Featured Article / Invited Peer-reviewed Article
Consumer Satisfaction and Extension Intention with Streaming Media Services
2024 Volume 44 Issue 1 Pages 27-36
Details
2024 Volume 44 Issue 1 Pages 27-36
Previous research on subscriptions has not focused on the characteristics of content where multiple content is consumed over a certain period, and evaluations of that are not easily associated. Therefore, considering these characteristics, this study theoretically and empirically examines the effects and interactions of conditions such as the ratio of good content, the content selection criterion, and the evaluation level of original content on consumer satisfaction and intention to extend. The results show that 1) the ratio of good content, i.e., the proportion of highly rated content, is the most influential factor for satisfaction and extension intention; 2) when the main criterion of content selection is the consumers themselves rather than recommendations by the streaming service, the difference in satisfaction based on the ratio of good content is significant; 3) the influence of original content exclusively planned and produced by the streaming service is relatively lower on satisfaction compared to the influence of the ratio of highly rated content, but similar on extension intention; and 4) the influence of original content on satisfaction is as high as that of the ratio of highly rated content, only when the content selection is done based on consumer preferences.
サブスクリプションの先行研究は，複数コンテンツが一定期間に消費されそれらへの評価が連動しにくいというコンテンツの特性に注目してこなかった。そこで本論は，その特性を踏まえて，良コンテンツ比率，コンテンツ選択軸，および独自コンテンツ評価水準という条件の水準や組合せが，消費者の満足度および継続意図に及ぼす影響を理論的および実証的に吟味した。分析の結果，1）良コンテンツ比率，すなわち高く評価されたコンテンツの比率は満足度および継続意図に対して最も影響力のある要因であるということ，2）コンテンツ選択の主軸が配信サービスによるオススメではなく消費者自身であるとき，高く評価されたコンテンツの比率の満足度への影響は大きいということ，3）配信サービスが独自に企画・制作したコンテンツの影響は，高く評価されたコンテンツの比率の影響に比べて，満足度については相対的に低く，継続意図については同程度であるということ，そして，4）配信サービスが独自に企画・制作したコンテンツの満足度への影響は，コンテンツ選択の主軸が消費者自身であるときに限っては，高く評価されたコンテンツの比率に匹敵して高いということが見いだされた。
A subscription involves a service where consumers pay a fixed amount once for a certain period, such as monthly or yearly, to consume a certain quantity of goods or content the way they want during that period. This study refers to services such as Netflix and Spotify that digitally deliver videos and music, as “content delivery subscriptions” or simply “streaming services.” One noteworthy characteristic of content delivery subscriptions is that the digitization of content allows consumers to access a vast and virtually unlimited selection. Another characteristic meriting discussion is that since content such as videos and music possesses qualities of experiential goods (Nelson, 1970) and, at times, credence goods (Darby & Karni, 1973), the evaluations of different content offerings tend not to be easily associated.
However, instead of discussing subscriptions in terms of the characteristics of content and the digital delivery mentioned above, previous research has tended to offer overly general insights into subscriptions. For instance, Savary and Dhar (2020) examined consumers’ intentions to continue and to join subscription services, but they assumed that the goods involved, such as magazines, health care services, and snacks, had characteristics that were arguably the opposite of those of typical content. Furthermore, while Basu and Ng (2021) and Roy and Ortiz (2023) researched content delivery subscriptions, they focused solely on the effectiveness of a specific pricing strategy of the subscription business model (yearly vs. monthly payment), overlooking the characteristics that content is thought to have and the impact of these characteristics on consumer psychology.
Therefore, this study aims to examine the effects of levels and interactions of conditions such as the ratio of good content, the content selection criterion, and the evaluation level of original content (the extent to which consumers evaluate original or exclusive content) on consumer satisfaction and intention to extend, while focusing on the unique characteristics of content delivery subscriptions that have not been directly addressed in previous research, namely, the vast and virtually unlimited selection of accessible content and the disconnected evaluations across multiple content offerings.
The analysis of this study yielded four main findings. First, the ratio of good content, i.e., the proportion of highly rated content, is the most influential factor for satisfaction and intention to extend. Second, when the main criterion of content selection is consumers’ own preferences rather than recommendation-oriented, the satisfaction level differs considerably depending on the ratio of highly rated content. Third, the influence of content planned and produced by the streaming service is relatively low in terms of satisfaction compared to the influence of the ratio of highly rated content, but similar to it in terms of intention to extend. Fourth, the influence of content planned and produced by the streaming service on satisfaction is comparable to that of the ratio of highly rated content only when the main criterion of content selection is consumers’ self-orientation. Managerial implications for the content business can be derived from these findings.
Previous research on subscriptions has primarily focused on providing insights into the subscription business model itself, rather than considering unique factors related to selling specific goods through subscriptions. For example, Savary and Dhar (2020) found that self-concept clarity, defined as “the extent to which self-beliefs are clearly and confidently defined, internally consistent and stable” (Savary & Dhar, 2020, p. 889), influences consumers’ intentions to continue subscription services related to their identity (e.g., language learning or fitness gym memberships). They argued that goods such as magazines, healthcare services, and snacks can be divided into those that reflect an identity, such as The Economist, and neutral ones, such as People. They showed that when consumers have low self-concept clarity, they are more likely to continue using identity-related subscription services but find it difficult to subscribe to new ones that reflect their identity. While their study is noteworthy for exploring factors influencing subscription continuations and new subscription intentions, it did not consider content delivery subscriptions.
Dazai (2022) investigated the relationship between usage and satisfaction in subscriptions. Like Savary and Dhar (2020), his survey categorized services into those reflecting consumer identity and those that did not. His findings revealed that there are thresholds in the relationship between subscription usage and satisfaction, with these thresholds differing based on the service category (identity-relevant vs. not relevant). While his study considered digital content, he did not delve into the characteristics associated with content or digital streaming.
Additionally, some previous studies have examined the pricing strategy of the subscription model. Basu and Ng (2021), applying regulatory focus theory, found that compared to prevention-focused consumers, promotion-focused consumers tend to perceive amounts as larger when framed in aggregate terms (e.g., a yearly payment of 6,000 yen) than when framed in periodic terms (e.g., a monthly payment of 500 yen), even if the total amounts are the same. Roy and Ortiz (2023) also examined the effectiveness of the pricing strategy in subscriptions. They focused on perceived control, defined as a “person’s belief that he or she is capable of obtaining desired outcomes, avoiding undesired outcomes, and achieving goals” (Roy & Ortiz, 2023, p. 367), and found that consumers with low perceived control tend to choose monthly payments, with risk aversion strengthening this tendency.
These two studies regarding pricing are worthwhile as they have specified conditions for decision-making by firms on whether to present prices yearly or monthly, focusing on content delivery subscriptions. However, neither study examined the impacts of the characteristics of the content or the resulting effects on consumer psychological factors, such as satisfaction with streaming services.
As discussed above, previous subscription research has either not considered streaming services, or if it has, it has failed to fully capture their characteristics or has neglected consumer psychology. Therefore, this study aims to investigate the impact of the following three factors, which are considered to be prominent in content delivery subscriptions, on consumer satisfaction and intention to extend.
The first factor is the “ratio of good content.” If consumers frequently encounter “good content,” that is, content that they find interesting and highly rated, their satisfaction will likely be proportionally high. On the other hand, there will likely be instances where they find content uninteresting and give it a low rating. In such cases, one can easily imagine that the high satisfaction levels derived from the high ratings of some high-quality content may be slightly diminished by the low-rated content. Therefore, the “ratio of good content,” which is the proportion of favorable content to the total consumed content, may determine consumer satisfaction and intention to extend.
The second factor is the “content selection criterion.” Analyzing one’s preferences and searching for content that is likely to be interesting every time is a tedious task. In this regard, the constant recommendations provided by content delivery subscriptions play a significant role. This enables consumers to make a choice between selecting content on their own “self-axis” or based on the “recommendation-axis.” If consumers select content based on the “self-axis,” the joy may be great if it turns out to be interesting content, but conversely, the disappointment may also be significant if it is uninteresting. On the other hand, if they select content based on the “recommendation-axis,” the range of emotions may be narrower, as there is no risk of spending time choosing. This “content selection criterion” is an essential element in content delivery subscriptions that cannot be overlooked.
The third factor is the “evaluation level of original content.” The tactic of recommendations, if perceived as unique to the streaming service, can become a differentiating factor. Moreover, an even greater differentiating factor could be the existence of “original content” that can only be consumed through a specific content delivery subscription. The amount of original content is incomparably smaller than that of other content. Considering this scarcity, its impact on satisfaction may be limited. However, it is also necessary to note that the magnitude of the presence of a single piece of original content, as a differentiating factor, is considerably larger compared to other content. Considering the magnitude of its presence, even a single piece of “original content with a high evaluation level” might be equivalent to a large amount of good content and might even inhibit switching to other streaming services. Therefore, as a concept that forms a pair with the ratio of good content, it is necessary to consider the “evaluation level of original content,” rather than the “ratio of good original content.”
Consumer satisfaction is determined by comparing expectations with outcomes (Oliver, 1980, 2010). However, many consumers who lack sufficient processing capacity may find it difficult to form overall expectations for the aggregate outcomes brought about by multiple content in streaming services. Therefore, it is natural to assume that the process of forming satisfaction by comparing expectations with outcomes is carried out individually for each content item. If this is the case, then satisfaction will clearly be high for particular good content and low for particular bad content. Thus, if the ratio of favorable content is high among all consumed content, overall satisfaction with the streaming service will be high; conversely, if the ratio of favorable content is low, overall satisfaction with the streaming service will be low. Therefore, hypothesis 1a is proposed.
H1a Satisfaction with streaming services is higher when the ratio of good content is high, compared to when the ratio of good content is low.
Satisfaction and repurchase are considered to have a positive relationship (Aurier & N’Goala, 2010). This is because consumers anticipate that the service provider’s actions that bring about the desirable emotion of satisfaction will also evoke the same desirable emotion through similar actions in the next purchase (Sirdeshmukh et al., 2002). Additionally, while it is indicated that the initial favorable emotions felt by consumers after starting a subscription will eventually decline (Iyengar et al., 2022), encountering good content could prevent such a decline in satisfaction, leading to a high intention to extend. Therefore, hypothesis 1b is proposed.
H1b Intention to extend streaming services is higher when the ratio of good content is high, compared to when the ratio of good content is low.
2. Ratio of good content * content selection criterionDue to the existence of recommendations provided by content delivery subscriptions, consumers can decide whether to choose content on their own or based on recommendations. If consumers choose content based on their “self-axis,” they bear some responsibility for the success or failure of that content. On the other hand, if consumers choose content based on the “recommendation-axis,” they can avoid the risk of investing time to consider, and part of the responsibility for the success or failure of that content lies with someone other than themselves.
In the attribution theory, which discusses the differences in the attribution of causes for success or failure, selection based on one’s own preferences would be considered internal attribution, while selection based on recommendations (not self) would be considered external attribution. According to this theory, emotional reactions to a given outcome are maximized when success or failure is internally attributed and minimized when externally attributed (Reimer, 1975). Applying this to the context of subscription services, encountering favorable content might lead to increased joy due to the effort of selecting it while encountering unfavorable content might lead to increased sadness due to the sense of responsibility for choosing it. Conversely, selecting content based on recommendations may not evoke a sense of responsibility for the choice, leading to a smaller range of emotions and, therefore, less joy or sadness. Therefore, hypothesis 2a is proposed.
H2a The ratio of good content and the content selection criterion interact to influence satisfaction with streaming services. Specifically, when the content selection criterion (the type of criterion that consumers use when selecting content) is self-axis, the difference in satisfaction levels resulting from the high or low ratio of good content is large. Conversely, when the content selection criterion is the recommendation-axis, the difference in satisfaction levels resulting from the high or low ratio of good content is small.
When the dependent variable is not satisfaction but rather intention to extend, the opposite may occur. Specifically, when the main criterion of content selection is consumers themselves, the variation in intention to extend is small, whereas when the main criterion is recommendations, the variation in intention to extend is large. This is because recommendations are unique to each streaming service, and when consumers highly rate content selected based on the recommendation-axis, it means that they highly evaluate the capability of that streaming service. Therefore, hypothesis 2b is proposed.
H2b The ratio of good content and the content selection criterion interact to influence the intention to extend streaming services. Specifically, when the content selection criterion is self-axis, the difference in extension intention levels resulting from the high or low ratio of good content is small. Conversely, when the content selection criterion is the recommendation-axis, the difference in extension intention levels resulting from the high or low ratio of good content is large.
3. Evaluation level of original contentOriginal content, which is produced by specific content delivery subscriptions and can only be consumed on their streaming service, can become a differentiating factor if it successfully receives high ratings, due to its exclusivity. Although the amount of original content is small, it can be a factor that creates satisfaction through a sense of uniqueness that is specific to that streaming service. On the other hand, the ratio of good content, while not accurately calculated, can be a factor that creates satisfaction through its breadth, calculated with the total amount of content as the denominator.
In this sense, it is anticipated that the positive impact of the evaluation level of original content and the ratio of good content on satisfaction will be similar. However, it has been pointed out that satisfaction, unlike intention to extend, has an aspect of being a cumulative indicator (Ono, 2010). Here, it is considered that the impact of the ratio of good content on satisfaction will be greater than the impact of the evaluation level of original content. Moreover, if there is joy associated with selecting the content themselves, the influence of the evaluation level of original content is likely to be comparable to that of the ratio of good content. Therefore, hypotheses 3a and 4a are proposed.
H3a Satisfaction with streaming services is lower when the evaluation level of original content is high, compared to when the ratio of good content is high.
H4a There is no significant difference in satisfaction with streaming services when the content selection criterion is self-axis and the evaluation level of original content is high, compared to when the ratio of good content is high.
The claim that intention to extend is an accumulated indicator has not been made, unlike satisfaction. If that is the case, then the positive impact of the evaluation level of original content and the ratio of good content on extension intention would be similar. Furthermore, if original content is selected based on recommendations, which can be a differentiating factor, the impact of the original content evaluation on extension intention is likely to be greater than the impact of the ratio of good content. Therefore, hypotheses 3b and 4b are proposed.
H3b There is no significant difference in the intention to extend streaming services when the evaluation level of original content is high, compared to when the ratio of good content is high.
H4b Intention to extend streaming services is higher when the content selection criterion is the recommendation-axis and the evaluation level of original content is high, compared to when the ratio of good content is high.
An online survey was conducted to collect data from consumers who were currently using or had experience canceling content delivery subscriptions. There were 410 participants, 321 (78.29%) of whom were male. Among all the participants, 208 were current users. The 202 respondents who had canceled subscriptions were asked about their experiences using the service before canceling.
The design of this study was a between-subjects design with three factors at two levels each: the ratio of good content (high/low), the content selection criterion (self/recommendation), and the evaluation level of original content (high/low). Since there are many users of content delivery subscriptions, it was judged that it would be easier for participants to respond based on their current or past experiences rather than based on manipulations. Therefore, no manipulation was conducted for any of the factors, and grouping was done based on responses to the questionnaire items.
2. ProcedureAfter providing information on their gender, age, and occupation, participants were first asked about the streaming services they were currently using. Next, they were asked about the streaming services they had experienced canceling (see the Appendix for the wording of questions). For participants who were both current users and had cancellation experience, they were asked to respond about either one of the experiences. Additionally, for participants who responded about multiple streaming services, they were asked to consider the streaming service that was closest to the option “Not listed above.” This is because it is believed that people deliberate more when subscribing to less popular streaming services than when subscribing to more popular ones.
Next, participants were asked to respond to questions regarding the ratio of good content, the content selection criterion, the evaluation level of original content (Raney et al., 2003), satisfaction with streaming services (Reynolds & Beatty, 1999; Voss et al., 1998), and intention to extend streaming services (Garbarino & Johnson, 1999; Sirdeshmukh et al., 2002). Confirmatory factor analysis using the maximum likelihood estimation was conducted to calculate the factor loadings for each item, and the reliability and validity of each construct were confirmed through these loadings and reliability analysis. The mean, standard deviation (SE), and factor loadings for each questionnaire item, as well as Cronbach’s alpha coefficient, composite reliability (CR), and average variance extracted (AVE) for each construct, are shown in Table 1.
Construct measurement assessment (Mean, SE, Loading, α, CR, and AVE)
For constructs other than the content selection criterion, the average values of the responses were calculated for each participant. Regarding the ratio of good content, participants were divided into two groups, a high group and a low group, based on the median.
Since the data did not satisfy the assumption of normality (H1a: W=0.976, p<0.001. H1b: W=0.911, p<0.001), the Mann‒Whitney U test, a non-parametric alternative to the independent t-test, was conducted for H1a and H1b. The results are shown in Table 2. The mean rank of satisfaction when the ratio of good content was high (277.92) was significantly higher than the mean rank of satisfaction when it was low (131.65) (Z=−12.526, p<0.001). Therefore, H1a could be supported. Similarly, the mean rank of extension intention when the ratio of good content was high (262.17) was significantly higher than when it was low (147.72) (Z=−9.880, p<0.001). Therefore, H1b could also be supported.
Results of H1a and H1b (Mann‒Whitney U test)
For the content selection criterion, the values of participants’ responses to the first item (CSC1: Find videos based on your own preferences at the time, rather than recommendations from [selected streaming service].) were compared with the average values of participants’ responses to the second item (CSC2: Choose videos to watch from recommendations by [selected streaming service].) and the third item (CSC3: Find videos to watch based on the recommended videos from [selected streaming service].). Participants were then divided into two groups: those who selected based on the self-axis if the former exceeded the latter, and those who selected based on the recommendation-axis if the reverse was true.
The results of the two-way ANOVA and subsequent multiple comparison tests for H2a are shown in Figure 1. In the two-way ANOVA, the interaction between the ratio of good content and the content selection criterion was significant (F=4.302, p<0.05). Subsequent multiple comparison tests confirmed that this interaction was a synergistic effect rather than a counteracting effect. Specifically, the mean of satisfaction when the ratio of good content was high (5.413 and 4.905) was significantly higher than the mean of satisfaction when the ratio of good content was low (3.881 and 4.086) at both levels (Figure 1 left). Therefore, H2a could be considered supported.
Results of H2a (two-way ANOVA and multiple comparison tests)
The results of the two-way ANOVA and subsequent multiple comparison tests for H2b are shown in Figure 2. In the two-way ANOVA, the interaction between the ratio of good content and the content selection criterion was not significant (F=0.583, p>0.05). Therefore, H2b could be considered unsupported.
Results of H2b (two-way ANOVA and multiple comparison test)
For the evaluation level of original content, participants were divided into two groups based on the median: a high-evaluation group and a low-evaluation group. For H3a and 4a, since the data satisfied the assumption of normality (H3a: W=0.983, p>0.05. H4a: W=0.968, p>0.05), independent t-tests were conducted. The results are shown in Table 3. Satisfaction when the evaluation level of original content was high (4.711) was significantly lower than satisfaction when the ratio of good content was high (5.209) (t=−2.691, p<0.01). Therefore, H3a could be supported. Similarly, there was no significant difference between satisfaction when the content selection criterion was the self-axis and the evaluation level of original content was high (4.703) and satisfaction when the ratio of good content was high (4.572) (t=0.538, p>0.05). Thus, H4a could also be supported.
Results of H3a and H4a (independent t-tests)
For H3b and 4b, since the data did not satisfy the assumption of normality (H3b: W=0.937, p<0.001. H4b: W=0.936, p<0.01), Mann‒Whitney U tests were conducted. The results are shown in Table 4. The mean rank of extension intention when the evaluation level of original content was high (45.400) did not differ significantly from the mean rank of extension intention when the ratio of good content was high (36.870) (Z=−1.621, p>0.05). Therefore, H3b could be considered supported. Similarly, there was no significant difference between the mean rank of extension intention when the content selection criterion was the recommendation-axis and the evaluation level of exclusive content was high (31.500) and the mean rank of extension intention when the ratio of good content was high (30.850) (Z=−0.122, p>0.05). Thus, H4b could not be supported.
Results of H3b and H4b (Mann‒Whitney U test)
Content delivery subscriptions have rapidly expanded their market size and attracted attention from both academic and practical perspectives. However, previous studies have not focused on analyzing this area. Therefore, this study is considered significant in that it analyzed content delivery subscriptions and, for the first time in subscription research, considered antecedents of satisfaction. In addition, by focusing on the characteristics of content and digital delivery, we identified the ratio of good content, the content selection criterion, and the evaluation level of original content as conditions influencing satisfaction and intention to extend streaming services.
Our study produced the following four main findings. First, the ratio of good content, i.e., the proportion of highly rated content, is the most influential factor on satisfaction and extension intention. Streaming service providers should focus on ensuring that each consumer encounters content they find interesting. They need to make accurate and continuous content recommendations to consumers, such as “If you think this is good, you should like this too.”
Second, when the main criterion of content selection is the consumers themselves rather than recommendations by the streaming service, the difference in satisfaction based on the ratio of good content is significant. Streaming service providers need to identify how consumers choose certain content. If content selected away from recommendations is judged as boring, companies should be concerned about consumer churn. On the other hand, consumers who frequently select content from the recommendation lists may become desirable supporters of subscription services.
Third, the influence of content exclusively planned and produced by the streaming service is relatively low on satisfaction compared to that of the ratio of good content, and similar on extension intention. Since the high rating of exclusive content can encourage consumer intention to extend, service providers should produce exclusive content as much as possible. However, to truly gain support through high satisfaction, there is no better tactic than to successfully match general content (not exclusive content) with consumers.
Fourth, the influence of original content on satisfaction is as high as that of the ratio of good content alone, only when the content selection is done based on consumer preferences. When exclusive content is chosen, the process of selection is also important. Streaming service firms need to understand this point well and enhance not only the essence of exclusive content but also its superficial appeal. If consumers willingly search for original content from among a large amount of content, the subscription service might gain a solid position.
2. Limitations and future researchThe following limitations of this study should be noted. First, the refinement of the items for measurement. Because several hypotheses included new concepts for which there were no previous studies to refer to, many items were created independently. Second, the scale of the survey. Few participants selected content based on recommendations of the streaming service, which might have resulted in hypotheses 2b and 4b being unsupported. It is expected that these limitations will be overcome by collecting larger data and expanding the survey.
Regarding future research directions, first, since there is room for refinement in the measurement of each construct, it may be possible to conduct research aimed at scale development. Second, it may be possible to precisely manipulate conditions such as the ratio of good content and the content selection criterion through laboratory experiments. In particular, interesting insights may be obtained regarding the behavior of consumers who perceive themselves as having found content on their own despite selecting content based on recommendations. Third, defection behavior could be measured and analyzed instead of extension intention. In subscription services, defection is an active behavior, contrary to the usual purchasing behavior, whereas continuation is a passive behavior. By identifying asymmetric factors for continuation and defection, it may be possible to gain insights into these two behaviors.
By continuing to develop theories based on these prospects, it may be possible to make further contributions to subscription research and content business research in the future.
Our sincere thanks go to Professor Akinori Ono (Keio University) for helpful discussions and valuable support in enhancing the quality of this work. We thank Claude, an AI assistant created by Anthropic, for refining the language of our paper.
Items
Takahiro Chiba
Takahiro Chiba is a professor of marketing at the Faculty of Business and Commerce, Kansai University. His research interests include service marketing and customer satisfaction.
Ryohei Kitazawa
Ryohei Kitazawa is a doctoral student at the Graduate School of Business and Commerce, Keio University. His research interests lie in the area of consumer behavior.
The datasets generated and analyzed in this article are openly available in J-STAGE Data.
