Although academics and practitioners have embraced customer engagement as a major objective of marketing, the conceptualization and measurement of engagement is challenging. Prior research largely has relied on conventional “one-size-fits-all” measures with a fixed set of scale items. The current, more flexible approach measures engagement based on context-specific experiences that can vary across brands and products. Three studies examining engagement when consuming (a) live jazz music, (b) newspapers, and (c) television programming provided evidence that a flexible approach to measuring engagement can help predict consumer behavior. The third of these studies also provided new evidence that engagement with television programming increases advertising effectiveness.
Most conventional measures of engagement take a “one-size-fits-all” approach by generating a fixed set of scale items.
The authors instead suggest a flexible approach for measuring engagement based on qualitatively rich, context-specific experiences that can vary across brands and products.
This flexible-measurement approach is well suited to the engagement construct and highly predictive of consumption behavior, sometimes more so than traditional fixed-scale measures such as satisfaction.
In addition to monitoring satisfaction, marketers should develop and study engagement metrics (ideally by focusing on experiences that relate to consumers' goals) to attain a multifaceted understanding of their customers.
Over the past 15 years, the construct of engagement has gathered considerable momentum as a way of expanding marketers' insight into consumers. Practitioners and academics have devoted substantial attention to both explaining and measuring the engagement construct, as evidenced by over 16,400 articles currently returned in Google Scholar from searches for either “customer engagement” or “consumer engagement” (searches conducted on August 17, 2015).
Yet, a recent list outlining the top research priorities shared by many of the largest marketing organizations in the world suggests that more work remains to be done on the topic of engagement. This list, published in 2014 by the Marketing Science Institute, a learning organization that bridges marketing theory and practice, included the following topic of interest: “How should engagement be conceptualized, defined, and measured?” (Marketing Science Institute, 2014)
The current research sought to further advance the engagement construct and relate it to advertising practice. To that end, the authors developed a new measurement approach that examines the impact of engagement on consumption. Although most prior definitions of engagement have agreed—or at least have not precluded—that the construct is likely to be context-specific (i.e., variable across domains, product categories, and brands), most existing engagement metrics largely have failed to take this into account and instead posit a one-size-fits-all set of items to measure engagement.
In contrast, the authors of the current article contended that the experiences comprising engagement with any one product or brand can be different from those associated with another. A fixed set of items cannot capture such differences with any specificity.
This research accordingly proposed an approach to measuring engagement that would be flexible enough to accommodate context-specific indicators of experiences without altering the higher order meaning of the engagement construct. Three empirical studies examined consumer engagement with live jazz music, newspapers, and television programming.
Across disparate categories, the authors believe that these three studies provided evidence that the proposed flexible approach to measuring engagement would be predictive of consumption.
What Is Engagement?
Engagement has been a difficult concept to define, and many different definitions have been proposed. After an extensive survey of the marketing, management, and social science literatures, the current authors aligned their definition of engagement with the following (italics provided in the original definition):
Customer engagement (CE) is a psychological state that occurs by virtue of interactive, co-creative customer experiences with a focal agent/object (e.g., a brand) … under a specific set of context-dependent conditions … and exists as a dynamic, iterative process …in which other relational concepts… are antecedents and/or consequences. It is a multidimensional concept subject to a context- and/or stakeholder-specific expression of relevant cognitive, emotional and/or behavioral dimensions. (Brodie, Hollebeek, and Ilic, 2011, p. 260)
Although this definition was formulated with service encounters in mind, the authors of the current article believe that it represents an emerging view in marketing that locates engagement as a state that arises from experiences that are context specific (i.e., domain, category, or brand specific).
Engagement, under this definition, arises out of the different ways in which a product or service is experienced. These experiences reflect the individual's interaction with the product over time as a way of accomplishing personal goals (Calder and Malthouse, 2008). The level of engagement arising from these experiences can, in turn, affect behavioral variables, including the extent to which the product is consumed.
Others have defined customer engagement as “the intensity of an individual's participation in and connection with an organization's offerings or organizational activities, which either the customer or the organization initiates” (Vivek, Beatty, and Morgan, 2012). This definition captured the notion that experiences exist on a continuum ranging from detached to intense: High engagement comes from rich qualitative “felt” experiences that produce “a proactive, interactive customer relationship with a specific engagement object” (Brodie et al., 2011, p. 257).
The current research proposed a new definition for engagement that built on prior work on the topic (Calder and Malthouse, 2008; Calder, Malthouse, and Schaedel, 2009); the studies conducted demonstrate the utility of this definition.
Engagement is a multilevel, multidimensional construct that emerges from the thoughts and feelings about one or more rich experiences involved in reaching a personal goal.
Having different experiences makes engagement multidimensional. The aggregate of the experiences is the second-order engagement construct, making it multilevel.
The current authors conducted three studies that identified five broad categories of experiences that may constitute engagement:
Interaction (to connect with others),
Transportation (to escape or become diverted),
Discovery (to gain insight, knowledge, or skills),
Identity (to affirm or express one's identity), and
Civic Orientation (to contribute to society).
It is important to note, however, that these categories did not comprise an exhaustive, fixed checklist for engagement. Consistent with the notion of “context-specific expression” (Brodie et al., 2011), the specific experiences that contribute to engagement would vary depending on the type of domain, product category, and brand under investigation.
Further, because engagement is context specific, the current authors did not expect high levels of all five of the experience categories for engagement to emerge. For example, experiences that are high in “transportation” (i.e., feeling captivated or carried away) likely would be more relevant to one's engagement with the arts (e.g., music or theater) than one's engagement with, say, consumer packaged goods.
Although there has been research on the use of neural measures of engagement (Pynta et al., 2014; Stelle et al., 2013), most engagement metrics to date have used a fixed set of items that assumed all products or brands could be assessed using the same set of scale items (Brakus, Schmitt, and Zarantonello, 2009; Hollebeek, Glynn, and Brodie, 2014). The primary objective of this type of approach has been to develop a single measurement scale that is valid, reliable, and generalizable across multiple contexts. A fixed-scale approach offers many benefits, including the ability to easily compare levels of a construct across multiple domains, product categories, or brands.
In contrast, the authors of the current article advocated a different approach that could be adapted to multiple contexts (e.g., domains, product categories, or brands) without altering the meaning of the underlying engagement construct. Furthermore, such a contextual sensitivity is, in the authors' view, inherent to engagement because it is the product of intense, qualitatively rich experiences that are highly context specific.
Thus, the authors believe, a context-specific approach should yield a superior engagement measure because it provides more explanatory power and diagnostic value. Analogously, tailor-made suits offer a superior fit compared with those that are mass-produced.
To clarify the distinction between a flexible approach and a fixed-scale approach, take the example of a Twitter user and a museum patron who each are highly engaged in their respective pursuits. Although their level of engagement may be similar, the Twitter user's engagement likely will arise from social interaction, whereas the museum patron's engagement more likely will arise from the process of discovery and learning. A single scale consisting of items related to social interaction and discovery may not adequately account for the unique experiences that comprise engagement in these two domains and, in many cases, a relevant experience may be omitted during scale development (e.g., a “discovery” measure may not have been included in the scale).
As previously mentioned, fixed-scale approaches to measuring engagement have been the rule rather than the exception in both the practitioner and academic literature (e.g., Brakus et al., 2009; Hollebeek, Glynn, and Brodie, 2014; Mollen and Wilson, 2010; Sprott, Czellar, and Spangenberg, 2009). In contrast, the current approach used more proximate, qualitatively rich, and direct measures of experiences.
Because the current research conceptualized engagement as a higher order factor that arises out of specific experiences (Calder et al., 2009; Malthouse and Calder, 2010; Mersey, Malthouse, and Calder, 2010), it used a context-specific set of direct-experience items to derive an overall measure of the degree of engagement. Subsequently, it examined the predictive power of this measure of engagement on consumption behavior, in isolation and in contrast with traditional marketing measures, such as satisfaction, that typically are measured using fixed scales.
The items used to measure experiences were developed through qualitative research (i.e., a series of exploratory interviews) and analyzed using exploratory factor analysis (EFA) and coefficient alpha. Rather than serving as a formal investigation of the engagement construct, the qualitative research allowed the authors to develop a relevant pool of items for possible use.
The items developed for measuring experiences then were subject to confirmatory analyses in connection with the three studies and the three different contexts—live jazz, newspapers, and television—discussed in this article.
Capturing Consumer Experiences
Identifying intense, qualitatively rich experiences was the basis for the measurement approach to engagement adopted in the current research. There are two alternative ways to capture experiences:
providing the real-time description of experiences as they happen (Kahneman, 2011). Though potentially valuable, the problem with this approach is that measurement itself may intrude on (or alter) the experience.
acknowledging that the remembered experience determines future behavior and that much of the real-time experience is inaccessible after the immediate experience is over (Kahneman, 2011; Kahneman, Wakker, and Sarin, 1997).
One could go so far as to postulate two selves, the experiencing self and the remembering self: “The remembering self is sometimes wrong, but it is the one that keeps score and governs what we learn from living, and it is the one that makes decisions” (Kahneman, 2011, p. 381).
The current research therefore focused on the remembered experience, measured after the immediate experience. Retrospective measures should be preferred in that they capture the accessible aspects of experiences that can affect future behavior (Kahneman, 2011). But it is also the case that retrospective measures necessarily involve beliefs about experiences rather than direct access to them (Robinson and Clore, 2002). Thus, the engagement measures used in the current article were based on beliefs about experiences.
The first step in measuring engagement was the development of a pool of items that may be potential indicators of separate experiences. The five broad experience categories mentioned earlier in this article and the specific items contained within each category were generated following a set of qualitative interviews with consumers who described their experiences consuming media and the arts in detail. Study 2, for example, involved conducting more than 300 hour-long personal interviews with newspaper readers. The current authors looked for common statements and themes and developed 275 items for a discovery-oriented survey (Malthouse, Calder, and Tamhane, 2007, p. 9).
Validation of these items was subject to subsequent EFAs and confirmatory factor analyses (CFAs). The items generated for each study were incorporated into surveys (described later in this article) and factor analyzed to determine whether they indicated separate experience categories. As will be discussed when describing the three empirical studies, the five specific-experience categories that emerged from this quantitative analysis were
The list is not meant to be a comprehensive catalog of experience categories but merely a way to organize the myriad of qualitative experiences uncovered that characterize engagement with live jazz music, newspapers, or television programming.
Although the current authors' final list of experience items was the product of their own qualitative research, many of the same broad experience categories can be located in existing typologies, including the uses and gratifications theory, which has been referenced for nearly 75 years in communications research to explain why people use media (McQuail, 1983). And although the uses and gratifications theory originally was intended to apply to communications, it has been applied to many marketing-related contexts, including engagement (Calder et al., 2009; Ko, Cho, and Roberts, 2005).
The experience categories, in fact, align closely not only with uses and gratifications theory, but with prior research on:
employee engagement (Salanova, Agut, and Peiro, 2005) and
There are, however, a few differences between the experience categories proposed in the current article and those that have been articulated in previous studies:
“Vigor” was eliminated as an experience category on the grounds that it is actually a behavioral consequence of engagement, usually represented as “the willingness to invest effort… persistence in the face of difficulties,” rather than a thought or feeling about an experience.
“Discovery,” which can be defined as “the acquisition of insight, skills, and knowledge,” was added because it is a potential experience category that seemed relevant to both media engagement and arts engagement according to the exploratory interviews.
“Civic orientation” was added as a potential experience category largely on account of the qualitative interviews conducted with newspaper readers across seven newspaper markets. Items that reflect civic-oriented experiences (e.g., “reading the newspaper makes me a better citizen”) likely will contribute to engagement in the newspaper category but not other categories (e.g., arts engagement). The approach to measuring engagement advocated in this article is flexible enough to accommodate these category-level differences.
The authors of the current article summarized their approach as follows:
Qualitative in-depth interviews generated survey items, which were included on surveys of relevant populations.
EFA identified first-order experience factors.
CFA was used to confirm measurement models for each consumption context (Calder et al., 2009).
A second-order CFA tested the overall engagement measure, where engagement is a second-order factor manifested by first-order experience factors.
In subsequent regression analyses, a simple average of the experience measures was used to estimate overall engagement.
Three studies were conducted to test this flexible approach to measuring engagement:
Study 1 demonstrated that the measure of arts engagement developed using this flexible approach was associated with different types of consumption behavior.
Studies 2 and 3 demonstrated that the proposed measures of media engagement also were associated with consumption behaviors.
Within the context of newspaper readership, Study 2 demonstrated that the explanatory power of the proposed engagement measure was greater than that of another important marketing metric (satisfaction) that is typically measured using a fixed-item approach.
Using the context of television programming, Study 3 contributed to prior evidence linking media engagement with advertising effectiveness by showing that high engagement with a television program was associated with high evaluations of embedded advertisements.
Additionally, Study 3 demonstrated the importance of developing highly context-specific experience items by showing that different experience categories provided the best indication of engagement for different types of television programs.
Study 1 proposed the following research question:
RQ1: Can a measure of engagement composed of context-specific experiences predict consumption?
This investigation developed a measure of engagement among concert attendees at a single event, the annual Chicago Jazz Festival. There were two main consumption-related dependent variables of interest:
attendees' likelihood to return to the concert the following year;
the extent to which concertgoers consume other arts-related events.
The study was done in partnership with the Jazz Institute of Chicago to collect data from concertgoers approximately six weeks after they attended the Chicago Jazz Festival. The sample included 7,020 jazz enthusiasts in the Chicago area who were on the Jazz Institute's e-mail distribution list. An invitation asked them to complete an online survey if they had attended the Jazz Festival. A total of 490 Jazz Festival attendees completed the survey, yielding a 7 percent response rate.
On the basis of the qualitative research discussed earlier, three experience categories (“Interaction,” “Discovery,” “Transportation”) were identified that comprise arts engagement in this particular context.
Questionnaire items for the three experiences included (See Table 2):
“Interaction”: “I enjoyed talking with someone else about it” and “I enjoyed going to it with family and friends.”
“Discovery”: “It gave me a broader, richer perspective” and “I learned about what kind of jazz I like best.”
“Transportation”: “I liked to imagine myself being on the stage” and “It made me think of actually playing an instrument or singing myself.”
The researchers also determined the loadings from the second-order CFA showing how the second-order engagement construct was related to the first-order experiences (See Figure 1). Across all three studies, the engagement measure was the simple average of the experiences, with each experience estimated by the simple average of the items loading on it.
One dependent variable was the likelihood of repeat consumption of the Jazz Festival in the subsequent year. Specifically, respondents were asked to rate how likely they would “attend the Chicago Jazz Festival next summer” on an unmarked scale. This question required consumers to evaluate the festival relative to other activities in which they might engage and determine their “anticipated repeat consumption.”
The other dependent variable was the total number of instances in which the respondent attended classical music concerts, art museums, live jazz concerts, and dance performances during the past year. This measure of past participation in arts events captured consumers' overall level of arts consumption and formed a unidimensional scale of “category-level consumption.”
As predicted, a regression analysis determined that the direct effect of engagement on anticipated repeat consumption was positive and significant (β = 0.25, SE = 0.12, p < 0.04). The direct effect of engagement on category-level consumption was also positive and significant (β = 0.079, SE = 0.031, p < 0.02).
These results suggested that a flexible approach to measuring engagement could generate effective indicators for key consumption metrics such as repeat and category-level consumption.
Study 2 proposed the following research question:
RQ2: Does context-specific engagement yield greater explanatory power than the fixed-scale measure of satisfaction?
This investigation tested whether a measure of engagement was a better predictor of consumption behavior than a measure of satisfaction, a traditional, fixed-scale metric often used in marketing research. Newspaper readership was selected as the context for this study because reading a newspaper—whether in print or online—may be highly experiential, making it an appropriate context for assessing both engagement and satisfaction.
The goal was to show that engagement can offer independent insights about consumption behavior that cannot be derived from traditional metrics (e.g., satisfaction).
Satisfaction has been characterized as the ultimate goal of marketing, if not all business (Converse and Huegy, 1946). Researchers have generally agreed that satisfaction is a response to an evaluation process that often occurs following consumption (Giese and Cote, 2000; Yi, 1990). Irrespective of whether the focal object has been defined narrowly (e.g., a single product attribute or feature) or broadly (e.g., the product as a whole), satisfaction judgments are both retrospective and integrative—the product of a backward-looking aggregation process on the part of the consumer.
This study measured engagement with context-specific, qualitatively rich, and goal-oriented experiences such as, “I show things in this newspaper to others in my family” (one of the items in the “Interaction” experience category).
In contrast, satisfaction was measured using items that are neither qualitatively rich nor related to personal goals. As one of the satisfaction measures, for example, participants evaluated different aspects of newspaper content (e.g., “Government: National,” “comics,” “food,” etc.) on 5-point scales, ranging from 1 (poor) to 5 (excellent).
The sample came from a large-scale survey of newspaper readers.1 After first identifying newspaper readers in 52 U.S. markets, surveys were sent to a random sample of 19,575. A total of 10,858 surveys were returned, giving a 55 percent response rate from the list of readers.
The approach adopted followed the conceptualization of engagement in Study 1. Qualitative interviews (Calder and Malthouse, 2004) determined that all five previously identified experience categories (i.e., “Interaction,” “Transportation,” “Discovery,” “Identity,” and “Civic orientation”) were appropriate indicators of newspaper engagement. There were 22 items across the five experience categories, including (See Table 3)
Interaction: “I bring up things I've read in this newspaper in conversations with others.”
Transportation: “I like to kick back and wind down with it.”
Civic orientation: “I count on this newspaper to investigate wrongdoing.”
Discovery: “This newspaper has columns that give good advice.”
Identity: “I like for other people to know that I read this newspaper.”
Overall engagement was the average of the five experiences, the same as in Study 1.
Prior research has used a wide array of self-reported satisfaction measures (e.g., Fornell et al., 1996). To enhance the generalizability of the current study's findings, three satisfaction measures were selected that comprise the main ways in which satisfaction data is typically collected:
a single-item measure (Overall Satisfaction),
a multi-item measure (Aggregate Satisfaction), and
a weighted multi-item measure that weights satisfaction responses by importance (Weighted Satisfaction).
The single-item measure (Overall Satisfaction) was the following question: “Overall, what rating would you give to this newspaper?” Responses were measured on a 5-point semantic differential scale.
The multi-item satisfaction scale (Aggregate Satisfaction) was constructed from 42 standard satisfaction questions about the content of the newspaper, such as
The question wording was, “Please rate this newspaper on each of the following kinds of content. To answer, use a 5-point rating scale, ranging from 1 (poor) to 5 (excellent). The Aggregate Satisfaction measure was the simple mean of all 42 items.
The final satisfaction measure (Weighted Satisfaction) was the importance-adjusted mean of these same 42 items. More specifically, after rating each of the 42 aspects of newspaper content, respondents were instructed as follows: “Then please indicate how important each is to you personally [on a 1–3 scale].” The researchers first calculated the product of the satisfaction rating for each aspect of the newspaper and its importance to the respondent. Weighted Satisfaction was then computed as the simple mean of the 42 products, which was then transformed to be on a 1 to 5 scale.
The dependent variable for this analysis was consumers' level of newspaper readership, measured by their reader behavior score (RBS; Calder and Malthouse, 2003). RBS quantifies a person's overall pattern of usage of the newspaper with a single numerical value.
The authors first tested a direct link between engagement and readership independent of satisfaction with hierarchical linear models, since there were two sampling stages (newspapers then respondents). Readership was found to be related to engagement, without satisfaction being included in the model. Intercept and slope for engagement had a t statistic of 37 and was thus significantly different from zero (See Table 4).
The authors next showed that engagement was related to satisfaction. The model was estimated using each of the three different measures of satisfaction. In all three cases, the effect of engagement on satisfaction was highly significant. The authors then modeled the relationship between satisfaction and RBS consumption. The effect of satisfaction on RBS consumption was significantly different from zero in all cases (See Table 4).
Finally, the relationship of RBS consumption on both engagement and satisfaction simultaneously was modeled. The significant indirect effect of satisfaction on RBS consumption suggested that satisfaction partially explained the relationship between engagement and RBS consumption noted above. In other words, engagement led to satisfaction, which led to consumption.
However, the direct effects of engagement on RBS consumption (coefficients of engagement in the model where both engagement and satisfaction were used as predictors) were all significantly different from zero. This indicates that engagement explained RBS consumption beyond satisfaction alone. These results, furthermore, suggest that weighted satisfaction and engagement were independent indicators of newspaper RBS consumption. In other words, engagement explains consumption behavior over and beyond the effects of satisfaction.
The above results were supported by a bootstrapping analysis that examined whether satisfaction mediated the relationship between engagement and RBS consumption. A significant indirect effect of engagement on RBS consumption was observed through weighted satisfaction (Indirect effect = 0.23, SE = 0.013, 95 percent confidence interval = 0.20 to 0.25), which established satisfaction as a mediator.
The direct effect of engagement on RBS consumption, however, was also positive and significant (β = 0.38, SE = 0.020, p < 0.0001), which suggests that a direct path existed from engagement to RBS consumption independent of satisfaction. Thus, taken together, the two analyses above show that engagement affects consumption both directly and by virtue of its effect on satisfaction.
Next, variable “importance” was compared using t statistics. The t statistic for engagement exceeded those for overall and aggregate satisfaction (34 versus 5.7 and 32 versus 7.7, respectively). For each satisfaction measure, a formal test of the hypothesis H0: β1 = β2 found p < 0.0001, indicating that the effect of engagement on RBS consumption was stronger than the effect of satisfaction.
Thus, irrespective of which satisfaction measure was used in the model, engagement was found to be the more powerful explanatory variable.
On the basis of the combined results of Studies 1 and 2, engagement has a significant association with consumption behavior, as measured by anticipated repeat consumption (Study 1), category-level consumption (Study 1), and the depth and frequency of consumption (RBS, Study 2):
Not only did the engagement measure used in Study 2 incrementally explain consumption beyond satisfaction measures alone, it was a superior predictor of consumption than any of the three satisfaction measures. This result further validates this paper's basic premise:
A context-specific measure of engagement, measured retrospectively in a rigorous but rich way based on specific consumer experiences, can independently and incrementally explain consumption behavior.
Study 3 proposed the following research question:
RQ3: Does a measure of engagement composed of context-specific experiences influence advertising effectiveness?
Study 3 examined media engagement with cable- and network-television programming in another country to establish that the approach generalizes beyond the United States. This study focused on cable television viewers in Mexico, demonstrating cross-cultural validity.
The authors acknowledge that conclusions from a Mexican sample do not necessarily apply in other geographical markets. They note, however, that because cable viewers in Mexico tend to be more affluent than the country's noncable viewers, they reasonably may be compared with consumers from the rest of North America and Europe.
This phase of the research program had two primary objectives:
to demonstrate that high levels of engagement not only contribute to program loyalty but also are associated with high evaluations of embedded advertisements;
to demonstrate that different television experience categories have different effects on these relevant outcome measures, highlighting the importance of using context-specific experience items.
There were two main classes of outcome measures in Study 3:
recommending the television program to a friend;
responding to advertising.
The study examined two programs and four advertisements. The first program was a “how-to” home decorating program and the other is a soap opera (telenovela):
Advertisement A was for a well-known luxury car;
Advertisement B was for a leading brand of beer;
Advertisement C was for an action movie; and
Advertisement D was for an air freshener.
The survey was executed online. Potential respondents were screened for regular viewership of the two programs, and qualifying respondents were shown a 10-minute segment of one program. Embedded in the segment was a pod with all four advertisements in random order (a pretest showed no order effects). After viewing the segment, respondents were asked a series of questions about the network, program, and four advertisements, so that each respondent rated all four advertisements but only one program.
The program stimuli were selected based on the assumption that they would create different experiences for their viewers. For example, the home-makeover program focused on a different house in each program and showed how it looked before the makeover. Professional decorators and remodelers made recommendations on how to improve the house, and the viewer was shown the transformation from beginning to end.
The program was expected to create “Discovery” experiences for the viewer, giving utilitarian ideas and insights about how to improve their own homes, and inspiring them to make changes to their own homes. The expectation was that viewers would watch the soap opera to be transported into the lives of the characters.
The sample consisted of 150 Mexican adults selected from a marketing-research panel and included 75 regular viewers of each of the two programs. The average age of respondents was 34.8 and exactly half were female. Items for the “Transportation,” “Discovery,” and “Interaction” experience categories were taken from prior research on understanding experiences with television news in the United States (Calder and Malthouse, 2008; Peck and Malthouse, 2011; Peer, Malthouse, Nesbitt, and Calder, 2007).
Media professionals from the Latin American Cable Association and the Ipsos/OTX Latin American marketing research company also provided input on the questionnaire. These efforts allowed the authors to better understand Mexican television preferences and programming.
In the context of television programming, “Civic orientation” was not deemed a likely contributor of engagement and, therefore, was excluded. Although “Identity” experiences may be relevant in determining engagement with television programming—particularly in the case of the home improvement program—the flexible-measurement model advanced in the current study did not require all possible experiences to be captured but merely a sample from the construct domain. Three experiences were measured, including the following questionnaire items (See Table 5):
Interaction: “This program comes up in conversations with many other people.”
Discovery: “This program gives me good tips and advice.”
Transportation: “This program takes my mind off of other things that are going on.”
The soap opera rated higher on all three dimensions, indicating that it was more engaging than the home-makeover program.
The first analysis examined the effect of experiences on recommending the program to a friend, and the second measured the effects on the advertisements. The dependent variable was the net promoter score question, “How likely is it that you would recommend this program to a friend or co-worker?” All three experiences were the predictors, allowing for interactions with the program.
This regression model allowed for different slopes and intercepts for the two programs (See Table 6). The slope of “Discovery” was 0.706 for the home makeover program but only 0.177 for the soap opera. The difference between the two is 0.706 – 0.177 = 0.529, which is significant (p = 0.0044). Thus, having a “Discovery” experience was a more important driver of loyalty for the home-makeover program than for the soap opera. The slope of “Transportation” was 0.298 for the home-makeover program and 0.597 for the soap opera.
The difference, 0.597 – 0.298 = 0.299, also was significantly different from 0 (p = 0.0413), indicating that “Transportation” was more strongly associated with recommending the program to a friend for the soap opera than the home-makeover program. Social interaction had a significant effect on recommendations for both programs, although the difference was not significantly different from 0, indicating that it is plausible that social interaction is equally important for the two programs.
The second analysis examined the effects of engagement with the program on reactions to the advertisements. Previous research has investigated whether engagement with a print and online advertising vehicle affects reactions to the advertisement itself (Calder et al., 2009; Malthouse and Calder, 2010; Malthouse et al., 2007), but to the current authors' knowledge, this is the first study examining carryover of television program engagement to advertising evaluations.
For each of the four advertisements, respondents were asked about their attitude toward the advertising (Aad), purchase intent, and recall of the advertisement.
The objective of understanding how each of the three advertising effectiveness measures depends on the engagement with the program was complicated by the fact that each respondent rated the four advertisements, creating four observations for each respondent. Mixed-effect models—including a random intercept for each subject—were used to account for customer heterogeneity and obtain correct standard errors. Several models with different levels of complexity were run (See Table 7).
The baseline model included different fixed intercepts for the four advertisements and two programs. There were no significant differences across advertisements, but the soap opera produced significantly higher ratings of the advertisement than the home-improvement show. More importantly, the program engagement effect was 0.070 (p < 0.0001), which indicates that the more engaged a viewer was in the program (ad vehicle), the more favorable the viewer was to the advertisement.
A similar program engagement effect was found for purchase intent but not for recall. Thus, engagement with surrounding context of an advertisement has an effect on the attitude toward an advertisement and purchase intent (See Table 7).
Several other more complex models were estimated, but they were not significant:
Allowing for interactions between advertisement and program produced results that were not significant.
Using the three experience factors in place of overall program engagement and allowing them to interact with the program resulted in nonsignificant interactions.
This analysis also had issues with multicollinearity; none of the individual variables were significant, yet including all three at once significantly improved the model.
Certainly consumers are attracted to products through hedonic pleasure. That people want and like things that give them pleasure and avoid things that do not is an age-old idea and one that marketers have used in many forms. The Greeks referred to the yearning as “hedonia.” Hedonia was distinguished from “eudaimonia,” which referred to experiencing life as being meaningful (Berridge and Kringelbach, 2011). Whereas the former has been much studied, even to the point of progress in understanding the brain mechanisms involved, the latter has received far less attention.
In the authors' view, engagement can play an important role in marketing theory by representing eudaimonia and balancing the long-standing focus on hedonia. Consider the social experience category in Study 2, where some respondents believed that talking about and sharing the content of a newspaper with others makes them more interesting and better connected to others. Or the “Civic orientation” experience, where some people believed that reading a newspaper empowers them and makes them a part of their community. Such experiences give rise to a sense of engagement in which reading the newspaper gives increased meaning to their lives.
Contrast this with how much a person likes the newspaper overall, or its Food section, or other sections of the newspaper. It is not that one construct is necessarily more important to marketers than the other. But it is key not to lose the potential importance of engagement by focusing only on the hedonic evaluation.
In other words, it is imperative for savvy marketers to measure both hedonia—by asking customers to “lean backward” and evaluate their satisfaction—and eudaimonia—by asking customers to “lean forward” and evaluate their engagement.
In this article, the authors outlined a methodology for measuring engagement that arose from the beliefs people have about their different experiences with a product. This approach examined beliefs about intense and qualitatively rich experiences that are potentially meaningful to the consumer. The authors showed that these measures of different experiences themselves are related to a higher order common factor that reflects the overall level of meaningful experience with the product. And that experience encompasses the consumer's level of engagement with it.
The authors chose to compare this engagement measure with satisfaction in Study 2 because satisfaction is the most prevalent hedonic measure used in marketing (Keiningham et al., 2015, p. 3) and typically is measured using a fixed-scale approach. Moreover, to be useful, any measure of engagement must contribute to an understanding of consumer behavior beyond considering satisfaction alone.
The current research indicated that engagement can relate to certain variables that are of interest to marketers more strongly than satisfaction, and this effect is not merely mediated by the relationship between engagement and satisfaction.
As a caveat, it is important to note that the findings reported in this article are derived from surveys, which are limited in their ability to establish causality. As previously stated, the purpose of this article was not to deny the importance of hedonic concepts and measures. Satisfaction may well be related to other variables of marketing interest more than engagement.
In the authors' view, the need exists to work with both eudaimonia and hedonia constructs and to explore their differences. But given the ubiquity of satisfaction and other hedonic metrics, the construct of engagement and its measurement via consumer experiences warrant more attention and investigation.
On a practical note, the applicability of considering both types of metrics is apparent from a consideration of a product like Facebook. In a 2014 American Customer Satisfaction Index study, Facebook received a low user-satisfaction score (67/100) relative to other e-business websites.2 This weakness was in sharp contrast to Facebook's dominance in the amount of time consumers spend with social media.
As a result, Facebook may be vulnerable to competitive threats even though it is highly engaging. Scenarios like this underscore the need to routinely examine engagement as well as satisfaction to attain the deep understanding of customers that marketers seek.
This research has a number of implications for advertisers:
The results of Study 3 indicate that media engagement has carryover benefits on advertising effectiveness, at least among Mexican cable viewers. Although this finding is consistent with research investigating engagement with print and online advertising (Calder et al., 2009; Malthouse and Calder, 2010; Malthouse et al., 2007), the authors of the current research believe this is the first to examine the carryover of television-program engagement to advertising evaluations.
Thus, advertisers would be well advised to monitor customer engagement with the specific media context in which their advertisement is embedded.
Although none of the three studies reported in the current article measured advertising engagement directly, the authors' approach may help inform advertisers who seek to measure engagement with different types of advertising (e.g., banner advertisements, television commercials, print advertisements, etc.).
An important conclusion from this research is that engagement needs to be conceptualized and measured in an appropriate way. The authors recognize that there are many benefits to one-size-fits-all measurement approaches and that the flexible approach to measuring engagement that they advocate may sometimes be prohibitive given the effort required to develop context-specific scales from qualitative research.
They have defined engagement, however, as a multi-level construct that emerges from the thoughts and feelings about one or more rich experiences involved in reaching a personal goal. Given this definition, which identifies engagement as the sum of intense, qualitatively rich experiences, the authors believe that any measure that seeks to truly measure engagement must attempt to capture these important goal-relevant experiences in a context-specific way.
ABOUT THE AUTHORS
Bobby J. Calder is Kellstadt Professor of Marketing in the Kellogg School of Management at Northwestern University. His research specialties include consumer psychology, media, advertising, and brand strategy. His work has appeared in major journals such as the Journal of Marketing Research, Journal of Consumer Research, and Journal of Consumer Psychology.
Mathew S. Isaac is an assistant professor of Marketing in the albers School of Business and economics at Seattle University. His research, which primarily examines consumer judgment and decision-making, has been published in the Journal of Marketing, Journal of Consumer Research and Journal of Consumer Psychology. He previously worked as a consultant and manager for Bain & Co. and ZS associates, where he advised media, healthcare, and private equity clients on sales, marketing, and corporate strategy.
Edward C. Malthouse is the Theodore R. and annie Laurie Sills Professor of Integrated Marketing Communications and Industrial Engineering and Management Science at Northwestern University, and the Research Director for the Spiegel Center for Digital and Database Marketing. His research interests center on engagement, media marketing, new media, integrated marketing communications, customer lifetime value models, marketing applications of big data, predictive analytics, and unsupervised learning. He was the co-editor of the Journal of Interactive Marketing from 2005 to 2011.
The authors gratefully acknowledge the Jazz Institute of Chicago, the Northwestern Media Management Center, and the Latin American Cable Association for their assistance with data collection for this article.
↵1 See http://www.readership.org/new_readers/newreaders.asp for survey details.
↵2 “ACSI: Customer Satisfaction with E-Business Rebounds as Social Media, Search Engines and News Sites Improve.” (2014). Retrieved July 30, 2015, from American Customer Satisfaction Index: https://www.theacsi.org/news-and-resources/press-releases/press-2014/press-release-e-business-2014.
- © Copyright 2015 The ARF. All rights reserved.