Direct and Indirect Tourism Online Channels. Do They Have a Different Potential for Customer Loyalty?

Abstract

To manage their competitive goals, e-tourism service companies, in direct and in indirect channels respectively, need to know the antecedents of customer loyalty. Customer loyalty, generated by satisfaction in its various forms, is the cornerstone of the company’s assets and financial sustainability. Current literature does not provide a comparative analysis on this issue. To fill this gap, this research presents a model that includes customer satisfaction and participation as the main drivers of customer loyalty. The empirical research relies on one survey conducted by a market research company addressed to Internet users in Spain with experience in online purchases of tourism products. The estimation method is 3SLS (Three-Stage Least Squares), a simultaneous equations model applied to the database obtained. The results reveal a different potential of the two types of e-channels in producing higher levels of loyalty through customer participation. Increasing the participation of customers in indirect e-tourism channels results in higher returns on loyalty while the impact is lower in the direct channels. These findings are especially interesting for tourism service providers.

1. Introduction

In the tourism industry, the final service providers (hotels, restaurants, airlines, rail transportation, car rental companies, etc.) have historically relied on their own direct channels to reach potential customers for booking their services. These channels, walk-ins, telephone, and other telematics (fax, e-mails) made the most of direct bookings [1]. In the last decade, almost all tourism companies in the sector—final service providers and retail intermediaries—have developed their own technological means (websites, apps) not only to foster direct bookings adding this online channel, but more than that, as the basic instrument for customer management and loyalty. In parallel, service providers have incorporated indirect online channels to complement the direct service through online travel agencies, OTAs (Online Tourism Agencies) [2,3,4,5]. The growth of these two online channels has caused most of the hospitality service providers are actually involved in an e-multichannel operation combining both direct and indirect channels.

In this e-multichannel scenario, it becomes relevant for both tourism service providers and OTAs to have a clear view on their potential to build up a loyal base of customers. Customer loyalty, generated by satisfaction in its various forms, is the cornerstone of the company’s assets and financial sustainability. In a recent literature review about the information technology and tourism management, Navío-Marco et al. [4] suggest that customer loyalty will become one of the key questions with regard to tourism agents’ strategies. Despite the acknowledged relevance of customer loyalty for tourism companies [6], the research on this issue shows some gaps. One major one is that most research exploring the antecedents of customer loyalty in tourism is focused on the product—i.e., the actual services delivered by providers and experienced by consumers at destination—[7,8,9,10,11]. Much less attention is focused on the purchase or booking process. Certainly, there are varied studies tackling the issue of customer loyalty and its antecedents in the online booking process [12]. Some of them are focused on direct channels—websites of service providers—[13,14], and others focused on OTAs in indirect channels [2]. All of them give interesting findings, however they do not provide a comparative framework to analyze the potential differences between direct and indirect e-channel alternatives to achieve different levels of customer retention.

With the aim to fill this gap, the research objective of this paper is to identify differences in the potential of direct and indirect online distribution channels to build customer loyalty through their booking processes. Discovering a different potential to retain customers by the tourism agents, service providers and OTAs, would contribute to a better understanding on how to manage the booking processes pursuing customer loyalty by improving customer engagement.

The paper is structured in sections. Section 2 presents a discussion of prior literature about customer loyalty and its antecedents based on the theoretical ground of the distribution channels studies [15]. Section 3 proposes the research model resulting from the previous discussion. Section 4 describes the materials and methods. Section 5 addresses the estimation of the model and the discussion of results. Finally, Section 6 provides a discussion with the synthesis of the findings in terms of conclusions, managerial implications, and suggestions for future research.

2. Customer Loyalty in Tourism e-Channels

When consumers are asked about their intentions to revisit or repurchase at one website, their answer is based on their past experience of consumers with service providers or OTAs in the booking process. The models explaining the e-loyalty should capture the relevant components of this experience. The literature on this topic provides theoretical frames and empirical models that have been used in the research using different components of the experience as antecedents of intentional loyalty. Some of them are centered on one particular feature of online shopping: the e-trust derived from the separation of customers from tourism agents in the purchase process [16,17]. More frequently, the antecedents of e-loyalty are identified with the service–profit chain: the quality–satisfaction–loyalty links [9,18]. In a recent review, Sánchez-Rebull et al. [19] show that, altogether or separately, these links have been widely analyzed in the tourism research ground. This research strand, however, does not consider the participation of customers in the booking process. With the exception of the study about offline travel agencies by Grisseman and Stokburger-Sauer [20] and Berne et al. [21], prior research in tourism does not provide an analytical frame of the booking process taking into account the co-production of distribution services and the transaction costs associated to customer participation. Thus, it is important to extend the prior literature including the participation of the customer, specifically in the booking process, considering this as the set of distribution services that are used discretionally by consumers to complete the purchase [22].

Regarding direct and indirect e-channels, the potential of tourism companies to foster customer loyalty has been studied. The own direct e-channels have been mainly studied in the airline online booking market, where some determinants of loyalty are identified, as the hedonic and functional (efficiency) quality through the perceived value of the online flights booking [13]. In this context, it has been argued that when the customer base of the service providers is characterized by a high price-loyalty elasticity, service providers could be more inclined to put aside OTAs due the provider can control the prices offered [23]. Direct e-channels should therefore be more effective for tourism service providers that follow a retention strategy based on prices. Differently, in the hospitality sector, there are other factors, mostly related with the customer acquisition, suggesting a more favorable predisposition to combine the direct and indirect e-channel options through OTAs [1,3,24]. Some studies have found that the number of online bookings in the hotels is a result of the number of channels used, particularly for non-categorized establishments [3]. In this direction, other studies discuss for hospitality businesses the advantages of supplementing the great reach of indirect channels for selling bookings with the superior potential of their web pages in offering incentives and customized services to engage and retain customers [24]. This would explain that most of the hospitality service providers are deploying multichannel marketing strategies combining different types of direct and indirect channels [1].

Regardless the sector, the multichannel operation aims to capture the advantages of direct and indirect channels to acquire and retain customers. Then, it becomes relevant for tourism service providers to have a clear view on the potential of both types of channels to build up a loyal base of customers.

Nonetheless, since both direct and indirect online channels rely on the same technological base and they are involved in a similar innovation dynamic, these factors are not expected to have different impacts in either channel as antecedents of loyalty. Accordingly, differences in loyalty, if any, must come from the consumer’s experience with service providers and OTAs in the past online booking.

The approach of this study thus departs from the previous ones because it considers booking as the outcome of a process in which the output of tourism providers and OTAs is the set of distribution services that are used discretionally by consumers to complete the booking. While the distribution services might be obtained directly from last providers (direct channel), the reason of being of OTAs (indirect channels) is that they provide an added value, which might be appreciated by consumers, across and along varied purchase incidences. This essential differential factor should bring differences to the consumer contribution and participation in the completion of the booking process, and eventually to the capacity to generate customer loyalty.

In this way, in the service–profit chain framework, this paper considers customer inertia, customer satisfaction with the last purchase, and consumer participation in the booking process, as main antecedents of attitudinal loyalty. Hereafter, a model is proposed in order to explain customer loyalty depending on the type of online tourism distribution channel.

2.1. Antecedents of Intentional Loyalty

The link between satisfaction and intentional loyalty in distribution channels has been extensively researched in a wide range of empirical contexts. Stemming from this extensive research, the review of Kumar et al. [25] provides evidence to conclude one empirical generalization: overall, there is a positive relationship between customer satisfaction and loyalty intentions. This positive relationship is enhanced in online channels since: “… due to the higher competition exacerbated by the Internet and the customers’ empowerment, satisfaction acquires much more importance in affecting loyalty online than offline” [25] (p. 250). This relationship has been confirmed in tourism online channels [26,27].

Moreover, the participation of customers in the purchase process has been suggested as a main source for creating and enhancing loyalty to the brand or the company. It happens in two paths, the personalization of the service and the economic benefits from productivity gains passed to the customer. First, the effect of personalization on loyalty comes not only from increases in the bundle of benefits to the customer; it also raises the costs of switching, and thereby loyalty [28]. Second, when properly designed and executed, the customer participation in the co-production of the service may provide productivity gains for the company that, seeking for competitive advantages, are passed in the form of lower prices to customers. In this context, co-production has an important role in building customer loyalty [21,29]

The inertia, as an antecedent of loyalty intentions, is the effect of past experience on the future usage intentions [30], that is, the purchase repetition based on habit [31]. On the one hand, it is needed to separate this conceptualization of inertia from the channel loyalty based on the perceived service superiority. Inertia might thus appear as a specific type of loyalty but, unlike traditional notions of loyalty, channel experience does not require superior channel performance on specific channel attributes. On the other hand, previous research is not conclusive: although it has been found that it has a strong effect on future usage intentions [32], some recent research in customer loyalty in OTAs does not support this positive direct effect [31].

2.1.1. Components of Customer Satisfaction

When end consumers purchase goods and services, they request a set of distribution services at different levels. In these terms, the deliberate output of a distribution channel at the retail level is to meet the consumers’ demands providing the right services at the right levels. The retail services are purposely offered so that they can be used at the best convenience of end-users in the purchase process. There are different enumerations of them and are usually referred to as a part of the channel attributes. The taxonomy from Betancourt [15] includes ambience, assortment, accessibility, and assurance of product delivery in form and time, and information. The ambience service is the design, organization, flow, and amenities provided by the web page or application. The level of assortment offered encompasses the width—the number of different categories of services—and depth—the number of alternatives within one category—dimensions. The level of accessibility is the convenience in finding and accessing the web page. The assurance of delivery of the product includes two dimensions, time and form; the level of assurance of delivery reflects the compliance of the customer expected reception time and product conditions. The level of information provided by the web page to the consumer reflects the quantity and adequacy of useful information on products, prices, availability, and other characteristics. This taxonomy covers all the elements that the consumer can appreciate when making a purchase and this makes it appropriate for its application to the analysis of any channel. It should be added that these services are supported by the so-called channel policies (privacy policies, means of payment, return, security, prices) that, unlike distribution services, are fixed stably over time and for the generality of clients. In tourism channels, these services have been approached and studied individually or in different combinations [21,33,34].

Betancourt et al. [35], in an online purchase process context, consider the same taxonomy of distribution services, adding two interesting features. The first one is that it captures directly the co-production elements leading to the effective purchase and delivery of the final product; distribution services are viewed as outputs of channel firms and as fixed inputs into the household production functions of consumers [36]. The second feature is that the customer evaluation of the mix of the different levels of services distribution provided to the completion of the transaction is equivalent to the concept of service quality. With this perspective, distribution services are the main mechanism through which retailers can influence customer satisfaction with a transaction.

Together with the provision of distribution services at a fixed level in the short term, companies must back up their offer of distribution services with policies that support the co-production process. The most common policies are those concerning privacy, safety, payment modes, product returns, and complaint handling issues. Privacy and safety policies have been found to be particularly relevant for consumers at the time of choosing online travel agencies [37]. In this supporting role, privacy/safety and payment channel policies have been modelled as a moderator of the satisfaction influence on loyalty [35]. In hotel websites, security and privacy have been found to be “… influential factors to develop customers’ perceived flow, resulting in customer satisfaction and purchase intentions” [38] (p. 224). However, approaching the profit–service chain from the perspective of the participation of customers, channel policies are intimately related to both the co-production process leading to satisfaction and the transaction costs, which are also a part of the participation in the purchase process.

In the context of tourism online distribution channels, the study of customer satisfaction about the booking process in OTAs and service providers has not deserved too much attention. Only a few studies propose an empirical model with a selection of variables as antecedents of customer satisfaction [39,40,41]. These studies use a similar set of basic antecedents, although they are grouped in different ways as components of different concepts.

Based on this discussion, satisfaction is composed by online tourism distribution services and channel policies.

2.1.2. Components of the Customer Participation

On the one hand, the transaction costs have been modelled in online shopping as antecedents of satisfaction [12]. The online channels open new interactive ways of participation of consumers in the distribution processes of the tourist services. The tourist consumer is not subject to an exchange in costs prefixed by the seller but can act on the level of costs incurred in the search and booking tasks. In this direction, the consumers may pursue higher levels of participation in exchange of lower prices. Depending on their perception of transaction costs implied in each purchase process, they might decide on higher or lower participation levels if there is a counterbalance in the prices [29].

To identify the different components of the transaction costs in the multichannel context, this research will rely on the components provided by Verhoef et al. [42]: search and purchase costs. During the booking process, consumers first fall into the search costs of visiting other alternative websites prior to the effective booking in one company of either a direct or an indirect channel. In doing so, they look for information as a risk reduction strategy to protect themselves and to maximize their satisfaction [43]. On the purchase costs side, and according Verhoef et al. [42], the basic ingredients are the effort and time employed, the perceived purchase risk, and the price level. In the current study context, the participation of customers in the booking process requires both time (to complete the booking) and the effort involved in executing the tasks. Both the time spent in the company’s website and the effort undertaken to complete the booking have a positive relation with the customer participation in the purchase process.

Purchase risks are those related to the customer’s uncertainty concerns about payment issues, and lack of privacy. To overcome the uncertainty, companies design channel policies about privacy and payment modes for shaping the assurance of the exchange in different ways. The concern about privacy is a relevant purchase cost that makes consumers reluctant to interact with companies in the purchasing processes. The commitment of companies with convincing privacy policies should have a positive impact on customer participation. In an interesting paper, Lee and Cranage [44] explore the interaction effect of the consumer perceived privacy assurance with the personalization level on the willingness to adopt the website’s services. In the same direction, customers perceive transaction costs associated to the different payment modes available. An adequate array of modes would let customers to choose the least costly alternative and they have been found influential in the consumers’ choice of on-line channels in tourism [45].

Finally, the most evident purchase cost is the monetary cost measured as the price level perceived compared to other online services. The price paid should reflect the opportunity cost of participation fairly, as found in tourism research; when customers perceive price justice, their intentions of revisit increase [46]. In the same direction, when booking was enforced to eliminate best price clauses in their contracts with hotel companies, losing its position of price leadership, the OTA company decided to add valuable services to the customers [47].

3. Model of Research

Table 1. Correspondence of transaction costs between Verhoef et al. [42] and this study.

Transaction Costs	Verhoef et al. [42]	In This Study
Search costs	Search Effort: The perceived time required (time costs) and perceived difficulty for consumers to gather product and service information	Search Effort: Search frequency in alternative websites prior to purchase through this company’s website
Purchase costs	Purchase Effort: The perceived difficulty and time costs that consumers experience when purchasing a product through a specific channel.	Purchase Effort: The consumers perceived effort in the purchase processTime: time taken to complete the purchase on the website
	Purchase Risk: The perceived uncertainty in buying products through a specific channel, due to things such as payment issues, and lack of privacy	Purchase Risk: The consumer’s perception of a channel’s payment modes and privacy policies
	Price Level: Consumers’ perceptions of prices on a specific channel	Consumers’ perceptions of prices compared to other online alternatives

includes the transaction costs by Verhoef et al. [42], and their adaptation to the context of this study. Participation is considered as composed of perceived search effort and purchase costs, the latter including purchase effort, time, the purchase risks, and price level. Purchase risks—privacy/safety and payment modes—are considered also as components of participation.

Consequently with the previous exposition, the empirical model is proposed as shown in Figure 1. According to the empirical model, there is a set of three linear equations with three endogenous variables: Loyalty, Participation, and Satisfaction. The exogenous variables are Distribution Services, Purchase Risks, Transaction Costs, and Inertia.

Equation (1): Satisfaction = Distribution Services, Purchase Risks, (Channel Type):

$$\begin{array}{ll}{\mathrm{SAT}}_{\mathrm{i}}=& {\mathsf{\alpha }}_0+{\mathsf{\beta }}_1{\mathrm{INFO}}_{\mathrm{i}}+{\mathsf{\beta }}_2{\mathrm{AMB}}_{\mathrm{i}}+{\mathsf{\beta }}_3{\mathrm{ACCESS}}_{\mathrm{i}}+{\mathsf{\beta }}_4{\mathrm{ASSORT}}_{\mathrm{i}}+{\mathsf{\beta }}_5\mathrm{DEL}\_{\mathrm{TIME}}_{\mathrm{i}}\\ & +{\mathsf{\beta }}_6\mathrm{DEL}\_{\mathrm{FORM}}_{\mathrm{i}}+{\mathsf{\beta }}_7{\mathrm{PRIVACY}}_{\mathrm{i}}+{\mathsf{\beta }}_8{\mathrm{PAYMENT}}_{\mathrm{i}}\\ & +\left({\mathsf{\beta }}_9{\mathrm{CHANNEL}}_{\mathrm{i}}\right)+{\mathsf{\epsilon }}_{\mathrm{i}}\end{array}$$

(1)

Equation (2): Participation = Purchase Risks, Transaction Costs, (Channel Type):

$$\begin{gathered} {\mathrm{PART}}_{\mathrm{i}}={\mathsf{\alpha }}_0+{\mathsf{\beta }}_1{\mathrm{PRIVACY}}_{\mathrm{i}}+{\mathsf{\beta }}_2{\mathrm{PAYMENT}}_{\mathrm{i}}+{\mathsf{\beta }}_3{\mathrm{TIME}}_{\mathrm{i}}+{\mathsf{\beta }}_4{\mathrm{EFFORT}}_{\mathrm{i}} \\ +{\mathsf{\beta }}_5{\mathrm{SEARCH}}_{\mathrm{i}}+{\mathsf{\beta }}_6{\mathrm{PRICE}}_{\mathrm{i}}+\left({\mathsf{\beta }}_7{\mathrm{CHANNEL}}_{\mathrm{i}}\right)+{\mathsf{\epsilon }}_{\mathrm{i}} \end{gathered}$$

(2)

Equation (3): Loyalty = Satisfaction, Participation, Inertia, (Channel Type)

$${\mathrm{LOYAL}}_{\mathrm{i}}={\mathsf{\alpha }}_0+{\mathsf{\beta }}_1{\mathrm{SAT}}_{\mathrm{i}}+{\mathsf{\beta }}_2{\mathrm{PART}}_{\mathrm{i}}+{\mathsf{\beta }}_3{\mathrm{INERTIA}}_{\mathrm{i}}+\left({\mathsf{\beta }}_4{\mathrm{CHANNEL}}_{\mathrm{i}}\right)+{\mathsf{\epsilon }}_{\mathrm{i}}$$

(3)

The three equations are structural equations as they stem from the theories and findings discussed in the previous section, and each of them tries to reflect one of the basic components of the service–profit chain. The equations are interdependent: Loyalty is determined by Participation and Satisfaction, and the latter shares some exogenous variables (Purchase Risks: Privacy, Payment) with Participation. As such, the set of equations has to be treated as a simultaneous equation model, in which all the equations are estimated jointly.

To tackle the research question, whether the channel type has an impact or not on the attainment of customer loyalty, we estimate the model with and without including the channel type variable (direct channel, CHANNEL = 0; indirect channel, CHANNEL = 1) in the three equations.

4. Materials and Methods

In a way very similar to that of countries with a relevant tourist structure, the distribution channels that bring tourist services closer to final consumers in Spain have evolved following the guidelines imposed by the development and implementation of Information and Communication Technologies (ICT) [48,49]. The most relevant consequence in the tourism distribution channels is the dominant position of online channels supported by the widespread use of the Internet by consumers in their purchasing processes. In this area, the empirical framework of the research is limited to Internet users in Spain with past experience in online purchasing of tourist products.

Consequently, the database for the empirical analysis is aimed at the segment of those consumers who, being Internet users, have made at least one purchase of tourism products in the last year. To target this segment and to gather the data, a market research company with a panel of Internet users was hired. Within this panel, a total of 445 users answered the questionnaire. To ensure the representativeness of the sample, a quota sampling was conducted to reflect the proportion of the major characteristics of the digital consumer population purchasing tourism services online (

Table 2. Sample demographics (%).

Gender	Male	51.4
	Female	48.6
Age (years)	18–30	27.8
	31–55	48
	Above 55	24.2
Education Level	Elementary	1.0
	Lower-level secondary	15.2
	Higher-level secondary	21
	Uncompleted university	10.8
	Bachelor’s degree	42.5
	Postgraduate degree	9.4

). As such, it is not a probabilistic sampling technique. For the specific research objectives, only 367 questionnaires were eligible (a confidence interval of 5.12 for a confidence level of 95%). Firstly, those in which the respondents did not know how to correctly identify the name of the company where they had made the last purchase and whether it was an intermediary or a service provider were rejected. Moreover, those questionnaires that did not have an answer in any of the variables of the empirical model were eliminated.

Gender is balanced and age and education level reflect the demographic bias of online tourism customers: a lower percentage of elders and a higher education level. More than 75% of respondents are in the range of millennial people, which have been shown as the digital tourist more loyal through the influence of their satisfaction level with hotel bookings [50].

The questionnaire was structured in three blocks of questions: demographic variables, online shopping behavior, the evaluation of the retail services provided in the last purchase, and the evaluation of the last purchasing experience. As it can be checked in the literature review, every variable included in the model has been used and measured previously in different studies and contexts. They are measured through a direct question and the model does not include measurement models.

The endogenous variables customer satisfaction level, participation, and loyalty and their antecedents were presented in Likert-scales from 0 to 10 points adapting the polar statements to the issue, like “from completely inadequate (0) to completely adequate (10); much more cheap (10) much more expensive (0); or completely satisfied (10) completely unsatisfied (0)”. This scale of eleven points has statistical advantages over others with 7 or 5 intervals and is more convenient for the respondents as they are familiar with this scale for a wide variety of evaluations and judgements. The only exception is the variable “Time employed” that is measured through a scale with four intervals: less than 30 min (1), 30–60 min (2), 60–90 min (3), and more than 90 min (4).

Since the explanatory variables Satisfaction and Participation (Equation (3)) are dependent variables from other equations in the system, the model must be estimated using simultaneous equations methodology [51]. The specific estimation method used, 3SLS (3 simultaneous least squares), is recommended to tackle the endogeneity bias problem, commonly present in data from surveys in distribution channels [52]. This happens when independent variables are correlated with the error term. In the context of distribution channels in tourism, 3SLS has been used to analyse the impact of ICTs on the power balance of the channel [53].

5. Results

Table 3. Descriptive statistics of the variables.

Variable	Mean	Standard Deviation	Scale
EXOGENEOUS VARIABLES
Distribution Services			0–10
Information	7.90	1.41
Ambiance	7.54	1.40
Accessibility	8.17	1.59
Assortment	7.61	1.69
Delivery in time	7.82	1.65
Delivery in form	7.77	1.66
Channel Policies			0–10
Privacy	7.81	1.66
Payment modes	7.73	1.75
Transaction Costs
Previous search in other sites (Research Shopping)	7.51	2.03	0–10
Time employed	1.72	0.75	1–4
Effort in the purchase	7.48	1.91	0–10
Price compared to other online alternatives	7.19	1.60	0–10
Inertia
Past patronage	5.87	2.57	0–10
ENDOGENEOUS VARIABLES
Satisfaction
Satisfaction with last purchase	7.86	1.69	0–10
Loyalty
Repurchase intentions	7.89	1.73	0–10
Participation
Participation in the purchase process	7.32	1.64	0–10

presents the descriptive statistics of the variables average, standard deviation and type of scale. Skewness and Kurtosis coefficients showed moderate levels indicating the data of our study are close to the univariate normal distribution [54].

Hereafter, the results of the model estimation and the effect of including channel choice (

Table 4. Loyalty model estimation with and without channel type.

	3SLS with Channel Type		3SLS without Channel Type
Loyalty
Constant	0.57		0.27
Satisfaction	0.53 ***		0.51 ***
Participation	0.45 ***		0.48 ***
Inertia	0.02		0.02
Channel (indirect)	−0.29
Adjusted R2	0.47		0.46
	Satisfaction
	Constant	−0.56 **		−0.47 *
	Information	0.09		0.08
	Ambience	0.18 ***		0.19 ***
	Accessibility	0.04		0.04
	Assortment	0.04		0.05
	Del_Time	0.30 ***		0.29 ***
	Del_Form	0.33 ***		0.33 ***
	Privacy	0.13 ***		0.14 ***
	Payment	−0.04		−0.04
	Channel (indirect)	0.13
	Adjusted R2	0.76		0.76
	Participation
	Constant	0.95 **		1.20 ***
	Privacy	0.15 ***		0.16 ***
	Payment	0.18 ***		0.18 ***
	Time costs	−0.04		−0.02
	Effort costs	0.08 **		0.09 **
	Price level	0.32 ***		0.32 ***
	Search costs	0.08 **		0.08 **
	Channel (indirect)	0.39 **
	Adjusted R2	0.37		0.36

*** Significance level < 0.01; ** Significance level < 0.05; * Significance level < 0.1.

) are discussed.

5.1. Covariates of Satisfaction (Equation (1))

The variance of customer satisfaction is explained to a great extent by the covariates (adjusted R² of 0.76). Concerning channel policies, only privacy has a significant contribution to customer satisfaction (0.14). This result confirms the preeminent role of the distribution services in the formation of satisfaction. These results apply to direct and indirect online channels, the inclusion of channel type choice does not bring any change. The adjusted R² value is the same (0.76), and the individual coefficients do not show significant changes. Customer satisfaction is equally determined by the customers’ demand of services and the channel policies in the two types of channels. Once the channel choice is made, consumers have adapted their demand of services to their purchase task and expectations, and the channel type does not bring any further effect on the satisfaction level.

A more detailed view shows that information, accessibility, and assortment services do not have a significant effect in either online channel, being ambiance and assurance of product delivery, in time and form, the services that play a significant and relevant role. Ambiance includes functional utility elements (clear identification of products, and neat definition of products) and hedonic utility (the appeal of the web page design). This is important because there is an influencing role of these aspects in the outcome of the co-production process: the customer satisfaction with the purchase. The assurance of delivery in the desired time of the transaction ending with the booking confirmation contributes strongly to customer satisfaction (0.29). This could be a somewhat unexpected large impact since confirmation procedures are simple and very similar across tourism companies. However, it shows that completing in time is a critical part of the service as is the culmination of the purchase process, and any performance below the expected standard would have a strong negative on the satisfaction level. For example, errors in the confirmation that lead to a time-costly repetition of the purchasing process with the uncertainty of getting the product (i.e., the same of seat, the same access to a performance, the same room …). The most influential service is the assurance of delivery at the expected form (0.33). The purchase is completed at the distribution channel with the reservation -the right to enjoy a tourism service sometime after the purchase-, but consumption takes place in a different time and space. In tourism services, consumers perceive high risks derived from product variability and changing conditions during the lapse time between reservation and consumption.

5.2. Covariates of Consumer Participation (Equation (2))

The variance of customer participation is explained to a relevant extent by its covariates: channel policies and transaction costs (adjusted R² of 0.36). Channel policies—privacy and payments—have a higher positive effect on participation (0.15 and 0.18, respectively). This result remarks the importance of the positive role of policies in supporting the co-production process. Only one of the transaction costs covariates—time spent in the web page—does not have a significant effect on the dependent variable. Transaction costs in terms of search and effort have significant coefficients (both values of 0.08) indicating that they are positively and equally associated with participation. The perception of a lower level of prices is the covariate more strongly associated with higher customer participation (0.32). This finding stresses the compensatory role of prices relative to the customer’s participation in the purchase process: high participation level is associated to lower prices.

Channel choice does not significantly improve the explanatory power of the model (adjusted R² of 0.37), but the channel type coefficient (0.39) is significant. All the other coefficients do not show any significant change except for the constant. This fact indicates that the type of channel has a direct impact on participation so that indirect channels are positively associated with higher levels of customer participation.

5.3. Covariates of Loyalty Intentions (Equation (3))

The variance of loyalty intentions is explained to a relevant extent by the customer satisfaction and customer participation (adjusted R² of 0.46). These results confirm the expected contribution of the co-production process of distribution services embedded in customer satisfaction, and the transaction costs compensation to the enhancement of loyalty. However, inertia has no effect on repurchase intentions, suggesting that customers do not commit their repurchase intentions based on habit. Rather, customers will be open to other alternatives depending on the context in which their next purchases take place.

The inclusion of the channel type slightly improves the explanatory power of the model from an adjusted R² of 0.46 to 0.47. It has not a direct impact on the loyalty intentions, but there is an indirect effect of channel type through customer participation that has to be kept in mind because it means that OTAs can manage to generate higher levels of customer participation through price-effort compensations.

6. Discussion

The online channels have become the basic playground for tourism companies. Direct and indirect e-channels are actually involved in an e-multichannel operation. In this e-multichannel scenario, it becomes relevant for both tourism service providers and OTAs to have a clear view on their potential to build up a loyal base of customers. OTAs on one side and service providers on the other side seek for developing competitive advantages built on a loyal customer base.

Thus, this research analyses the potential advantages that both direct and indirect channels might hold pursuing customer loyalty goals. The analysis has been focused on the differences between the two types of channels in the attainment of loyalty intentions, and from the perspective of consumers.

In accordance with the results obtained, some differences between the capabilities of the two types of e-channels to foster customer loyalty have been identified. They do not include customer satisfaction since, although this is the most relevant factor in determining loyalty, the type of e-channel does not show any significant influence. Since consumers choose the online channel at their best convenience, this result corresponds to the rational response offered by the channel agents: it must fit the customer needs for the specific purchase they are seeking. Both types of channel thus provide similar levels of customer satisfaction, and it is remarkable that ambience service, as its most important component, is the one that exerts the greatest indirect influence on customer loyalty. The finding supports the one obtained by Llach et al. [13] for direct e-channels, and extending it to the indirect e-channels, which means an important contribution to the knowledge about the differences/similarities between the two e-channels.

The major finding of this study regards the role of customer participation in the purchase process on customer loyalty. The type of channel has a direct impact on participation. Interestingly, the findings fit the transaction costs theory in the household economics context: consumers exchanging more effort (transaction costs) for lower prices are more likely to increase their participation in the purchase process. Based on this, the results open an interesting path for identifying customer participation as a differential source for customer retention between the two channels. This study specifically shows that when customers undertake more nonmonetary transaction costs and less monetary costs, and they purchase in an indirect channel, they increase their participation in the purchase process. That is, they are more likely to repurchase at the same agent in the future. Therefore, compared to direct channels, OTAs in indirect channels show a higher competitive capability to retain customers when providing benefits linked to lower prices to outweigh the participation costs. Indirect e-channels show a higher capability to foster customer intentional loyalty due that they are positively associated with higher levels of customer participation. This conclusion agrees with the argument provided by Koo et al. [23].

Beyond the differences between direct and indirect channels, the nonsignificant influence of inertia on loyalty, independently of the type of e-channel, comes to confirm one of the most relevant characteristics in the shopping and consumption behaviors of consumers for tourism products: the heterogeneity of purchase over time. This result agrees with Gonçalves et al. [31]. Every purchase might be substantially different from previous due to changes in the shopping task, combined with the fact that the market conditions in the tourism industry are very volatile. Moreover, consumers might enjoy getting involved in prior search, and, ultimately, the variety-seeking behavior is very present in the consumption of tourist products. Separately or combined, these factors weaken the loyalty commitment of customers towards the companies at which they purchase.

The results obtained comparing direct and indirect tourism e-channels in getting loyalty provide relevant managerial implications. First of all, both types of tourism agents, service providers and OTAs, should be concerned about the distribution services of ambience and assurance of product delivery in time and form, because they are the main drivers of customer satisfaction and consequently of customer loyalty. By improving these services, tourism agents would permit customers to better personalize the tourism product, and the booking process itself, favoring the retention objectives. This fact would increase the competitiveness level in the market providing benefits for final users. But tourism companies should keep alert since satisfied customers are very sensitive to a quality loss and would switch to competitors [31].

Regarding the differences finding between the direct and indirect e-channels, indirect channels have shown a higher capability to retain customers than direct channels due to their intrinsic advantage in generating customer engagement with the booking process. That is, being equal the consumers’ perceptions of the channel policies (payments and privacy), and of the transaction costs (effort, search, time and perceived price), indirect channels get a higher participation level. This participation gap is the basis upon which service providers should project their initiatives for improving their competitiveness in the online tourism market.

Four recommendations for service providers stem from the research conclusions. The first recommendation is that their customers should perceive that they receive a fair treatment, getting a favorable price compensating the effort and costs involved in completing the booking. This is a requirement for services providers to convert increases in customer participation on improvements in the loyalty level of their customers.

The second recommendation is about closing the gap with OTAs in the width and depth of the product assortment. Service providers are mostly specialists of one type of product; they have a short product mix giving little space to their customers to personalize the product through their websites alone. To moderate their disadvantage in this service, tourism service providers should seek co-operation. Firstly, to offer a broader range of complementary product categories from other service providers, which are commonly combined in tourism transactions, such as accommodation, car rental, transportation, destination services, tickets and insurance. This movement is already widespread in some sectors such as airlines that sponsor access to the websites of other providers for services at the destination. Secondly, co-operation would also greatly help to improve a service category by grouping service providers on web pages that add the offer of that category.

The third recommendation concerns improvements in the booking process. Service providers should incorporate into its booking process participation paths based on the display of different ways of savings linked to product and service alternatives, which means assuming more transaction costs by the customer. They could extend their product portfolio following criteria of price.

A fourth managerial implication for service providers concerns the balance between costs and profits involved in sustaining their customer acquisition and retention strategies. Initially, the tourism service providers developed their websites for customer acquisition but more and more they orientate their effort to customer retention objectives. Since the major source of profitability is a solid customer base with high records of loyalty, service providers invest in incentives to retain their customers. But when service providers consider to combine direct with indirect distribution e-channels, the costs associated to the benefits passed to loyal customers (discounts, bonuses, complementary services) through the direct channel should be lower than the fees paid to the OTAs for the intermediation.

The approach of exclusively covering online channels is a limitation on the scope of the results obtained. Some caveats should be made. This study is exclusively focused on online channels, but the multichannel operation of service providers and OTAs includes in most cases offline alternative channels. As such, there is an interdependency between the offline and the online channels of these tourism companies that needs to be considered to give a complete perspective for the achievement of customer acquisition and retention objectives. The conclusions point at some logic managerial recommendations about the offline channel operation, but we cannot go any further since the offline–online interactivity precludes doing so. Future research should be addressed to the consumers’ omnichannel behavior, including the offline channels to analyze how they interact with the online channels and the consequences on the potential for customer acquisition and retention strategies. The empirical model could be improved by considering other customer satisfaction antecedents to complement the co-production process implicit in the distribution services. Attending the technological advances in tourism and hospitality, co-creation processes could play an important role in this task [55]. Moreover, in order to confirm the inertia effect on loyalty, future studies should control shopping task and different product categories.