their attention on encounter. In other words, they launched the inter- either the technology or its users. Most current re- net script on the fl oor and the subsequent interac- search on technology’s social implications focuses tion then unfolded as a face-to-face version of an on how people use a technology. Because drama- internet encounter. All three encounters were de- turgy highlights roles, scripts, interactions, and role void of moves by the salesman to create pressure; 54 Academy of Management Discoveries March all three were devoid of moves by the customer to members of distributed teams to interact face-to- gain advantage over the salesman; and none evinced face from time to time to facilitate coordination awkward moments. In fact, one was among the fi ve (Hinds & Cramton, 2012). Evidence from automo- successful fl oor sales we observed. Conversely, we bile sales challenges the inherent superiority of also witnessed an encounter in which the customer face-to-face encounters by extending the research attempted to initiate an internet script with a fl oor lens to interactions in which the parties have op- salesman at the Toyota dealership. The customer posing, rather than ostensibly shared, goals. One began by asking the salesman to sit down and re- might speculate that the relative effectiveness of in- view the options on the models in which he was teracting from a distance via computer-mediated interested. He also attempted to persuade the sales- technologies may largely depend on whether par- man to share the invoice price. The salesman re- ticipants deem the face-to-face version of an inter- acted to the customer’s request with disbelief and action to be culturally and interpersonally noxious. anger. From that point on, the interaction became This study shows that distance can mollify interac- increasingly contentious and ultimately degener- tions marked by stress, wariness, and distrust. At ated to the point where the salesman terminated the very least, our fi ndings suggest the utility of ex- the encounter, leaving the customer standing alone amining a broader range of interactions and en- in the lot while the salesman walked angrily back counters that can be conducted either face-to-face to the showroom. or through distancing technologies to move toward These four encounters underscore several a more contingent theory of computer-mediated points. First, an interaction order may be enacted communication. even in the absence of the technology that ini- tially occasioned its emergence. Second, it may be easier for the situationally advantaged and I am indebted to Kathy Lee, Christopher Han, Elijah more powerful party, in this case the salesman, in Rabek, Alison Wong, and especially Rosanne Sino whose favor the situation is stacked to modify or who conducted fi eldwork for this project. I am transform an interaction ritual. Third, and more grateful for my colleagues Diane Bailey, William practically, organizations might do well to ask Barley, Paul Leonardi, Woody Powell, Daisy themselves when it makes sense to transfer an in- Chung, and the members of the Networks and teraction order from the setting where it emerged Organizations workshop and the Work Technology to settings where it is not the dominant line of and Organizations Lab Meetings at Stanford who action. The data make clear that the internet offered many useful comments on earlier and much script could work just as easily on the fl oor as in muddier drafts of this paper. The research was the backrooms of the dealership. Moreover, inter- conducted under the auspices of and with funding net salesmen typically sold more cars than did from the Stanford/General Motors Collaborative fl oor salesmen. What most likely kept the dealers Research Laboratory. from recognizing that internet scripts could be profi tably transferred to the fl oor was that doing so required abolishing a long standing institution REFERENCES which would have violated both the fl oor sales- men’s and the owners’ notions of “the way things Barker, R. G. 1963. The stream of behavior. New York: are done around here.” Appleton-Century-Crofts. In closing, it is worth noting that the fi ndings of Barley, S. R. 1990. The alignment of technology and this study pose a corrective to our knowledge about structure through roles and networks. Administra- distributed or virtual coordination, currently the tive Science Quarterly, 35: 61–103. largest body of research on how the internet is changing work. Most studies of how people use Barley, S. R. 1986. Technology as an occasion for struc- computer-mediated technologies to interact from a turing: evidence from observations of CT scanners and the social order of radiology departments. Ad- distance have examined work groups whose mem- ministrative Science Quarterly, 31: 78–108. bers need to collaborate on a common task (Cram- ton & Hinds, 2005; Hinds & Bailey, 2003; Hinds & Barley, S. R. 2004. Puddle jumping as a career strategy. Kiesler, 2002). The majority of these studies indi- In R. Stablein & P. Frost (Eds.), Renewing research cate that distributed interaction poses signifi cant practice: Lessons from scholar’s journeys: 67–82. challenges for coordination, and suggests that all Stanford, CA: Stanford University Press. else being equal, face-to-face encounters are supe- Barley, S. R. Meyerson, D. M., & Grodal, S. 2011. Email rior to virtual encounters. Indeed, students of dis- as a source and symbol of stress. Organization Sci- tributed work now advise employers to allow ence, 22: 262–285. 2015 Barley 55

Becker, H. S., & Strauss, A. L. 1956. Careers, personality, Eriksson, O., & Goldkuhl, G. 2013. Preconditions for pub- and adult socialization. American Journal of Soci- lic sector e-infrastructure development. Information ology, 62: 253–263. and Organization, 23: 149–176.

Bernard, H. R., Killworth, P., Kronenfeld, D., & Sailor, L. Evans, J. A. 2008. Electronic publication and the nar- 1985. The problem of informant accuracy: The validity rowing of science and scholarship. Science, 321: of retrospective data. Annual Review of Anthropology, 395–399. 13: 495–517. Fischer, C. S. 1992. America calling: A social history of Boczkowski, P. J., & Orlikowski, W. J. 2004. Organiza- the telephone to 1940. Berkeley, CA: University of tional discourse and new media: A practice perspec- California Press. tive. In D. Grant, C. Hardy, C. Oswick, & L. L. Putnam (Eds.), The handbook of organizational discourse: Fulk, J., Steinfi eld, C. W., & Schmitz, J. 1987. A social in- 359–377. Thousand Oaks, CA: Sage. formation processing model of media use in organi- zations. Communication Research, 14: 529–552. Browne, J. 1973. The used-car game: A sociology of the bargain. Lexington, MA: Lexington Books. Garfi nkel, H. 1967. Studies in ethnomethodology. Engle- wood Cliffs, NJ: Prentice Hall. Buhalis, D., & Zoge, M. 2007. The strategic impact of the internet on the tourism industry. In Geertz, C. 1973. Ideology as a cultural system. In Geertz, M. Zigala, L. Mich, & J. Murphy (Eds.), Information C. (Ed.), The interpretation of cultures: 193–233. and communication technologies in tourism: 481– New York: Basic. 492. Heidelberg, Germany: Springer. Gerwin, D. 1979. The comparative analysis of structure Castells, M. 1996. The rise of the network society. Ox- and technology: A critical appraisal. Academy of ford, UK: Blackwell. Management Review, 4: 41–51.

Collins, R. 1988. Theoretical continuities in Goffman’s Ghose, A., Mukhopadhyay, T., & Rajan, U. 2007. The im- work. In P. Drew and A. Wootton (Eds.). Erving Goff- pact of internet referral services on a supply chain. man: Exploring the interaction order: 41–63. Information Systems Research, 18: 300–319. Cambridge, UK: Polity. Giddens, A. 1987. Social theory and modern sociology. Collins, R. 2004. Interaction ritual chains. Princeton, NJ: Stanford, CA: Stanford University Press. Princeton University Press. Goffman, E. 1959. The presentation of self in everyday Cramton, D., & Hinds, P. J. 2005. Subgroup dynamics in life. Garden City, NY: Doubleday. internationally distributed teams: ethnocentrism or cross-national learning. Research in Organizational Goffman, E. 1961a. Encounters. Indianapolis, IN: Bobbs- Behavior, 26: 231–263. Merrill.

Dabbish, L. A., & Kraut, R. E. 2006. Email overload at Goffman, E. 1961b. The moral career of the mental pa- work: An analysis of factors associated with email tient. In E. Goffman (Ed), Asylums: 125–170. New strain. In Proceedings of the ACM Conference on York: Anchor. Computer Supported Cooperative Work: 431–440. Goffman, E. 1967. Interaction ritual: Essays on face-to- New York: ACM Press. face behavior. Garden City, NY: Doubleday. Demaerschalk, B. M., Vargas, J. E., Channer, D. D., Noble, Goffman, E. 1974. Frame analysis. New York: Harper & B. N., Kiernan, T. J., Gleason, E. A., Vargas, B. B., Row. Ingall, T. J., Aguilar, M. I., Dodick, D. W., & Bobrow, B. J. 2012. Smartphone teleradiology application Goffman, E. 1981. Forms of talk. Philadelphia: Univer- is successfully incorporated into a telestroke net- sity of Pennsylvania. work environment. Stroke, 43: 3098–3101. Goffman, E. 1983. The interaction order. American So- Dougherty, J. W. D. (Ed.). 1985. Directions in cogni- ciological Review, 48: 1–17. tive anthropology. Urbana, IL: University of Illinois Press. Hage, P., & Harary, F. 1991. Exchange in Oceania: A graph theoretic analysis. Oxford, UK: Oxford Uni- Edmondson, A. C., Bohmer, R. M., & Pisano, G. P. 2001. versity Press. Disrupted routines: Team learning and new tech- nology implementation in hospitals. Administra- Harvey, E. 1968. Technology and the structure of organiza- tive Science Quarterly, 46: 685–716. tions. American Sociological Review, 33: 247–259.

Emery, F. E., & Marek, J. 1962. Some socio-technical Hinds, P. J., & Bailey, D. E. 2003. Out of sight, out of sync: aspects of automation. Human Relations, 15: 17– Understanding confl ict on distributed teams. Organ- 25. ization Science, 14: 615–632. 56 Academy of Management Discoveries March

Hinds, P. J., & Cramton, C. D. (2012). Collaboration in Morton, F. S., Zettelmeyer, R., & Silva-Risso, J. 2001. In- distributed work: Transforming distant into close ternet car retailing. Journal of Industrial Econom- relationships through site visits. Center for Work, ics, 49: 501–519. Technology and Organization, Stanford, CA: Stan- Morton, F. S. 2003. Consumer information and discrimi- ford University. nation: Does the internet affect the pricing of new Hinds, P. J., & Kiesler, S. 2002. Distributed work. Cam- cars to women and minorities. Quantitative Mar- bridge, MA: MIT Press. keting and Economics, 1: 65–92.

Hughes, E. C. 1937. Institutional offi ce and the person. Nadel, S. F. 1957. The theory of social structure. Glen- American Journal of Sociology, 43: 404–413. coe, IL: Free Press.

J. D. Power and Associates. 2007. 2007 new autoshop- Nakao, K., & Treas, J. 1994. Updating occupational pres- per.com study. Westlake Village, CA: McGraw-Hill. tige and socioeconomic scores: How the new measures measure up. Sociological Methodology, 24: 1–72. Jacobs, S. 2001. Confessions of a car salesman. Accessed National Research Council. 1999. Collaboratories: Im- Aug 2013. http://www.edmunds.com/car-buying/ proving research capabilities in chemical and bio- confessions-of-a-car-salesman-pg2.html. medical sciences. Washington, DC: National Acad- Jerolmack, C., & Khan, S. (Forthcoming). Talk is cheap: emies Press. Ethnography and the attitudinal fallacy. Sociologi- O’Leary, M. B., & Mortensen, M. 2010. Go (con)fi gure: cal Methods & Research. Subgroups, imbalance, and isolates in geographically Kühnle, H. (Ed.). 2010. Distributed manufacturing: dispersed teams. Organization Science, 21: 115–131. Paradigm, concepts, solutions and examples. Lon- Orlikowski, W. J. 1992. The duality of technology: Re- don: Springer. thinking the concept of technology in organizations. Kuruzovich, J., Viswanathan, S., Agarwal, R., Gosain, S., Organization Science, 3: 398–427. & Weitzman, S. 2008. Marketspace or marketplace? Orlikowski, W. J. 1996. Improvising organizational trans- Online information search and channel outcomes in formation over time: A situated change perspective. auto retailing. Information Systems Research, 19: Information Systems Research, 7: 63–92. 182–201. Orlikowski, W. J. 2007. Sociomaterial practices: Explor- Lang, T. C. 2000. The effect of the internet on travel consum- ing technology at work. Organization Studies, 28: er purchasing behaviour and implications for travel 1435–1448. agencies. Journal of Vacation Marketing, 6: 368–385. Orlikowski, W. J. & Robey, D. 1991. Information technol- Latour, B. 2007. Reassembling the social: An introduc- ogy and the structuring of organizations. Informa- tion to actor-network-theory. Oxford, UK: Oxford tion Systems Research, 2: 143–169. University Press. Orlikowski, W. J. & Scott, S. V. 2008. Sociomateriality: Lawson, H. M. 2000. Ladies on the lot: Women, car Challenging the separation of technology, work and sales, and the pursuit of the American dream. Lan- organization. Academy of Management Annals, 2: ham, MD: Rowman & Littlefi eld. 433–474.

Leonardi, P. M. 2013. Theoretical foundations for the Pentland, B. T. 1992. Organizing moves in soft- study of sociomateriality. Information and Organi- ware support hotlines. Administrative Science zation, 23: 59–76. Quarterly, 37: 527–548.

Leonardi, P. M. & Barley, S. R. 2010. What’s under con- Perrow, C. 1967. A framework for the comparative analy- struction here? Social action, materiality, and power sis of organizations. American Sociological Review, in constructivist studies of technology and organiz- 32: 194–208. ing. Academy of Management Annals, 4: 1–51. Pinch, T. 2010. The invisible technologies of Goffman’s MacDuffi e, J. P. 1995. Human resource bundles and sociology from the merry-go-round to the internet. Technology and Culture, 51: 409–424. manufacturing performance: Organizational logic and fl exible production systems in the world auto Pike, K. L. 1967. Language in relation to the unifi ed the- industry. Industrial & Labor Relations Review, 48: ory of the structure of human behavior. The Hague: 197–221. Mouton.

Messinger, S. L., Sampson, H. S., & Towne, R. D. 1962. Poole, M. S., & DeSanctis, G. 1990. Understand- Life as theater: Some notes on the dramaturgic ap- ing the use of group decision support systems: proach to social reality. Sociometry, 25: 98–110. The theory of adaptive structuration. In J. Fulk & 2015 Barley 57

C. Steinfi eld (Eds.), Organizations and communi- Turner, R. E., & Edgley, C. 1976. Death as theater: A cation technology: 173–93, Newbury Park, CA: Sage. dramaturgical analysis of the American funeral. So- ciology and Social Research, 60: 377–392. Ribes, D., Jackson, S., Geiger, S., Burton, M., & Thomas, F. 2013. Artifacts that organize: delegation in the dis- Van Hoose, David D. 2011. E-commerce economics. Mil- tributed organization. Information and Organiza- ton Park, UK: Routledge. tion, 23: 1–14. Viswanathan, S., Kuruzovich, J., Gosain, S., & Rice, A. K. 1963. Productivity and social organization: Agarwal, R. 2007. Online infomediaries and price The Ahmedebad experiment. London: Tavistock. discrimination: Evidence from the automotive retail- ing sector. Journal of Marketing, 71: 89–107. Ross, P., Sepper, R., & Pohjonen, H. 2010. Cross-border teleradiology-experience from two international Wellman, B., & Haythornthwaite, C. (Eds.). 2002. The teleradiology projects. European Journal of Radi- internet in everyday life. Oxford, UK: Blackwell. ology, 73: 20–25. Wilshire, B. 1982. The dramaturgical model of behavior: Schank, R. C., & Abelson, R. P. 1977. Scripts, plans, Its strengths and weaknesses. Symbolic Interaction, goals, and understanding. Hillsdale, NJ: Lawrence 5: 287–298. Erlbaum. Wise, G. L. 1974. Differential pricing and treatment Schultze, U., & Orlikowski, W. J. 2004. A practice by new-car salesmen: The effect of the prospect’s race, perspective on technology-mediated network re- sex, and dress. Journal of Business, 47: 218–230. lations: the use of internet-based self-serve tech- Womack, J. P., Jones, D. R., & Roos, D. 1990. The machine nologies. Information Systems Research, 15: that changed the world: The story of lean produc- 87–106. tion. Cambridge, MA: MIT Press.

Smite, D., Moe, N. B., & Agerfalk, P. J. (Eds.). 2010. Woodward, J. 1958. Management and technology. Agility across time and space: Implementing ag- London: Her Majesty’s Stationary Offi ce. ile methods in global software projects. Berlin: Springer. Zettelmeyer, F., Morton, F. S., & Silva-Risso, J. 2006. How the internet lowers prices: Evidence from matched Sproull, L., & Kiesler, S. 1991. Connections: New ways survey and automobile transaction data. JMR, Jour- of working in the networked organization. Cam- nal of Marketing Research, 43: 168–181. bridge, MA: MIT Press.

Trist, E. L., & Bamforth, K. W. 1951. Some social psycho- logical consequences of the longwall method of coal Stephen R. Barley ([email protected]) is the getting. Human Relations, 4: 3–38. Richard Weiland Professor of Management Science Turban, E., Lee, J. K., King, D., Peng, L. T., & Turban, and Engineering and the Co-Director of the Center D. 2009. Electronic commerce 2010. Upper Saddle for Work, Technology & Organization at Stanford’s River, NJ: Prentice Hall. School of Engineering. He received his PhD from MIT in Organization Studies. His research interests Turner, R. H. 2006. Role theory. In J. H. Turner (Ed.), include work and occupations, technological change, Handbook of sociological theory: 233–254. New and corporate power in representative democracies. York: Springer. Address: Department of Management Science and Engineering, 208 Huang Engineering Center, 475 Via Turner, R. H. 1968. Role: sociological aspects. In Ortega, Stanford University, Stanford, CA, 94305. D. L. Sills (Ed.), The international encyclopedia of the social sciences: 552–557. New York: Macmillian. 58 Academy of Management Discoveries March

APPENDIX A Defi nitions and Examples of Sales Moves, Customer Moves, and Award Interactions

Move/Interaction Defi nition Example

Sales Moves

All dealers tell you the same thing Salesman tells customer that no “You know there’s no dealer that would sell the car, an LE other car dealer will give them a model, for that. So, help us to help you. You’ve been better price. here. You spent time here. I want to step up to the plate, just to make this transaction.” Want to buy or not? Salesman asks customer if they The salesman comes back after a minute and says in a are going to buy the car or not. soft tone, “We can do $11,600 plus tax and license.” (Usually after negotiating for The customer puts both hands on his lap, his shoulders some time.) tense, and breathes. He then says, “Let me get your business card. I’ll need to think about it and.” The salesman interjects, “If I do $11,500, will you buy it?” The customer smiles, “That helps.” “Then will you buy it?” asks the salesman. The customer leans back and replies, “I think very likely. 90% yes.” Manager’s not going to like that Salesman tells customer that “If I go to my manager and ask them for $20,000 out the his manager is not going to like door, he’s basically gonna chew my head off for coming the offer the customer has made. back because I know I can’t give it. Basically what I got to do is I got to go to that guy. See how big that guy is.” Need to make profi t/ commission Salesman tells the customer The customer again asks if there are any graduate or that he can’t go as low as the repeat buyer discounts. After being told no, he says, customer wants because there’d “$21,431? So that’s it. Nothing below that?” The be no profi t or commission. salesman replies, “I can’t do it. I don’t want to lose.” Create urgency Salesman states that if the The salesman says that he doesn’t want to put any customer doesn’t buy, the price pressure on them to buy but that the HG (the package will go up or the vehicle may with the smart key) only has the discounted price be sold to someone else. through the end of the day and that the price is only $300 above dealer cost. Deprecate alternatives Salesman deprecates another car The salesman says that the Rav4 is cheaper than the the customer is considering. CRV because the CRV comes with “a lot of extra crap you don’t need.” He also says that the CRV is based off the Civic, which is not good. He explains that Honda had to recall the Civic because of safety issues and that the Rav4 is less likely to roll over in the case of an accident. Ignore customer comment Salesman ignores a customer’s After telling Lawrence that he didn’t have a trade-in, statement or question. the brother mentions that he wants to see the price on the Ford F150 with the supercharger because Ford was going to give them an estimate. The salesman seems to ignore this statement and continues collecting personal information for a form he’s fi lling out. Double team A second salesman joins the fi rst Somewhat abruptly Anne Marie peeps her head into and together they attempt to sell the offi ce on the other side of the cubicle, and asks the the customer. salesmen there, “Would you guys help me and split a deal?” Bob agrees. Anne Marie turns to the customers and says, “We work together and Bob is good.” Anne Marie explains how the salesmen are like family. Turn the customer The fi rst salesman turns the The salesman walks to the sales manager’s offi ce. The customer over to another customer waits in the cubicle, looking at the papers the salesman usually to “turn up salesman has left with him. Ten minutes later a man the heat.” in his late 50’s named “Todd” (according to his badge) comes in and shakes the customer’s hand, “Hello, I’m Todd. I’m the fl oor manager. I slash the prices to make a deal. All right?” 2015 Barley 59

Neutralize internet data Salesman provides reasons for The customer looks like he’s thinking and then says, “I got why information that the a quote from Cars.com, and I think it’s $100 less.” The customer has seen on the web is salesman doesn’t fl inch and says, “Well, Cars.com don’t probably inaccurate. sell cars. They go through dealers like us. The price they give you is their suggested price.”

Customer Moves

Disparage dealer or maker Customer complains or makes a (Talking about fi nancing) “That’s not what I saw on snide comment about auto TV—it said 1.99,” the customer says. “That’s for the 07s dealers or manufacturers. not the 08,” the salesman replies. The customer retorts, “That’s why it’s not easy to buy a car—it looks easy, but it’s not.” He is looking at the Yaris as he says this. The salesman says, “For some people it is.” Ignore salesman’s comment Customer dismisses or ignores a The customer is listening expressionless as the salesman’s statement. salesman continues to talk. He just nods occasionally, but his eyes are wondering off to the trucks. Negative comment about car Customer makes a negative The salesman fi nds a Tahoe in the color the customer comment about the car the requested. The husband seems suspicious. He looks salesman is showing. at it inquisitively and walks around the car to see it fully. “It changes on you.” It is an iridescent color that does seem to change as one moves around it. “It’s like a pimp car,” the wife says. Obstinate about price Customer insists on a particular The customer wrote a number on a piece of paper and price and won’t bend. taps the paper, “We are at the max that I can go. This is my max. I won’t be willing to go beyond that. I don’t want to go beyond that.” Threaten to go to another dealer Customers tells the salesman The customer says to the salesman, “OK. So if you that they will go to another don’t mind if I shop around and see what kind of deal dealer for a better price or I can get from (another dealer).” The salesman replies, treatment. “You can shop anywhere that you’d like to shop.” Use internet to check truthfulness Customer uses internet data to The customer asks, “Do other dealerships have the assess the truthfulness of a hybrid?” The salesman says, “I don’t think so.” The salesman’s claims. customer then says, “I went online today and it said you had, like, seven.” The salesman shakes his head and says he doesn’t know why it said that. Use other quote to bargain Customer uses a quote from Customer: “Before coming here, we visited Chevy on another dealer or the internet El Camino. They were offering something like $4,500 to get salesman to lower off MSRP.” Salesman: “We can make a good deal. his price. I want to sell you a truck today.” Customer: “Well, it ain’t gonna happen unless you make me an offer I can’t refuse.” Deal is unacceptable Customer tells the salesman The husband tells the salesman: “ So you say, if I pay that his offer is simply cash you can give me $500 rebate?” His wife interjects: unreasonable and unacceptable. “It’s nothing!” The husband says, “It’s not worth it for me. “

Awkward Interactions

Customer confrontational Customer confronts a salesman’s Three Hispanic men are talking in low voices in Spanish behavior or statement, as they look at Corvettes. The salesman is hovering over communicating anger. them, one turns and says “Man, we’re just looking at cars today—it’s not like we’re gonna take one.” Customer contradicts salesman Customer disputes a salesman’s The salesman explains that there’s a dual air conditioning statement. setting, one for the driver and another for the passenger. The husband looks at it and replies that it’s for inside/outside air, not a dual setting. The salesman doesn’t say anything. Customer hangs up the phone Customer hangs up the phone (a phone encounter) “Hi, is this Heath?” (pause) “This is abruptly in the middle of a Dan from California Toyota.” (pause) “OK. Where did phone conversation. you end up buying it?” (customer hangs up on Dan.) 60 Academy of Management Discoveries March

Customer implies salesman lies Customer suggests that the Many minutes into a long and tense negotiation the salesman is lying. customer suddenly tells the salesman, “I think it’s used.” It is not clear why customer now believes the car is used. The salesman, indignantly, “It’s not a used car, it’s a brand new car!” Frame break Customers attempt to get outside After looking at the cars for a short time the customer of the customer/salesman says, “I know you’re supposed to get me in the car,” and relationship by stating, “I know the saleswoman responds, “All I’ve got to do is fi nd you what’s going on here.” the right car.” She adds “I’m not going to play games with you,” at which point both she and the client smile and laugh. Interrupting Either the customer or the The salesman says something about an ’07 and the salesman interrupts the other customer interrupts, raising his right hand in front of while the other is talking. him, “Don’t start on the ’07 discussion—I don’t want an ’07—you are not the fi rst one to push that so that’s why I am ticked with you.” Rapport building fails A salesman’s attempt to establish The salesman notices the customer’s iPhone strapped rapport backfi res immediately. to his belt and asks if he likes it. The customer says he does but the salesman tells him that he had one for a while but didn’t like the software. He adds he returned the iPhone and got a Palm instead. The customer says that he “has to like the iPhone” and the salesman realizes that the customer works for Apple. Salesman is rude or belittling The salesman treats the (during a test drive) The salesman asks the young customer rudely, with sarcasm woman if she wants to go on the freeway but she says or disrespect. she doesn’t. (She has been driving really cautiously this whole time.) The salesman then jokes that she’s been holding her breath while driving and the woman replies, “That’s how I drive!” Salesman contradicts customer Salesman disputes a customer’s The customer asks if the salesman has any Tahoe statement. hybrids. The salesman laughs and says that “this is better than hybrid” (referring to the fl exfi ll engine). He starts talking about the improved gas mileage you can get with ethanol but the customer mentions that some hybrids get the carpool benefi t. The salesman laughs again and says that they “don’t have the carpool thing anymore.” Salesman implies customer lies Salesman suggests that the To a customer, “I have no problem if you say you’ll customer is lying. be back tomorrow, but you know, I’ve seen a lot of customers. They tell you tomorrow. They lie to you. They never come back.” Silence Long runs of silence in situations, The customer seems worried (about driving a Prius) coupled with an indication that and asks if everything besides putting the car in drive the silence is uncomfortable. is normal. The salesman tells her yes, and that you actually get taught how to drive a hybrid in a class the dealer offers. This is followed by a long period of silence. [The salesman told me later he was having an off-day and didn’t feel like helping anyone.]