Andrea Godfrey, Kathleen Seiders, & Glenn B. Voss  Enough Is Enough! The Fine Line in   Executing Multichannel Relational ...
has a negative influence on repurchase because customers          We use these data to estimate a system of simultaneouspe...
munication volume in relational exchanges, in which theory         and customer response beyond a certain threshold (Anand...
TABLE 1                                                                                   Recent Empirical Studies of Rela...
both channels, leading to increased repurchase. A customer’s      communication volume and repurchase, but results reporte...
In effect, both theoretical perspectives imply that align-                    For each customer, we matched survey respons...
chase visits and repurchase spending for each customer dur-       DVij  0 + 1TIME + 2TIME2 + 3iing quarters 1–12. W...
QUARTER2, 3, and 4 = dummy variables to control for                       4). For expositional brevity, we report results ...
TABLE 4                                  Simultaneous Equation Results for Repurchase Spending                            ...
positive, indicating that relational customers return regu-                                        FIGURE 1larly for routi...
FIGURE 2                                                                       FIGURE 3           Plotting Significant Vol...
communication. First, the results indicate that there is an        sitivity toward the telephone channel. Reactance is gen...
individual customer characteristics, such as channel prefer-      operationalize telephone communication as a dummyence, t...
The Fine Line in Executing Multichannel Relational Communication
The Fine Line in Executing Multichannel Relational Communication
The Fine Line in Executing Multichannel Relational Communication
The Fine Line in Executing Multichannel Relational Communication
Upcoming SlideShare
Loading in...5
×

The Fine Line in Executing Multichannel Relational Communication

1,044
-1

Published on

Published in: Technology, Business
0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total Views
1,044
On Slideshare
0
From Embeds
0
Number of Embeds
0
Actions
Shares
0
Downloads
15
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

The Fine Line in Executing Multichannel Relational Communication

  1. 1. Andrea Godfrey, Kathleen Seiders, & Glenn B. Voss Enough Is Enough! The Fine Line in Executing Multichannel Relational CommunicationIn an effort to build long-term, profitable relationships, many companies systematically engage in multichannelrelational communication—personalized messages sent to existing customers through various channels as part ofa broader relationship marketing strategy. In this research, the authors examine three key drivers of relationalcommunication effectiveness: volume of communication, mix of communication channels, and alignment of thosechannels with customers’ preferences. They hypothesize that customer response to relational communicationfollows a continuum in which reciprocity explains response to lower levels of communication, the classic ideal pointdescribes a transition phase, and reactance explains response to higher levels of communication. They empiricallytest the theoretical framework by examining the impact of multichannel communication on repurchase over a three-year period. The results indicate that after the ideal level of communication is exceeded, customers reactnegatively. This negative response can be exacerbated by the use of multiple channels but attenuated by aligningchannels with customer preferences. The findings suggest that the complex effects of multichannel communicationcan actually drive customers away from rather than closer to a company.Keywords: multichannel relational communication, ideal point, reactance, reciprocity, customer repurchase I spent two months assessing and segmenting the database itability. To encourage loyalty and build long-term relation- of past donors and creating a plan for scheduling personal meetings with 60 high-gift-capacity donors. Ten donors ships, firms must communicate with their customers in a were leery about speaking with me. I left messages for compelling way. However, what if such marketing commu- another 30 donors, but only three returned my call. Five nication drives customers away from rather than closer to donors hung up on me with varying degrees of anger. One the company? As the major gifts officer in the preceding person said that I was the third person to call in three quote discovered, even an organization’s most ardent sup- days. Thirteen donors also received a handwritten note to porters can be alienated when they feel bombarded by the inform them of our fundraising campaign. In the end, I did not book a single appointment. Because our organization communication. relies heavily on phone and mail contacts along with face- This research focuses on multichannel relational com- to-face fundraising, potential donors have already been munication, which we define as personalized communica- bombarded by mail and phone calls by the time I try to tion with existing customers through various channels as arrange a personal meeting. I am removing all of my part of a broader relationship marketing strategy. In an donors from the phone and mail contact lists to control how much communication they receive from us. effort to retain and cross-sell to existing customers, compa- nies use individual-level customer data to personalize this —Major gifts officer, national nonprofit agency communication. The communication can remind customers of needed services, announce new products and locations,M arketers champion the idea of focusing resources on customer retention strategies to avoid expensive survey customer satisfaction following a service encounter, acquisition initiatives and cultivate customer prof- and convey targeted promotional offers. Despite widespread use, multichannel relational com-Andrea Godfrey is Assistant Professor, Department of Management and munication has only recently attracted attention in the mar-Marketing, School of Business Administration, University of California, keting literature, and its effects on customer repurchase areRiverside (e-mail: andrea.godfrey@ucr.edu). Kathleen Seiders is Associ- not well understood. Although general consensus exists thatate Professor, Marketing Department, Carroll School of Management, some amount of communication is better than none, rele-Boston College (e-mail: seiders@bc.edu). Glenn B. Voss is Associate Pro- vant marketing theories and the limited empirical evidencefessor, Department of Marketing, Cox School of Business, SouthernMethodist University (gvoss@cox.smu.edu). All three authors contributed are equivocal as to whether there is an ideal level beyondequally. The authors thank their research partner for providing access to which additional communication leads to diminishing orits data, the Boston College (Carroll School of Management) Dean’s even negative returns. For example, reciprocal action theoryResearch Fund, and the University of California, Riverside, Regent’s Fac- implies that increasing relational communication positivelyulty Fellowship. They also thank Yuping Liu and Jacquelyn Thomas for influences repurchase because customers perceive greatertheir constructive comments, and the anonymous JM reviewers for their relationship investment by the firm. Conversely, reactancethoughtful and constructive suggestions. theory suggests that increasing relational communication© 2011, American Marketing Association Journal of MarketingISSN: 0022-2429 (print), 1547-7185 (electronic) 94 Vol. 75 (July 2011), 94–109
  2. 2. has a negative influence on repurchase because customers We use these data to estimate a system of simultaneousperceive the communication as invasive or obtrusive. The equations with endogenous variables. The results indicatereciprocity perspective is supported by empirical studies that there is an ideal level of communication volume thatreporting a positive linear association between relational varies across channels; the ideal point also shifts incommunication volume and repurchase behavior, while response to multichannel communication efforts and acrossreactance is supported by research reporting an inverted U- individual people, depending on their channel preferences.shaped effect between relational communication volume Because shifts in the ideal point directly influenceand repurchase behavior. repurchase behavior, our results provide substantive The challenge of understanding multichannel communi- insights that can lead to more effective customer relation-cation effects is complicated by the limited knowledge of ship strategies. The findings highlight the importance ofhow channels combine to influence customer repurchase. considering the impact of specific channels, individuallyWhen firms use different communication channels in com- and in combination, rather than aggregate volume as abination, it is a matter of speculation whether the effects are means to manage the communication channel mix moreadditive or multiplicative and, if multiplicative, whether the effectively. The results also underscore the need to avoidinteraction enhances or diminishes customer response. inefficient allocation of marketing resources by developingCommunication through multiple channels could enhance protocols that limit total communication through all chan-customer response by demonstrating greater resource nels and specify effective channel combinations.investment on the part of the firm or could alienate cus- In the sections that follow, we develop a conceptualtomers by signaling an inadequate appreciation of one-to- framework that predicts diminishing effects of multichannelone communication. Empirical studies involving the simul- relational communication on customer repurchase. Wetaneous use of multiple channels are particularly scant. extend the framework to examine interaction effects for The mixed evidence as to whether there is an ideal level combinations of channels and for customer-level channelof communication—and how that level might vary when preferences. We then present the empirical model and resultsfirms use multiple channels—undermines managerial prac- and conclude with implications for researchers and managers.tice in terms of effectively allocating marketing resources.Managers face additional challenges in aligning the com-munication channels used with customers’ preferences. Conceptual FrameworkAlthough we might assume that aligning communication Reciprocal action theory builds on social norm theory tochannels with customers’ preferences increases the utility of describe the obligation people experience to return “goodthe contacts, research has not examined the strength or for good,” in which the response is proportional to what isimportance of this moderating effect. received (Bagozzi 1995; Becker 1990). In market To explore these important and unresolved issues in cur- exchanges, firm investments in customer relationships canrent knowledge, we examine a set of interrelated research produce psychological bonding that triggers the normativequestions: Is there an ideal volume of relational communi- reciprocity of purchase (Dahl, Honea, and Manchandacation? How is customer response to multichannel commu- 2005). Customer reciprocity manifests when individualizednication influenced by the total volume of communication communication increases perceived relationship quality (Deand the mix of channels used? and To what degree do cus- Wulf, Odekerken-Schroder, and Iacobucci 2001) or feelingstomers’ channel preferences moderate the relationship of gratitude toward the firm (Palmatier et al. 2009). Reci-between relational communication volume and customer procity suggests that relational communication positivelyrepurchase? In addressing these questions, our study con- influences repurchase because customers perceive a greatertributes theoretical, empirical, and substantive insights into resource investment by the firm and enhanced informativethe complex nature of multichannel communication. or interpersonal value. We present a conceptual framework that integrates reci- Reactance theory posits that attempts to influenceprocal action theory and reactance theory perspectives to behavior generate a motivational state of reactance (Brehmdescribe the effects of multichannel relational communica- 1966). Customers resist marketing efforts that they perceivetion on customer repurchase behavior. Together, these as an attempt to manipulate them, exercise control overtheories describe a continuum of customer response to rela- their purchasing, or limit their freedom of choice (Clee andtional communication, with lower volumes eliciting reci- Wicklund 1980). As pressure from a firm’s persuasionprocity and higher volumes triggering reactance. An ideal attempts increases, the reactance response grows stronger,point along this continuum identifies the volume at which resulting in diminished influence on customer behavior or,customer repurchase shifts from positive reciprocity to at extremes, a backlash or boomerang effect (Fitzsimonsnegative reactance. After that ideal point, additional con- and Lehmann 2004; Wendlandt and Schrader 2007). Reac-tacts diminish rather than enhance customer repurchase. tance theory can explain negative response to personal sell- We offer empirical support for this framework by exam- ing (e.g., Wicklund, Slattum, and Solomon 1970), advertis-ining the impact of multichannel relational communication ing (e.g., Robertson and Rossiter 1974), direct marketingon customer repurchase over three years. The longitudinal (e.g., Morimoto and Chang 2006), and rewards programsanalysis provides a fine-grained examination of multichan- (e.g., Kivetz 2005). Prior studies have focused primarily onnel effects by combining three distinct data sets: (1) cus- reactance to message content characteristics in transactionaltomer contact records, (2) customer transaction data, and exchanges (e.g., Clee and Wicklund 1980; Fitzsimons and(3) survey data capturing customers’ channel preferences. Lehman 2004). In contrast, we examine reactance to com- Executing Multichannel Relational Communication / 95
  3. 3. munication volume in relational exchanges, in which theory and customer response beyond a certain threshold (Anandpredicts a negative customer response to communication and Sternthal 1990).perceived as invasive, obtrusive, or environmentally wasteful. The ideal point should describe customer response to The relational communication literature provides relational communication for any channel a firm uses. Inempirical evidence that both theoretical perspectives are this study, we examine communication volume throughrelevant to understanding customer response to firm con- three channels firms commonly use for targeted customertact. As Table 1 indicates, the limited empirical studies in contacts: telephone, e-mail, and postal mail. We formallythis area offer conflicting results. The reciprocity principle hypothesize the following:is supported by several studies that find positive linear H1: For each communication channel, there is an ideal level ofeffects of relational communication on outcomes such as relational communication volume, which implies ancustomer profitability, share of wallet, and relationship inverted U-shaped relationship between customer repur-duration (Reinartz, Thomas, and Kumar 2005; Rust and chase and (a) telephone contact volume, (b) e-mail contact volume, and (c) mail contact volume.Verhoef 2005; Verhoef 2003). Reactance receives support instudies that report an inverted U-shaped relationshipbetween communication and purchase frequency (Venkate- Communication Channels in Combinationsan and Kumar 2004) and customer retention (Drèze and The use of multiple communication channels in combina-Bonfrer 2008), as well as a U-shaped relationship between tion can stimulate either additive, independent effects, orcommunication and purchase timing (Kumar, Venkatesan, multiplicative interaction effects. Interaction effects implyand Reinartz 2008). Reconciling the mixed results from this that the ideal point for relational communication volume inemerging set of studies is difficult because of variations in one channel depends on the level of communication inthe operationalization of communication, the customer out- another channel. As volume in one channel changes, thecomes measured, and the research contexts. ideal point for volume in the other channel shifts to a higher or lower point. Cross-channel interactions are consistentThe Classic Ideal Point and Relational with both reciprocal action theory and reactance theory, butCommunication Volume the two theories differ as to whether the interactions areWe propose that reciprocity explains why customer repur- positive or negative.chase is likely to increase as communication volume Customers may respond positively when they receiveincreases from low to moderate levels, whereas reactance relational communication through a combination of chan-explains why repurchase is likely to decrease as communi- nels, because the use of multiple channels, rather than acation increases from moderate to high levels. To describe single channel, shows greater resource investment by thethe transition phase of the reciprocity–reactance continuum, firm. Because of their varying characteristics, differentwe draw on the classic ideal point concept, which posits channels provide distinctive advantages in communicating information to customers. For example, the telephone chan-that the perfect or utility-maximizing level of an attribute nel allows bidirectional communication and the capacity foroccurs at an intermediate level that elicits the most positive the customer to be involved in the communication (Mohrresponse (Teas 1993). Initially, as an attribute increases up and Nevin 1990), and the e-mail channel offers customersto the ideal point, utility also increases and customer the ability to view rich visual representations of the infor-response becomes more favorable (Green and Srinivasan mation being conveyed (Alba et al. 1997). Contacting cus-1978). After the ideal point is attained, further increases in tomers through multiple channels allows firms to offer cus-the attribute instead result in negative utility and less favor- tomers complementary benefits that enhance the overallable customer response (Lilien, Kotler, and Moorthy 1992). utility of the communication, signifying greater resourceApplied to multichannel relational communication, this investment. When customers attribute the enhanced utilitycontinuum suggests that customers reciprocate with greater to firm actions, they reciprocate with increased spending.repurchase as communication increases, up to the ideal point; This reciprocity-based perspective is consistent with abeyond that point, they exhibit reactance and repurchase central tenet of integrated marketing communication, whichless in response to further increases. Therefore, the ideal holds that the combined effect of using multiple mediapoint implies an inverted U-shaped relationship between channels is greater than the sum of the individual effects ofrelational communication volume and customer repurchase. each channel (Naik and Raman 2003). For example, using The ideal point conceptualization is similar to the two- mass-media channels that effectively build brand awarenessfactor paradigm for predicting customer response to mass can enhance the effects of channels used to communicatemarketing communication. This theory proposes that more detailed information (Prins and Verhoef 2007). Whileopposing factors determine customer response to repeated limited research systematically examines synergies betweenadvertising stimuli (e.g., Berlyne 1970; Rethans, Swasy, communication channels, especially at the individual cus-and Marks 1986). Initial exposures produce a positive tomer level, it is logical to expect such effects to be evenresponse due to reduced uncertainty and learning about the more pronounced for direct channels because the customersstimulus (Stang 1975), but higher levels of exposure pro- may value customized content more.duce a negative response due to tedium or reactance The theoretical implication is that communication(Sawyer 1981). The net effect of the opposing factors is a through multiple channels shifts the ideal point so that cus-diminishing relationship between the amount of advertising tomers exhibit reciprocity up to a higher volume in one or96 / Journal of Marketing, July 2011
  4. 4. TABLE 1 Recent Empirical Studies of Relational Communication Effects on Customer Repurchase Behavior Effect of Relational Relational Communication Variables Dependent Variables Communication on Theory Consistent Study Operationalization Operationalization Repurchase Behavior with Findings De Wulf, Direct mail Perceived relationship investment Positive linear effect (Europe) Reciprocity Odekerken- Three survey items on seven-point Likert scale cap- Three survey items on seven-point Likert Nonsignificant effect (United Schroder, and turing customer’s perception of firm’s efforts to keepscale capturing customer’s perception of States) Iacobucci regular customers informed through mail contacts firm’s efforts to contribute value to regular (2001) customers Drèze and Inter-e-mail time a. Customer retention a, b. Inverted U-shaped effect Ideal point Bonfrer Number of days since last e-mail contact Estimated customer retention probability (2008) b. Customer equity Calculated customer equity Kumar, a. Frequency of rich modes of communication Purchase timing a, b. U-shaped effect Ideal point Venkatesan, Number of face-to-face contacts Time (in months) between previous observed and Reinartz b. Frequency of standardized modes of communication purchase and current observed purchase (2008) Number of telephone and mail contacts Prins and Direct marketing communication Adoption timing a. Positive effect Reciprocity Verhoef a. Dummy indicating whether a telephone contact Time (in months) between service introduc- b. Negative interaction effect Reactance (2007) was sent in month t tion and customer’s adoption Direct marketing communication × mass marketing communication b. Dummy indicating whether a telephone contact was sent in month t ¥ advertising expenditures in month t Reinartz, Communication contacts f. Acquisition likelihood a, b, c. Positive linear effect ReciprocityExecuting Multichannel Relational Communication / 97 Thomas, and a. Number of face-to-face contacts Indicator variable showing whether customer d, e. Positive interaction Reciprocity Kumar (2005) b. Number of telephone contacts is acquired effect for all dependent c. Number of e-mail contacts g. Relationship duration variables Communication contact interactions Time (in days) of customer’s relationship with d. Monthly occurrence of both face-to-face and e-mail the firm contacts h. Customer profitability e. Monthly occurrence of both telephone and e-mail Total revenue—total costs contacts Rust and Ver- a. Transaction-oriented communication Change in gross profit per customer a, b. Positive linear effect Reciprocity hoef (2005) Number of direct mail promotion contacts Gross profit per customer = total number of b. Relationship-oriented communication services purchased ¥ contribution margin Number of relationship magazine contacts Venkatesan and a. Frequency of rich modes of communication Purchase frequency a, b. Inverted U-shaped effect Ideal point Kumar (2004) Number of face-to-face contacts Total number of products purchased b. Frequency of standardized modes of communication Number of telephone and mail contacts Verhoef (2003) Direct mailings Change in customer share Positive linear effect Reciprocity Number of mail contacts Customer share = number of services pur- chased from focal firm (observed)/number of services purchased from all firms (self-report)
  5. 5. both channels, leading to increased repurchase. A customer’s communication volume and repurchase, but results reportedrepurchase response to relational communication volume in in prior studies offer divergent support for the direction ofone channel is enhanced as the volume of communication the interaction. As such, we leave the question whether thein another channel increases, manifesting as a positive interactions are positive or negative to be determinedinteraction between any two channels used in combination. empirically. More formally, we hypothesize the following: Reinartz, Thomas, and Kumar (2005) provide empirical H2: Relational communication volume using more than onesupport for this perspective in a business-to-business con- channel produces shifts in the ideal point for at least onetext and report positive interactions when firms used (1) channel, which implies significant interaction effects onface-to-face and e-mail and (2) telephone and e-mail chan- customer repurchase for (a) telephone ¥ e-mail contactnels in combination. However, they measure the interaction volume, (b) telephone ¥ mail contact volume, and (c) e-between channels as the number of times two communica- mail ¥ mail contact volume.tion channels were used concurrently in a given month, anoperationalization that may understate multiplicative effects The Moderating Role of Channel Preferencebecause it does not capture the total volume of communica- Various perspectives imply that communication channeltion through each channel. preference is idiosyncratic and that heterogeneity in indi- An opposing perspective is also plausible: Customers vidual channel preferences influences customer response tomight demonstrate reactance and respond negatively when relational communication. For example, the telephonethey receive relational communication through a combina- channel is commonly perceived as one of the most intrusivetion of channels. The use of multiple channels might imply (Reinartz, Thomas, and Kumar 2005), but some customersthat the firm is employing an ad hoc rather than a cus- prefer its interpersonal nature, which enables them totomized approach to communicate with customers. Just as request clarification and elaboration of the message (Robertsdifferent communication channels offer distinct benefits, and Berger 1999). Some customers perceive a company’sthey also provide distinct disadvantages. For example, the use of costly communication channels as a proxy for rela-mail channel can be environmentally wasteful, and the e- tionship investment, whereas others regard such costs as anmail channel can heighten privacy concerns (Morimoto and inefficient expense that ultimately increases priceChang 2006). From this perspective, contacting customers (Palmatier 2008). Boulding et al. (2005) encouragethrough multiple channels implies an uninformed commu- researchers to examine such heterogeneity to understandnication program, because the combined channels require customer response to relationship marketing actions.the customer to manage a variety of annoyances that Reciprocal action and reactance theories both suggestdecrease the overall utility of the communication. Cus- that customers’ channel preferences moderate theirtomers ultimately feel trapped and victimized by a seem- response to relational communication and produce a shiftingly manipulative relational marketplace (Fournier, Dob- along the reciprocity–reactance continuum to a higher idealscha, and Mick 1997). point. The reciprocity principle suggests that identifying Theoretically, the reactance-based perspective predicts and using preferred channels enhances customers’ motiva-that the use of multiple channels shifts the ideal point in one tion to reciprocate because they appreciate the company’sor both channels so that reactance is triggered at a lower personalization efforts. As a result, customers respond morevolume. Customer response to relational communication positively to higher volumes of communication than theyvolume in one channel peaks at a lower level and then do when their preferred channels are not used. Althoughdiminishes as the volume of communication in another prior research has not examined the moderating effect ofchannel increases. This effect would manifest as a negative channel preferences theoretically or empirically, this expec-interaction between any two channels used in combination tation is consistent with the finding that direct communica-and results in a diagonal, downward shift in the response tion aligned with customer needs increases perceived rela-function that attenuates the peak response. tionship investment and encourages reciprocity (De Wulf, Although a paucity of research has examined the effects Odekerken-Schroder, and Iacobucci 2001).of multichannel communication on repurchase, Prins and Reactance theory suggests that matching the communi-Verhoef (2007) offer evidence of customer reactance to the cation channel with customers’ preferences attenuates reac-use of multiple channels. They report that telephone contact tance related to communication volume. When a customerand mass communication (e.g., television, radio, print, out- views a channel as intrusive, for example, managing thatdoor) have negative interaction effects on adoption timing channel’s contacts requires greater effort and creates morefor a new service. They suggest that the use of both com- reactance (Kivetz 2005). If customers prefer channels thatmunication channels in combination causes customers to provide greater utility, the matching process enhances thebelieve that the firm is placing too much attention on the overall value of the communication, thereby decreasingnew service and pressuring them excessively to adopt. reactance and increasing repurchase response. This perspec-Although that study examines the interaction between direct tive is consistent with Fitzsimons and Lehmann (2004),telephone and mass channel communication rather than who find that offering product recommendations that aremultiple direct, personalized channels, the findings suggest inconsistent with a customer’s a priori preferences triggersthat communication through a combination of channels can reactance, but providing recommendations that align withhave a negative interaction effect on purchase behavior. preferences enhances satisfaction with the purchase deci- In summary, both theory and empirical evidence suggest sion process. The authors advocate that firms directly mea-multiplicative, interaction effects between multichannel sure customers’ preferences.98 / Journal of Marketing, July 2011
  6. 6. In effect, both theoretical perspectives imply that align- For each customer, we matched survey responses withing channels with customers’ preferences increases recep- the corresponding objective data from the company’s con-tivity to the communication and shifts the ideal point to a tact records and transaction database. Contact recordshigher volume. Formally, included the dates each customer was contacted and the communication channel used for each contact. The transac- H3: Customer preference for a channel shifts the ideal point tion database captured the date of each customer’s visit and for relational communication through that channel and enhances customer response, which implies a positive the dollar amount spent. To facilitate a longitudinal analysis, interaction effect on customer repurchase for (a) telephone we aggregated the data into 13 quarterly periods (quarters channel preference ¥ telephone contact volume, (b) e-mail 0–12), which constitute the 39-month observation period. channel preference ¥ e-mail contact volume, and (c) mail The combined data sets represent panel data that are channel preference ¥ mail contact volume. subject to several forms of bias (Hsiao 2004). First, cross- sectional differences in the dependent variable not captured by the explanatory variables can manifest as a heterogeneity Empirical Application bias and produce inconsistent estimates of the coefficientsOur research design features data from three sources, com- of interest. Second, the survey data we use to capture thebining survey data that capture customer channel prefer- panel’s communication preferences are subject to selectionences with 39 months of customer contact history and bias if the panelists’ decision to respond to the survey isrepurchase behavior. We collected the data for customers of related to repurchase, the dependent variable of interest.a large automobile dealership with a high-volume service Third, our analysis of customer response to relational com-department. The survey sampling frame included 3370 ran- munication is subject to endogeneity bias because mar-domly selected customers who had visited the service keters tend to communicate more with customers whodepartment within the past year. We sent each customer a repurchase more, so causality may be circular. We nowpacket that included a letter from the owner of the dealer- describe the key variables and the methods we use to con-ship, a five-page survey, a postage-paid return envelope, and trol for all three forms of bias.an offer for a $5 gift card on return of the completed ques- Variablestionnaire. We mailed follow-up surveys to all nonrespon-dents four weeks after the initial mailing. The two mailings Table 2 presents descriptive statistics and variable cor-produced 180 undeliverable addresses and 1162 complete relations. The Appendix provides additional details on theresponses, for a 36% effective response rate. The majority measurement of each variable.of the respondents were men (57%) and were between the Dependent variables. Consistent with the theoreticalages of 35 and 64 years (60%). Sixty-nine percent had some framework, the dependent variable of interest should cap-technical or university education, and 66% had an average ture customer response to relational communication. Wehousehold income exceeding $63,000. examined two direct measures of customer response: repur- TABLE 2 Descriptive Statistics and Correlation MatrixVariables 1 2 3 4 5 6 7 8 9 10 1 Repurchase spendinga 1.0 2 Repurchase visitsa .59 1.0 3 Telephone contactsa .15 .27 1.0 4 E-mail contactsa .09 .18 .04 1.0 5 Mail contactsa .16 .29 .22 –.02 1.0 6 Telephone preference .01 .05 .05 –.08 .07 1.0 7 E-mail preference .02 .04 –.07 .21 –.02 –.16 1.0 8 Mail preference –.03 –.04 –.02 –.05 –.01 .13 .10 1.0 9 Warranty worka .43 .41 .10 .05 .11 .01 .03 –.01 1.010 Number of vehicles .15 .30 .04 .12 .15 .03 .07 .00 .09 1.0 Mean 122.3 2.12 1.25 .62 1.31 3.54 2.81 3.12 25.47 2.20 Standard deviation 295.9 2.80 1.39 1.38 1.92 1.15 1.28 1.08 112.0 1.49 Standard error 2.50 .02 .01 .01 .02 .01 .01 .01 .95 .01 Skewness 6.68 2.02 1.36 2.89 2.28 –.58 .13 –.21 15.62 .86 Minimum 0 0 0 0 0 1 1 1 0 1 Maximum 5376 25 14 14 16 5 5 5 4302 5 Median 25.40 1 1 0 1 4 3 3 0 1 Lower 99% CL 115.8 2.06 1.22 .59 1.27 3.51 2.78 3.10 23.03 2.17 Lower 95% CL 117.4 2.07 1.23 .59 1.28 3.52 2.79 3.10 23.61 2.18 Upper 95% CL 127.2 2.17 1.28 .64 1.35 3.56 2.83 3.14 27.32 2.22 Upper 99% CL 128.7 2.18 1.28 .65 1.36 3.56 2.84 3.14 27.91 2.23aQuarterly measures.Notes: Correlations greater than |.02| are significant at p < .05 (two-tailed test). Executing Multichannel Relational Communication / 99
  7. 7. chase visits and repurchase spending for each customer dur- DVij  0 + 1TIME + 2TIME2 + 3iing quarters 1–12. We log-transformed both dependent + 4lagSPENDi +5lagVISITi + 6WWi + 7VEHivariables to improve distribution normality. + 8PrefPHONEi + 9PrefEMAILi + 10PrefMAILi Control variables. We included mean-centered mainand quadratic terms for the 12 quarters (TIME) to allow for + 11PHONEi + 12EMAILi + 13MAILitrend effects. To control for heterogeneity across individual + 14PHONEi2 + 15EMAILi2 + 16MAILi2respondents, we included four exogenous variables that arenot directly related to our research questions but are related + 17PHONEi ¥ EMAILi + 18 PHONEi ¥MAILito auto service spending levels. All four variables were cap- + 19 EMAILi ¥ MAILi + 20PrefPHONEi ¥PHONEitured in the transaction database: lagged repurchase spend-ing (lagSPEND), lagged number of repurchase visits + 21PrefEMAILi ¥ EMAILi + 22PrefMAILi(lagVISIT), the amount of warranty work in the current ¥ MAILi + ei1,quarter (WW), and the number of vehicles each respondent where the time subscript t is implied for all variables and lagowned (VEH). We log-transformed the warranty work indicates that the measure was taken from the quarter beforevariable to improve distribution normality. the dependent measure observation. (i.e., t – 1) In addition, We controlled for selection bias using Heckman’s(1979) two-step procedure. We first estimated the probabil- DVij = customer i’s quarterly response,ity of responding to our survey using relevant information j= repurchase visits or spending,for all customers in the sampling frame: the number of rela- TIME = quarter from 1 to 12,tional communications before the survey through each com- i = selection control factor for customer i,munication channel; household income (HI), which we SPEND = quarterly repurchase spending by customer i,determined from census data by zip code; and whether the VISITi = number of quarterly repurchase visits byrespondent had moved to a different home address during customer i,the observation period (MOVE), as captured in the transac- WWi = amount of warranty work completed fortion database. We then created an inverse Mills ratio for customer i,each respondent (, which is a monotonic decreasing func- VEHi = number of vehicles serviced for customer i,tion of the probability that each customer responded to our PrefPHONEi = customer i’s self-reported preference forsurvey. Including the inverse Mills ratio in the empirical the telephone channel,model controls for the effect of unmeasured characteristics PrefEMAILi = customer i’s self-reported preference forrelated to the selection process. the e-mail channel, PrefMAILi = customer i’s self-reported preference for Exogenous independent variables. Channel preference the mail channel,was a self-reported measure that captured customers’ PHONEi = number of quarterly telephone contacts tar-responses to a single five-point Likert-scale question asking geting customer i,whether the customer preferred to be contacted through each EMAILi = number of quarterly e-mail contacts target-channel (PrefPHONE, PrefEMAIL, and PrefMAIL). We have ing customer i,no theoretical expectations for the effect of channel prefer- MAILi = number of quarterly mail contacts targetingence on repurchase, but because underlying curvilinearity customer i, andcan bias tests of interactions (e.g., Cortina 1993), we exam- ei1 = autoregressive error term.ined both main and quadratic terms to ensure that unexpectednonlinear relationships did not emerge as a significant inter- For the three endogenous relational communication vol- ume equations, we used exogenous, lagged variables as pre-action effect. None of the quadratic terms were significant. dictor variables: Endogenous independent variables. Three independent CVik  0 + 1TIME + 2TIME2 + 3QUARTER2variables capture the volume of relational communicationthe firm sent to its customers by telephone (PHONE), e- + 4QUARTER3 + 5QUARTER4 + 6lagSPENDimail (EMAIL), and mail (MAIL). We determined the num- + 7lagSPENDi2 + 8lagVISITi + 9lagVISITi2ber of contacts according to the date of each contact in eachquarter of the firm’s customer contact database. We mean- + 10WWi11VEHi + 12lagPHONEi + 13lagPHONEi2centered the contact measures to facilitate interpretation of + 14lagEMAILi + 15lagEMAILi2 + 16lagMAILithe quadratic and interaction effects. + 17lagMAILi2 + 18lagPHONEi ¥ lagEMAILiModel Specification + 19lagPHONEi ¥ lagMAILi + 20lagEMAILiHausman (1978) tests confirmed that the three relational ¥ lagMAILi + 21HIi + 22HIi2 + 23lagMOVEi + eij2,communication volume measures were endogenous. Tocontrol for this endogeneity, we estimated a simultaneous wheresystem of four equations for each of the endogenous CVik = communication volume for cus-variables. We specified the dependent variable (repurchase tomer i through channel k =visits or spending) equation as follows: telephone, e-mail, and mail;100 / Journal of Marketing, July 2011
  8. 8. QUARTER2, 3, and 4 = dummy variables to control for 4). For expositional brevity, we report results for the analy- programmatic seasonal changes ses using repurchase spending as the dependent variable. in communication volume; We used the Bayesian information criterion (BIC) to assess HIi = annual household income of overall model fit. Model 1 was a baseline model that did not customer i; and include quadratic or interaction terms. In Model 2, we MOVEi = dummy variable indicating that added the three relational communication volume quadratic customer i had moved to a new terms that test H1. Applying generally accepted standards home address. for model fit (Kass and Raftery 1995), we can conclude that We used full-information maximum likelihood analysis the addition of the relational communication volume qua-to estimate the system of equations (Chow 1964). We dratic terms (Model 2) results in very strong improvementexamined five hierarchical models to assess the robustness in fit (Table 4; BIC = 395.6).of individual results and overall fit of each model (see Tables We then compared the BIC for Models 3–5 with that for3 and 4). To explore whether multicollinearity might be Model 2. In Model 3, we added the three two-way relationalbiasing the estimates, we examined the condition numbers communication volume interactions that test H2. We alsoin each model; none exceeded 15, which is well below the examined higher-order terms to fully explore how the ideallevel that would indicate potential problems (Greene 1990). point shifts in the presence of two-way interaction effects. Specifically, we considered the three-way interactionModel Selection among telephone, e-mail, and mail contact volume, whichThe analyses produced similar empirical results for both assesses whether the total volume of communicationrepurchase visits (Table 3) and repurchase spending (Table through all three channels significantly altered the ideal TABLE 3 Simultaneous Equation Results for Repurchase Visits Model 1 Model 2 Model 3 Model 4 Model 5Intercept 1.04*** 1.09*** 1.06*** 1.09*** 1.04***Time –.01* –.01** –.01*** –.01** –.01***Time2 –.01*** –.01*** –.01*** –.01*** –.01***Selection control factor –.09*** –.08*** –.08*** –.08*** –.08***Lagged spending –.00 –.00 –.00 –.00 –.00Lagged visits .02*** .03*** .03*** .03*** .03***Amount of warranty work .18*** .19*** .19*** .19*** .19***Number of vehicles .07*** .08*** .08*** .08*** .08***Main Effects Telephone preference .03*** .03*** .02*** .05*** .04*** E-mail preference -.00 -.00 .00 .00 .00 Mail preference –.01* –.01* –.01* –.01* –.01* Telephone contacts .03* .04** .02 .04** .02* E-mail contacts .08*** .09*** .04*** .09*** .04*** Mail contacts .16*** .15*** .14*** .15*** .14***Volume Quadratic Effects Telephone contacts2 H1a – –.01*** –.01*** –.01*** –.01*** E-mail contacts2 H1b – –.01*** –.01*** –.01*** –.01*** Mail contacts2 H1c – –.01*** –.01*** –.01*** –.01***Volume Interactions Telephone contacts × e-mail contacts H2a +/– –.02*** –.02*** Telephone contacts × mail contacts H2b +/– .00 .00 E-mail contacts × mail contacts H2c +/– –.02*** –.02*** Telephone contacts2 × e-mail contacts .01*** .01*** Mail contacts2 × e-mail contacts .002*** .002***Preference × Volume Interactions Telephone preference × phone contacts H3a + .01*** .01*** E-mail preference × e-mail contacts H3b + .01** .01** Mail preference × mail contacts H3c + -.00 –.00 Telephone preference × phone contacts2 –.004** –.003**Model Fit R2 .42 .46 .46 .46 .47 BIC 170,212.2 169,849.2 169,731.7 169,835.0 169,722.7 BIC (compared with Model 1) 363.0 480.4 377.2 489.4 BIC (compared with Model 2) 117.5 14.2 126.5*p < .05.**p < .01.***p < .001 (one-tailed t-tests for directional hypotheses; two-tailed for nondirectional hypotheses). Executing Multichannel Relational Communication / 101
  9. 9. TABLE 4 Simultaneous Equation Results for Repurchase Spending Model 1 Model 2 Model 3 Model 4 Model 5Intercept 3.51*** 3.68*** 3.54*** 3.67*** 3.53***Time –.01 –.01* –.02** –.01* –.02**Time2 –.04*** –.03*** –.03*** –.03*** –.03***Selection control factor –.22*** –.20*** –.20*** –.20*** –.20***Lagged spending .00 .00 .00 .00 .00Lagged visits .05*** .07*** .06*** .07*** .06***Amount of warranty work .50*** .51*** .51*** .51*** .51***Number of vehicles .18*** .19*** .19*** .19*** .19***Main Effects Telephone preference .10*** .09*** .08*** .17*** .15*** E-mail preference .00 -.01 -.01 .00 .01 Mail preference –.02 –.03 –.02 –.02 –.02 Telephone contacts .11** .14*** .07 .15** .08* E-mail contacts .26*** .32*** .16*** .32*** .15*** Mail contacts .58*** .54*** .48*** .53*** .48***Volume Quadratic Effects Telephone contacts2 H1a – –.04*** –.03*** –.04*** –.03*** E-mail contacts2 H1b – –.04*** –.03*** –.04*** –.03*** Mail contacts2 H1c – –.04*** –.03*** –.04*** –.03***Volume Interactions Telephone contacts × e-mail contacts H2a +/– –.08*** –.08*** Telephone contacts × mail contacts H2b +/– –.01 –.01 E-mail contacts × mail contacts H2c +/– –.06*** –.07*** Telephone contacts2 × e-mail contacts .02*** .02*** Mail contacts2 × e-mail contacts .01*** .01***Preference × Volume Interactions Telephone preference × phone contacts H3a + .04*** .04*** E-mail preference × e-mail contacts H3b + .02* .02** Mail preference × mail contacts H3c + .00 –.00 Telephone preference × phone contacts2 –.02*** –.02**Model Fit R2 .34 .39 .40 .39 .40 BIC 203,537.6 203,142.0 203,006.7 203,120.6 202,991.3 BIC (compared with Model 1) 395.6 530.8 417.0 546.3 BIC (compared with Model 2) 135.3 21.4 150.7*p < .05.**p < .01.***p < .001 (one-tailed t-tests for directional hypotheses; two-tailed for nondirectional).points, and the interactions between the channel volume We estimated all theoretical relationships simultane-quadratic terms, which assess whether each curvilinear ously in Model 5, which produced strong improvements inchannel volume relationship varied as a function of the vol- fit over Model 2 (BIC = 150.7), Model 3 (BIC = 15.4),ume in other channels (e.g., Luo and Donthu 2006). We and Model 4 (BIC = 129.3). Therefore, we focus theincluded two significant higher-order terms in Model 3, remaining discussion on the Model 5 results.which produced strong improvement in fit over Model 2(BIC = 135.3). Results For Model 4, we added the channel preference ¥ chan- The results for the control variables are logical. The mainnel volume interactions that test H3. We also examined effect and quadratic coefficient for time are both negative,channel preference ¥ channel volume quadratic interac- suggesting a downward trend in repurchase over time. Thetions, which assess whether the curvilinear channel volume selection control factor is significantly negative, suggestingrelationship varied as a function of channel preference. We that nonrespondents repurchased more than respondents.included one significant quadratic interaction term in Model This might indicate that people with more free time (per-4, which resulted in a strong improvement in fit over Model haps older and retired) were more willing to respond to the2 (BIC = 21.4). A comparison of Model 4 with Model 3 survey than were younger, working-age people with moreoffers no theoretical insight but is practically noteworthy: vehicles in the household who subsequently repurchasedThe BIC and R-square values indicate that the relational more. The coefficient for lagged spending is not significant,communication volume interactions (Model 3) have greater consistent with the idea that large automotive repair expen-predictive ability than do the channel preference ¥ channel ditures in one quarter might preclude large expenditures involume interactions (Model 4). the following quarter. The coefficient for lagged visits is102 / Journal of Marketing, July 2011
  10. 10. positive, indicating that relational customers return regu- FIGURE 1larly for routine maintenance. The coefficients for warranty Ideal Points for Impact of Relationalwork and number of vehicles are also positive, indicating Communication on Repurchase Spendingthat more work performed under warranty and more cars inthe household translate into greater spending on mainte- A: Impact of Telephone Contacts on Repurchasenance and repairs. We now examine the results related to Spendingour three research questions. A positive main effect and a negative quadratic coeffi- $80cient for contact volume through a given channel supportthe classic ideal point hypothesis (H1). The results in Table $604 offer full support for H1, because the main effect for com- Spendingmunication volume is positive, and the quadratic coefficientis negative for telephone, e-mail, and mail contacts. These $40results indicate that the volume of relational communicationthrough each channel has a positive impact on repurchase $20until the ideal point is reached. After the ideal point isexceeded, increasing volume has a negative effect onspending. Figure 1 illustrates the inverted U-shaped effects $0 0 1 2 3 4 5 6 7 8 9 10in the plots. For each channel, the first few contacts create a Number of Phone Contactspositive response. The positive response peaks at approxi-mately three contacts for the telephone channel, betweenthree and four contacts for the e-mail channel, and between B: Impact of E-Mail Contacts on Repurchase Spendingnine and ten contacts for the mail channel.1 Increasing vol- $80ume beyond these points triggers negative reactance amongcustomers, causing them to repurchase less. A significant interaction effect for contact volume in $60two channels supports a shift in the ideal point (H2). The Spendingresults in Table 4 provide full support for two of the three $40predictions from H2. The telephone contacts ¥ e-mail con-tacts and e-mail contacts ¥ mail contacts interactions areboth significantly negative. These negative two-way inter- $20actions are consistent with a reactance theory perspective.The telephone contacts ¥ mail contacts interaction coeffi- $0cient also is negative but is only marginally significant 0 1 2 3 4 5 6 7 8 9 10using a two-tailed t-test (p < .10; t-value = –1.85) and is not Number of E-Mail Contactssignificant in the repurchase visits analysis (Table 3). We present graphs for the two significant cross-channel C: Impact of Mail Contacts on Repurchase Spendinginteraction effects in Figure 2. Panel A shows that customerresponse to e-mail contacts follows an inverted U-shaped $240pattern consistent with reciprocity followed by reactancefor all levels of telephone contact. There is a clear shift inthe ideal point for e-mail contacts as the number of tele- $180phone contacts increases. When there is one telephone con- Spendingtact, the ideal number of e-mail contacts is between five andsix, but the ideal number of e-mail contacts drops to $120between two and three when there are three to five tele-phone contacts. This ideal point shift is consistent with $60cross-channel reactance, which leads to lower repurchase asthe number of telephone and e-mail contacts jointlyincreases. The results in Panel B also indicate a shift in the $0 0 1 2 3 4 5 6 7 8 9 10ideal point consistent with reactance. When there is one Number of Mail Contactsmail contact, the ideal number of e-mail contacts is approx-imately five, but the ideal number of e-mail contacts dropsto one when the number of mail contacts is five. to communications through that channel. Consistent with We also find support for H3, which predicts that prefer- H3a, the telephone channel preference ¥ telephone contactence for a channel leads to more positive customer response volume coefficient is significantly positive. Figure 3, Panel A, shows that customer response follows the inverted U- 1All values in the graphs fall within the range of the data. Maxi- shaped pattern that is consistent with reciprocity followed bymum values were 14 telephone, 14 e-mail, and 16 mail contacts reactance when customer preference ranges from moderateper quarter. to high. There is no perceptible shift in the ideal point, which Executing Multichannel Relational Communication / 103
  11. 11. FIGURE 2 FIGURE 3 Plotting Significant Volume Interaction Effects Plotting Significant Preference ¥ Volume Interaction Effects A: Telephone ¥ E-Mail Interaction Effect on Repurchase Spending A: Telephone Preference ¥ Telephone Contacts Interaction Effect on Repurchase Spending $80 $80 $60Spending $60 Spending $40 $40 $20 $20 $0 0 1 2 3 4 5 6 7 $0 Number of E-Mail Contacts 0 1 2 3 4 5 6 7 1 phone contact Number of Phone Contacts Low preference Low preference Moderate preference Moderate preference High preference High preference 3 phone contacts 1Low preference phone contact contact contacts 3 phone contacts 5 phone contacts contacts 5 phone contacts 1 phone contact contact 3 phone contacpreference phone contacts Moderate ts contacts 5 contacts 1 phone contact contact 3 phone contacts contacts 5High preference phone contacts contacts B: Mail ¥ E-Mail Interaction Effect on Repurchase Spending B: E-Mail Preference ¥ E-Mail Contacts Interaction Effect on Repurchase Spending $120 $100 $80Spending $80 $60 Spending $60 $40 $40 $20 $20 $0 0 1 2 3 4 5 6 7 $0 Number of E-Mail Contacts 0 1 2 3 4 5 6 7 Number of E-Mail Contacts 1 mail contact Low preference Low preference Moderate preference Moderate preference High preference High preference 3 mail contacts 1Low preference phone contact contact 3 phone contacts contacts 5 phone contacts contacts 5 mail contacts 1 phone contact contact 3 phone contacpreference phone contacts Moderate ts contacts 5 contacts 1 phone contact contact 3 phone contacts contacts 5High preference phone contacts contactsoccurs at approximately three phone contacts. However,when preference for the channel is low, customers exhibit noresponse, either positive or negative, to any phone contacts. Discussion H3b receives support because the e-mail channel prefer- This research was motivated by three key questions: Inence ¥ e-mail contact volume coefficient is significantly multichannel relational communication, is there an idealpositive. Panel B in Figure 3 shows a shift in ideal point. volume? Are volume effects across channels additive orFor customers who prefer the e-mail channel, repurchase is multiplicative, and if the effects are multiplicative, are thehighest at five e-mail contacts; for customers who do not interactions positive or negative? and To what extent doprefer that channel, repurchase is highest between two and customers’ channel preferences moderate the relationshipthree e-mail contacts.2 between communication volume and repurchase? We use the classic ideal point to conceptualize a response contin- 2We also examined whether a “mismatch” between channel uum that reflects reciprocity followed by reactance inpreference and channel volume would produce a negative shift in the response to multichannel communication volume. Weideal point. Three of the 6 mismatch interactions were significantly extend this theoretical framework to explain volume inter-negative: telephone preference ¥ mail volume, e-mail preference ¥telephone volume, and e-mail preference ¥ mail volume. We do not action effects between channels and gauge the strength ofinclude these terms in our analyses because inclusion did not change channel preferences as moderating effects.the results for the hypothesized relationships and because our mea- We provide empirical support for our framework bysures of (lower/higher) preference for one channel do not necessar- matching longitudinal customer transaction data with rela-ily translate into (higher/lower) preference for another channel. For tional communication records and a large-scale customerexample, sending a high volume of mail communication to some-one who indicates a high preference for the telephone channel survey. Our results offer concrete answers to the researchdoes not necessarily imply a mismatch, because that person might questions and yield new insights into the unintended andalso have a high preference for the mail channel (r = .13, Table 2). potentially detrimental effects of multichannel relational104 / Journal of Marketing, July 2011
  12. 12. communication. First, the results indicate that there is an sitivity toward the telephone channel. Reactance is gener-ideal volume of relational communication that varies across ated at even lower volumes when customers receive highchannels; after the ideal point is reached, additional com- levels of e-mail contacts. For customers who prefer not tomunication generates reactance and increasingly negative be contacted by telephone, the ideal point for telephonecustomer response. Second, we find that volume effects contacts is zero because these customers exhibit no positiveacross channels are multiplicative and that, in contrast with response to the company’s efforts to build a relationshipprior studies that find synergies between channels, the inter- through telephone communication. On average, these cus-actions are negative, indicating that the ideal level of com- tomers also repurchase less than customers who are moremunication through one channel decreases as communica- receptive to telephone communication. Thus, for the intru-tion volume in other channels increases. Specifically, our sive telephone channel in which message avoidance is diffi-results show that combining e-mail contacts with either cult, reactance occurs early, and the subsequent negativetelephone or mail contacts generates reactance at a faster effects are strong.rate; the ideal volume of e-mail contacts decreases as the Following this logic, we believe that the e-mail channelnumber of telephone or mail contacts increases. Finally, we is somewhat less intrusive than the telephone channelfind that customers’ preferences for the telephone and e- because it does not necessarily disrupt the receiver in realmail channels positively moderate the impact of communi- time. Messages can be reviewed and deleted whenever thecation volume through each of those channels, respectively. receiver chooses, and the receiver has the ability to flag senders so that future messages go directly into a junkTheoretical Implications folder. As a result, e-mail contacts prompt reactance afterOur conceptual framework shares some characteristics with four contacts per quarter on average (Figure 1, Panel B).the two-factor model of mass marketing repetition, as we The intrusiveness of e-mail is amplified, however, for cus-noted previously. Experimental studies have uncovered tomers who receive higher volumes of communication byempirical support for the two-factor model, but there has telephone (Figure 2, Panel A) or mail (Figure 2, Panel B).been little support in field studies. For example, in a study It seems that respondents viewed the postal mail chan-of television advertising, Tellis, Chandy, and Thaivanich nel as the least intrusive and the easiest to ignore, which is consistent with prior reports that consumers view unso-(2000, p. 43) conclude that their results “do not provide licited e-mail as more intrusive than postal mail (Morimotosupport for the two factor theory of advertising response … and Chang 2006). Apparently, the ease with which direct[perhaps because] the effects of advertising frequency are mail is disposed of reduces activation of negative reactance.too weak to register in a field setting.” As a result, reactance did not appear until after nine mail- One explanation for why negative consumer reactance ings per quarter (Figure 1, Panel C). Given the high numbermay not manifest in a field study pertains to individual con- of contacts required to trigger reactance and the relativelytrol over message processing. If a consumer leaves the low message processing costs for direct mail, reactance toroom during commercial breaks or fast-forwards through postal mail may be infrequent.commercial breaks when watching prerecorded shows, We believe that the subsequent onset of reactancereactance does not occur because the consumer has not toward mail contacts may inform previous failures tobeen compelled to expend effort to process the advertise- uncover reactance toward advertisements in field studies.ment. Despite the widespread occurrence of message avoid- The receiver’s ability to retain control and avoid messageance, the phenomenon has received little theoretical or processing attenuates the onset of reactance so that mes-empirical examination. sages through channels such as television (in which the Our results offer tantalizing insights into the role of receiver has significant control over message processing)individual control over message processing on response to may rarely evoke negative reactance. These inferences arecommunication volume. Specifically, they suggest that consistent with experimental research demonstrating that anreactance occurs more quickly in the telephone channel, inverted U-shaped response occurs when participants per-more slowly in the e-mail channel, and much more slowly form deep processing of advertisements but does not occurin the mail channel. Furthermore, our findings suggest that with shallow processing (Nordhielm 2002).reactance occurs more quickly when customers receive The overall story that emerges is that customers reactcommunication through multiple channels and occurs more negatively to multichannel relational communication levelsslowly when customers receive communication through that exceed their ideal point. This negative reactance ischannels they prefer. We speculate that these results pertain muted if customers can control and avoid exposure to theto the intrusiveness of the communication and the person’s message. Customers may perceive greater control overcorresponding lack of control in avoiding the message. messages sent through channels that are less intrusive or for We take the perspective that the telephone channel is the which they have greater preference. Reactance is height-most intrusive among the three channels we examined. A ened if customers believe that control and avoidance is dif-telephone call disrupts the receiver in real time or, if a voice ficult, especially when messages come through multiplemessage is left, forces the receiver to process the message channels. These findings underscore the importance ofat least minimally to determine whether he or she wants to examining the effects of multichannel communication vol-delete it. As a result, telephone contacts prompt reactance ume by considering the impact of specific channels, indi-after three contacts per quarter on average (Figure 1, Panel vidually and in combination, rather than aggregate volume.A). However, this average reactance point understates sen- Our findings also highlight the importance of considering Executing Multichannel Relational Communication / 105
  13. 13. individual customer characteristics, such as channel prefer- operationalize telephone communication as a dummyence, that moderate the impact of relational communication variable capturing whether any telephone contact occurredon repurchase. in a given month; restricting the independent variable to two levels (0, 1) cannot capture curvilinear effects. DeReconciling Empirical Findings Wulf, Odekerken-Schroder, and Iacobucci (2001) measureOur review of empirical studies examining the effect of customers’ perceptions of firm communication using arelational communication on repurchase behavior indicates seven-point Likert scale, which creates a ceiling in assess-that a majority uncover effects consistent with reciprocity ing the level of communication. In contrast, studies sup-(see Table 1). Of the studies we examined, only one (i.e., porting an ideal point use independent measures with unre-Prins and Verhoef 2007) reports a negative interaction stricted distributions, such as the number of contacts andeffect consistent with reactance, and just three (i.e., Drèze number of days since the last contact (Drèze and Bonfrerand Bonfrer 2008; Kumar, Venkatesan, and Reinartz 2008; 2008; Kumar, Venkatesan, and Reinartz 2008).Venkatesan and Kumar 2004) report ideal points consistent In summary, plausible theoretical, methodological, andwith reciprocity followed by reactance. We offer several substantive explanations exist for the prevalence of reci-explanations for the prevalence of reciprocity effects. procity effects in prior research, which should encourage First, reactance effects may be attenuated in some con- additional research to try to understand when customers aretexts due to inertia or switching barriers. For example, Ver- likely to demonstrate reciprocity or reactance in response tohoef (2003) and Rust and Verhoef (2005) report effects con- relational communication.sistent with reciprocity in an insurance context that can bedescribed as an ongoing, contractual service with desig- Managerial Implicationsnated renewal periods. Generating reactance in a continuing Increased access to individual-level customer informationservice context, in which inertia and switching barriers are has accelerated the use of targeted, multichannel communi-high, might require far higher communication volume than cation. Despite significant financial investments, firms havein contexts featuring repeated, noncontractual transactions, surprisingly limited knowledge about how the simultaneoussuch as automobile service. use of multiple communication channels affects customer Second, the theory, research design, and analysis in response. Our results can inform multichannel communica-prior studies may not have considered curvilinear effects. tion practice to strengthen customer relationships effec-Cortina (1993, pp. 917–18) notes that there is a “bias tively without wasting firm resources.against nonlinear hypotheses” that leads to “complex, non- Managers should carefully identify which channels areadditive models without consideration of possible nonlinear most effective in reaching their customers and avoid theeffects.” Of the studies in Table 1 that report linear effects assumption that different communication channels exertconsistent with reciprocity, none reports the results of tests equivalent effects. As an example, the low cost per contactto explore whether the relationship between communication and quick execution of e-mail might encourage companiesand repurchase might be curvilinear. Thus, we cannot dis- to increase the proportion of e-mail contacts, under thecern whether the underlying effects of communication were assumption that e-mail communication is equally effective.linear or curvilinear. Given emerging evidence in support of Our industry partner increased the proportion of e-mailan ideal point of communication, we encourage further communication from 2% of total contacts in the first quarterresearch to assess curvilinear effects on repurchase explic- to 10% in the twelfth quarter. This increased allocation isitly by including quadratic terms. not warranted if the e-mail channel is less effective than Third, including quadratic terms is especially important other channels in driving customer repurchase. As Figure 1when examining interaction terms because underlying illustrates, this firm’s customers are far more responsive tocurvilinearity can bias tests of interaction effects (e.g., mail contacts than to either telephone or e-mail contacts.Cohen 1978; Cortina 1993). Two prior studies examining To maximize repurchase and minimize the likelihood ofcross-channel interactions have used empirical approaches negative customer reactance, managers should monitor totalthat do not include quadratic terms, which may explain the contact volume and explore how specific channel combina-variation in their findings; one reports a positive interaction tions may shift the ideal point. Our results show that e-mailconsistent with reciprocity (Reinartz, Thomas, and Kumar combines poorly with both telephone and mail contacts,2005), and the other reports a negative interaction consis- producing a decline in customer spending (Figure 2, Panelstent with reactance (Prins and Verhoef 2007). Another A and B). Conversely, the absence of a negative interactionexplanation for this difference is Reinartz, Thomas, and between telephone contacts and mail contacts suggests thatKumar’s (2005) operationalization of cross-channel interac- customers respond better to this combination than to thetions as the number of times both types of communication other combinations.occurred within the same month. This operationalization To exploit revealed channel preferences, firms shouldcreates a change in the intercept rather than a change in the develop protocols that limit total communication throughslope for one independent variable as a function of another. all channels and specify effective combinations of channels. Finally, uncovering curvilinear relationships consistent For example, the cross-channel interactions in our studywith an ideal point is sensitive to coarseness or range point to two segments: (1) customers who respond posi-restrictions in measuring the independent variable (Russell tively to traditional telephone and mail channels and (2)and Bobko 1992). For example, Prins and Verhoef (2007) customers who respond positively to the technological e-106 / Journal of Marketing, July 2011

×