Business and Customer-Based Chatbot Activities: The Role of Customer Satisfaction in Online Purchase Intention and Intention to Reuse Chatbots

Abstract

In the online shopping context, brands aim to achieve a high level of profit by providing better customer satisfaction by using various artificial intelligence tools. They try creating a satisfactory customer experience by creating a system that provides never-ending customer support by using dialog-based chatbots, especially in the field of customer service. However, there is a lack of research investigating the impact of business and customer-based chatbot activities together on online purchase intention and the intention to reuse chatbots. This research considers the use of chatbots as a marketing tool from both customer and business perspectives and aims to determine the factors that affect the customers’ intention to purchase online and reuse chatbots. Accordingly, the impact on customer satisfaction with chatbot usage, which is based on chatbots’ communication quality and customers’ motivations to use chatbots, on online purchase intention and intention to reuse chatbots was examined. Through an online questionnaire with two hundred and ten participants, employing structural equation modeling, we revealed that customer satisfaction with chatbot usage has a greater impact on the intention to reuse chatbots than on online purchase intentions. In addition, chatbot communication quality has a greater impact on customer satisfaction with chatbot usage than customers’ motivation to use chatbots. To solidify these findings, confirmatory factor analysis, along with reliability and validity assessments, were implemented within the analytical framework to provide robust support for the study’s hypotheses. These findings not only provide empirical evidence and implications for companies in online shopping but also extend the understanding of AI tools in marketing, highlighting their subtle impact on consumer decision-making in the dynamic digital marketplace.

1. Introduction

Companies have become more consumer-centric and renewed their business models to adapt products and services to the rapid inventions of internet technology and e-commerce [1,2]. It has become a crucial issue to respond to consumer demands as soon as possible and in the most accurate way given the increase in internet usage and online shopping [3]. In this respect, companies conduct various marketing activities and seek innovative ways and tools to meet increasing and unique consumer expectations in the online environment, such as social media channels, online marketplaces, and shopping websites [4,5]. Thanks to integrating new technologies into their marketing activities, companies try performing more effective and interactive marketing activities to ensure customer satisfaction and secure the competitive advantage that stems from superior customer satisfaction [6].

Especially in the context of e-commerce, due to the intense competition, it is crucial to attract customers and keep them satisfied. To achieve that in the pre-purchase phase and maintain it in the post-purchase process, it is necessary to establish fast, reliable, and solution-oriented online communication. At this point, by leveraging artificial intelligence (AI) capabilities, marketers strive to establish an emotional connection with customers that goes beyond mere satisfaction [7]. They are able to deliver versatile and impactful communication through online channels, tapping into the immense power and potential of AI. Chatbots, the fastest-growing global service channels [8], enable personalized communication with useful services, recommendations, or convenient shopping [9] while enabling computers to conduct business with humans from the human’s perspective [10]. According to Artasanchez and Joshi [11], a well-designed chatbot possesses four essential characteristics: adaptability, personalization, availability, and relatability. Adaptability refers to the chatbot’s ability to comprehend user input and respond appropriately to varying situations, including by providing polite responses when encountering unrecognized or unexpected input. Personalization is achieved by the chatbot’s ability to gather and retain user information, allowing it to recall previous interactions and make the user feel valued during subsequent engagements. Availability ensures that the chatbot is accessible whenever needed, offering prompt responses to user inquiries. Finally, relatability ensures that users feel as though they are engaging in a familiar, coherent conversation. Without these core features, a chatbot cannot function effectively or fulfill its intended purpose.

With the help of customer service assistants, decision support systems, smart logistics applications, and big data analytics, companies can better analyze customer needs and make operational and managerial decisions based on data [12]. At this point, benefits from customer service assistants such as chatbots improve the effectiveness and sustainability of the services that are provided to customers. In addition to delivering support, AI-driven customer service that utilizes chatbots powered by deep learning and machine learning can anticipate customer needs and offer individualized recommendations and guidance [13]. Chatbots are designed as text-based or voice-based channels using an artificial intelligence solution that works uninterruptedly and can interact with the customer as a marketing tool. Juniper Research [14] predicts that consumer retail spending via chatbots worldwide will reach $72 billion—up from just $12 billion in 2023—by 2028, because chatbots are growing in popularity among businesses and consumers alike, as consumers request 24 h assistance in areas ranging from banking and finance to health and wellness. According to Gartner, Inc. [15], by 2027, about 25% of organizations will rely on chatbots as their main customer service channel. According to a survey of more than 2000 customer service professionals with different roles, company sizes, and locations conducted by Intercom (2024), 44% of them believe that chatbots are the most promising area of investment for support teams. Furthermore, the rise in AI technology is creating new strategic support positions, such as chatbot analysts who evaluate the performance of chatbots and customer conversation data to identify new opportunities and insights, accounting for 42% of the focus. As reported by Invesp [16], two-thirds of consumers worldwide used a chatbot for customer support in the last year, and 40% of consumers did not consider whether they were communicating with a real customer service representative or a chatbot as long as they got what they needed.

In a study by the Capgemini Research Institute [17] on conversational bots, 74% of respondents revealed that they use chatbots to research and purchase products and services, create shopping lists, and check order status. Another study found that users expect chatbots to provide assistance around the clock, provide quick responses, and connect them to a human representative when requested [18]. Aivo [19] found that the chatbots applied in the e-commerce industry effectively solve 65% of customers’ questions and requests while only transferring 21% to live support.

Although there are many studies in the literature examining chatbots as a marketing tool, prior research has mostly approached the issue from either a business perspective, such as chatbot effectiveness, chatbot marketing efforts, or anthropomorphism [20,21,22,23], or from consumer perspectives, such as the factors affecting chatbot use or customer satisfaction [24,25,26,27]. While there are studies focusing on research on chatbots, their impact on consumer behavior, and findings on the consumers’ positive perception of the use of chatbots [28,29,30], there is indeed a gap in the literature specifically addressing the combined effects of consumer- and business-oriented chatbot activities on online purchase intention and chatbot reuse intention.

The investigation of this phenomenon through various theoretical perspectives is essential for understanding the emergence of human–chatbot relationships across different contexts and cultures. Additionally, there is a pressing need for future research in the domain of humanoid chatbots [31,32]. Based on our current knowledge, research that combines both consumer- and business-oriented chatbot activities, particularly in relation to consumers’ online purchase intentions and their intention to reuse chatbots, is lacking. To fill this gap, we conducted a study that evaluates the relationships between chatbot marketing efforts, motivation to use chatbots, chatbot communication quality, satisfaction with chatbot usage, purchase intention, and continued usage intention of chatbots, which will allow for a more comprehensive look at the subject. We extend knowledge by investigating a broader perspective on factors affecting customers’ purchase and continued usage intention toward chatbots in online shopping. This study seeks to address four research questions. First, do chatbot marketing activities lead to improved chatbot communication quality? Second, can chatbot communication quality be a predictor of customer satisfaction with chatbot usage? Third, which motivations behind the use of chatbots impact customer satisfaction with chatbot usage? Fourth, what is the impact of satisfaction with using chatbots on both (1) online purchase intention and (2) continuous usage intention?

To address these questions, we developed a model that examines the relationships between chatbot marketing efforts, chatbot communication quality, chatbot usage motivations, satisfaction with chatbot usage, reuse, and purchase intention. An online questionnaire with 210 Turkish participants facilitated the empirical examination of these relationships through structural equation modeling (SEM).

This study enriches the body of knowledge in digital marketing and consumer behavior by exploring the critical role of chatbot interactions in enhancing online shopping experiences and offers valuable insights for e-commerce businesses aiming to leverage artificial intelligence for superior customer engagement. Moreover, this study offers practical insights for practitioners and policymakers, enabling them to optimize chatbot functionalities and thereby elevate online reuse and purchase intentions and cultivate consumer satisfaction towards chatbot integration to achieve a competitive advantage in the digital marketplace.

The paper is organized as follows: Section 2 establishes a theoretical grounding by critically reviewing relevant literature. Section 3 then delves into the adopted methodology. Section 4 unfolds the study’s findings through analysis and results. Section 5 engages in a critical discussion of these insights, drawing out their theoretical and practical implications. Finally, the paper examines its limitations and offers valuable suggestions for future research.

2. Theoretical Background and Hypotheses Development

2.1. Chatbot Marketing Efforts

Brands try to influence users’ decision-making processes through various marketing efforts. Kim and Ko [33] evaluated the marketing efforts of luxury brands in the context of social media marketing with the dimensions of entertainment, interaction, trendiness, customization, and word of mouth (WOM). They found that these dimensions significantly positively affect value equity, relationship equity, and brand equity. Considering these exact dimensions as a holistic concept, Godey et al. [34] found that these dimensions positively and directly affect consumer response and brand equity. On the other hand, Chung et al. [35] used the dimensions of interaction, trendiness, customization, and problem-solving to measure the effectiveness of marketing efforts in communicating with brands online. Similarly, Cheng and Jiang [36] used interaction, information, accessibility, and customization as dimensions of marketing efforts to test the impact of chatbot marketing efforts on chatbot communication quality. Within the scope of chatbot marketing efforts, this study examines the roles of interaction, trendiness, customization, and problem-solving.

A well-designed chatbot can reduce marketing costs and increase conversion rates, creating an additional revenue channel for companies. In this context, chatbots can be defined as an important marketing tool for brands. The initial focus area, interaction, is defined as the effectiveness of communication between the chatbot and the customer. By creating an entirely new communication channel to interact with customers, thanks to the virtual customer support assistant, companies can more effectively improve their capacity to provide real-time customer support with fewer employees [37]. Being a 24 h active software, obtaining quick answers, solving simple questions, and having features such as easy communication make chatbots a useful communication tool for customers [38]. The second construct, trendiness, examines the extent to which utilizing the chatbot is perceived as being current and fashionable. Another important factor, problem-solving, examines the chatbot’s effectiveness in efficiently addressing and resolving customer issues. Although chatbots cannot apply humanoid features such as empathy, innuendo, and critical thinking very well, and their communicative and social competence is still insufficient [39] as they start to serve more effectively, it is expected that the chatbot will reach a solution by understanding and empathizing with the customer’s problems. It is also underlined that human intervention will be required only for extreme problems in the future. Last but not least, customization assesses the chatbot’s ability to create tailored experiences for individual customers. It is expected that chatbots will be able to access information from various online communication channels and provide faster and more effective service by using this information [11].

2.2. Chatbot Communication Quality

Communication quality is crucial for positive customer experiences and the maintenance of a unique market position for brands in online interactions [40]. Mohr and Sohi [41] assess communication quality by considering factors like timeliness, accuracy, completeness, credibility, and adequacy. In this respect, high-quality communication in chatbot-customer interactions hinges on designing chatbots that deliver prompt, accurate, and comprehensive responses, while ensuring the information is credible and relevant to individual customer needs. Besides that, Edwards et al. [42] propose drivers like attraction, communication competence, credibility, and intent to interact as crucial drivers for effective communication quality. In other words, leveraging chatbots that engage customers, demonstrate competence, build trust, and proactively address needs can lead to marked improvements in communication quality. Edwards et al. [42] conducted a study investigating potential differences in communication quality between human and bot agents on Twitter. Their findings suggest that, in terms of perceived source credibility, communication competence, and interactional intentions, users did not differentiate between the two types of agents.

Grounded in the existing literature on bot agent communication, this study explores communication quality through the lens of established dimensions like accuracy, credibility, and communication competence, as identified by Edwards et al. [42] and Chung et al. [35]. Accuracy signifies the chatbot’s ability to provide precise and reliable information, fostering trust and consistency within the interaction. Furthermore, credibility represents the perceived honesty and trustworthiness of the chatbot. Ultimately, communication competence evaluates the chatbot’s efficiency and effectiveness in facilitating interactions with customers. By prioritizing accuracy, credibility, and communication competence, brands can cultivate trust in digital interactions through chatbots. This not only meets users’ expectations and needs but also highlights the chatbot’s ability to facilitate smoother and faster customer service experiences.

Drawing upon the established literature, we anticipate a positive impact on communication quality due to enhanced chatbot marketing efforts. Such marketing efforts can improve communication quality by expediting inquiry resolution, improving information accuracy, and fostering personalized interactions that resonate with individual needs and preferences. Accordingly, the following hypothesis was developed:

H1: 

Chatbot marketing efforts significantly and positively impact chatbot communication quality.

2.3. Satisfaction with Chatbot Usage

Customer satisfaction stands as a cornerstone concept in contemporary marketing theory and practice, emphasizing the reciprocal relationship between delivering superior customer value and achieving sustainable profitability. It serves as a critical metric for gauging the extent to which an organization fulfills the diverse needs of both its customers and its firms [43]. Customer satisfaction occurs at the point where the performance of the products and services offered by the brand meets customer expectations.

In a highly competitive e-commerce environment, online businesses must provide effective customer service to ensure and increase customer satisfaction. Hassan et al. [44] suggest that companies implementing robust and reliable CRM practices can cultivate higher levels of customer satisfaction. Chatbots, as an integral part of CRM strategies, play a vital role in this process. It is essential for chatbots to be able to provide high-quality communication in order for them to be effective in customer service to satisfy customers. This highlights the significance of accuracy, reliability, and coordination in the responses that they provide.

The low quality of online communication causes complaints about customer experience [45], harms trust, and reduces customer satisfaction [46]. When customers encounter false information given by chatbots, it can raise concerns about the reliability of the source. Providing accurate information increases the trust in the chatbot source and is proof that the chatbot fulfills its task completely [24,35]. The lack of an appropriate and coordinated response from chatbots can have an unfavorable effect on the overall customer experience [47].

At this point, in order for chatbots to be an effective communication tool for online companies, they must be able to correctly understand the needs of customers and respond to the information they seek quickly and correctly [48]. For this reason, the ability of the chatbots to provide services to the specified features will help companies achieve their targeted results and increase customer satisfaction with chatbot usage by making a positive contribution to the customer experience [35,42,49]. Thus, the following hypothesis was formed:

H2: 

Chatbot communication quality has a significant and positive impact on customer satisfaction with chatbot usage.

2.4. Motivation for Chatbot Usage

To influence the decision-making process of consumers with effective marketing campaigns in the online environment, companies have to design chatbots that will respond to customers’ needs and expectations at the right time and in the right way. Achieving this success on the brand side is possible by researching the purposes of consumers using chatbots and implementing a system suitable for related factors. It is found that the most important reason behind the motivation for using chatbots is productivity, which includes ease of use, accessing information, and obtaining support quickly, followed by entertainment and social and relational aspects that refer to the desire to avoid loneliness and socialization, respectively [50]. Productivity-related perceived performance associated with chatbot use positively affects customer satisfaction [51]. On the other hand, in situations characterized by distress and social isolation, individuals may exhibit attachment formation towards chatbots perceived as offering emotional support, encouragement, and a sense of psychological security [52]. By potentially satisfying inherent human needs for social connection and self-exploration, chatbots may foster the development of emotional attachments and contribute to the cultivation of deeper relationships [53]. Additionally, entertainment assesses the ability of chatbot interactions to provide enjoyable and engaging experiences for customers. While purchase intention remains a significant goal, customer motivations extend to seeking enjoyment and engagement within the chatbot interaction itself [54].

However, chatbots can be designed to imitate human characteristics. Anthropomorphism refers to the act of attributing traits or characteristics that are typically associated with humans to agents that are not human [55], such as giving chatbots a name (e.g., “Alexa”) [56], ascribing them consciousness or emotions [57], and using conversational cues like empathy, language style, and emojis [58] to enhance interactions in specific customer contexts. Han [59] asserted that users find chatbots more entertaining if the anthropomorphism level of the chatbot is high. Rietz et al. [60] argued that chatbots with more anthropomorphic designs increase user satisfaction. Sheehan et al. [61] also claimed that chatbots with anthropomorphic features may satisfy the social expectations of customers who need more human interaction. A chatbot lacking anthropomorphic features has the potential to heighten dissatisfaction among customers, consequently leading to negative word-of-mouth and potentially influencing customer attitude and behavior in a negative manner [62]. Consequently, the following hypothesis was developed:

H3: 

Motivation to use chatbots has a significant and positive impact on customer satisfaction with chatbot usage.

2.5. Purchase and Reuse Intention

According to the Theory of Planned Behavior, the most effective determinant of behavior is behavioral intention [63]. Consumer behavior can generally be predicted by consumers’ intentions. Accordingly, purchase intention refers to the likelihood that a consumer will make an online purchase in the near future. Purchase intention can also be defined as an individual’s desire to purchase a product of a certain brand as a result of some evaluations [64]. Users who perform various e-commerce activities may tend to purchase if they are satisfied with the various services provided by the business, and also tend to repeat their positive experiences by using the same services again.

There are many studies in the relevant literature that examine the determinants of online purchase intention. It is clearly revealed in these studies that customer satisfaction is one of the key factors that affect online purchase intention [65,66,67,68,69,70]. In line with this, the relationship between online purchase intention and continuous intention of use has been investigated in many studies by using the Expectation-Confirmation Theory, the Cognitive Model, and the Technology Continuance Theory [71].

The Expectancy-Confirmation Theory (ECT) offers a comprehensive framework derived from consumer behavior studies and prior research in the field of Information Systems (IS) to understand the dynamics behind the engagement of users with IS platforms [72]. Mainly, this framework examines the factors that influence individuals’ continued use of IS. According to ECT, satisfaction, which is influenced by the confirmation of expectations and perceived usefulness, determines users’ continuance intention. In the Cognitive Model of Oliver [73], the model begins by establishing an individual’s initial attitude towards a product as a function of their expectations before use. After experiencing the product, the individual’s satisfaction level is assessed based on their expectations and the degree to which those expectations were met or not met, thus satisfaction is defined as a function of expectations and disconfirmation. Satisfaction, in turn, is believed to influence attitude change and purchase intention. Technology Continuance Theory (TCT), developed by Liao et al. [74], merges three influential information system models—Technology Acceptance Model (TAM), Expectation Confirmation Model (ECM), and Cognitive Model (COG)—to predict and explain users’ engagement with technology and their ongoing intention to use it. In short, TCT is a model that predicts how likely it is that people will keep using a technology, and it is applicable across different adoption stages. Its significant advancement lies in incorporating both satisfaction and attitude to enhance understanding of users’ continuous usage intentions, thereby offering a more comprehensive framework than its predecessors.

According to these theories, customer satisfaction has a positive impact on reuse intention [72,73,74,75]. Thus, the following hypotheses were formed:

H4: 

Customer satisfaction with chatbot usage positively affects online purchase intention.

H5: 

Customer satisfaction with chatbot usage positively affects the reuse intentions of chatbots.

Figure 1 shows the conceptual model of this study based on the five hypotheses proposed above.

3. Methodology

The items developed by Chung et al. [35] regarding chatbot marketing efforts were used to measure the sub-dimensions named engagement, trendiness, customization, and problem-solving. The items related to motivations for using chatbots, productivity, and social and relational motivation, were measured using the items developed by Rieke [76], and anthropomorphism was measured using the opposite expressions determined by the authors. Entertainment was measured by using items from Chung et al. [35]. Accuracy, credibility, and communication competence, which are specified as the dimensions determining communication quality, were also measured using the items developed by Chung et al. [35]. Their study was also used for items related to customer satisfaction. The items developed by Hsiao et al. [77] were adapted to measure the reuse intentions of chatbots. Items developed by Kim et al. [78] were adapted to measure purchase intentions. Participants were asked to respond to the statements in the questionnaire in the range of “strongly disagree” (1) to “strongly agree” (7).

Since it is impossible to define the population who has used chatbots on e-commerce sites before, the non-probability convenience sampling method was used. Potential participants were reached through social media platforms, professional networks, and particularly channels where discussions about technology and consumer behavior are prevalent. An online questionnaire on Google Forms was used for data gathering. The questionnaire was distributed electronically through social media and professional networking groups. A control question was first asked of the participants to ensure that participants had used chatbots at least one time in their prior online shopping experience. Ethical approval was received from the Ethical Board of Dokuz Eylul University, and participants were first asked to approve their consent. The questionnaire was conducted over a period of two months, and it was distributed to individuals in Türkiye. A pilot study was conducted with a small group of participants to ensure clarity and understanding of the questionnaire items. A total of 246 responses were obtained; however, 36 of them were excluded due to repetitive and incomplete data, representing a total of 210 participants. The structural equation model was used to test the study hypothesis. Seeing that the sample size is sufficient [79,80], data were analyzed using SPSS 26 and Amos 26.

Most participants are women (58 percent) between the ages of 18–34 (93.9 percent), have at least a graduate level of education (68.6 percent), and spend more than 4 h on the Internet (72.4 percent). Considering online shopping, most of them stated that they shopped online more than 15 times (51.4 percent) and used chatbots at least two times (77.1 percent) in a year. The most frequently used online shopping website for a chatbot is trendyol.com which is the biggest online shopping platform as a marketplace in Türkiye according to similarweb (accessed on 3 October 2024) [81]. The key demographic information is detailed in

Table 1. Summary of Participant Demographics.

	Frequency (n)	Percent (%)
Gender
Female	122	58.1
Male	85	40.5
Rather not say	3	1.4
Total	210	100.0
Age
18–24	107	51.0
25–34	90	42.9
35–44	7	3.3
45+	6	2.9
Total	210	100.0
Marital Status
Married	31	14.8
Single	179	85.2
Total	210	100.0
Education Level
High school graduate	12	5.7
Bachelor’s degree	144	68.6
Master’s or Doctorate degree	54	25.7
Total	210	100.0
Internet Usage Time (Daily)
0–3 h	58	27.6
4–6 h	91	43.3
7–9 h	26	12.4
10+ h	34	16.2
No response	1	0.5
Total	210	100.0
E-commerce Shopping
Those who do	209	99.5
Those who do not	1	0.5
Total	210	100.0
Number of E-Commerce Purchases (Annual)
0–15	102	48.6
16–30	62	29.5
31–45	13	6.2
46–60	19	9.0
61+	14	6.7
Total	210	100.0
Number of Interactions with a Chatbot on an E-Commerce Website (Annual)
1	48	22.9
2–5	94	44.8
6–10	36	17.1
10+	32	15.2
Total	210	100.0
E-commerce Websites You Use Chatbots on
Trendyol	123	36.72
Yemeksepeti	51	15.2
n11	39	11.6
Hepsiburada	39	11.6
Getir	7	2.1
AtasunOptik	4	1.2
VatanBilgisayar	16	4.8
Other (please specify)	56	28.7
Total	335	100.0

.

Potential non-response bias was assessed by comparing early (n = 108) respondents who completed the questionnaire within the first month of distribution and late (n = 102) respondents who completed it afterward on all constructs. Independent sample t-tests were performed on the variables. No statistically significant differences were identified between early and late responses in terms of gender (p = 0.35), age (p = 0.80), frequency of online shopping (p = 0.89), time spent on the Internet (p = 0.82), and frequency of chatbot use (p = 0.92).

4. Analysis and Results

This research incorporated two phases used to define and confirm the factors that affect online purchase intention and the intentions to reuse chatbots. First, confirmatory factor analysis (CFA) was conducted to confirm the factor structure of the measurement models and to examine its reliability and validity. Also, Hair et al. [82] state that in order to demonstrate the validity of the second-order model, there should be a strong and significant correlation between the second-order component and its dimensions (p < 0.05), and the R² of each dimension of the first-order models should be larger than 0.5. Thus, second-order components and related dimensions were analyzed. Second, the structural model was tested by second-order SEM analysis and the path coefficients were estimated. The variance inflation factor (VIF) scores were calculated to evaluate multicollinearity. The highest VIF value was 3409, indicating no multicollinearity among variables [83].

4.1. Measurement Model (First-Order Constructs)

Internal reliability was measured using Cronbach’s α Coefficients and Composite Reliability (CR) values. Convergent validity and discriminant validity were evaluated with Average Variance Extracted (AVE) and HTMT (Heterotrait-Monotrait Ratio), respectively. The item means, standard deviations, factor loadings, AVE, and reliability values are shown in

Table 2. Results of first-order factors.

Variable	Item	Mean	SD	Factor Loading	AVE	CR	α
Interaction	INT1	4.01	1.777	0.891	0.692	0.817	0.861
	INT2	4.20	1.911	0.767
Trendiness	TRE1	4.51	1.892	0.873	0.686	0.865	0.855
	TRE2	3.86	1.827	0.656
	TRE3	4.48	1.887	0.929
Customization	CUS1	3.57	1.84	0.716	0.664	0.855	0.861
	CUS2	3.65	1.806	0.820
	CUS3	4.14	1.693	0.899
Problem-solving	PRB1	3.81	1.793	0.778	0.708	0.879	0.878
	PRB2	4.03	1.884	0.820
	PRB3	3.71	1.781	0.920
Productivity	PRO1	4.03	1.956	0.895	0.650	0.880	0.889
	PRO2	4.54	1.932	0.818
	PRO3	4.54	1.959	0.820
	PRO4	3.39	1.919	0.675
Social & Relational Motivation	SOC1	2.08	1.69	0.898	0.798	0.922	0.921
	SOC2	2.39	1.796	0.888
	SOC3	2.01	1.647	0.894
Anthropomorphism	ANT1	3.47	1.556	0.759	0.547	0.783	0.783
	ANT2	3.54	1.553	0.752
	ANT3	4.04	1.622	0.705
Entertainment	ENT1	3.00	1.904	0.768	0.638	0.875	0.866
	ENT2	3.17	1.951	0.688
	ENT3	2.61	1.744	0.915
	ENT4	2.35	1.657	0.808
Accuracy	ACC1	3.50	1.772	0.811	0.759	0.904	0.903
	ACC2	3.56	1.758	0.891
	ACC3	3.34	1.800	0.909
Credibility	CRE1	4.37	1.857	0.807	0.660	0.853	0.855
	CRE2	4.54	1.838	0.774
	CRE3	4.32	1.879	0.855
Communication Competence	COM1	3.17	1.876	0.902	0.626	0.825	0.787
	COM2	3.02	1.82	0.904
	COM3	4.70	1.961	0.498
Satisfaction with chatbot usage	SAT1	4.06	1.792	0.956	0.914	0.977	0.976
	SAT2	3.97	1.741	0.966
	SAT3	3.94	1.793	0.973
	SAT4	4.06	1.775	0.929
Reuse Intention	REU1	4.21	1.803	0.935	0.685	0.866	0.904
	REU2	3.18	1.737	0.744
	REU3	3.56	1.747	0.791
Purchase Intention	PUR1	3.58	1.908	0.852	0.800	0.941	0.940
	PUR2	3.42	1.896	0.937
	PUR3	3.40	1.902	0.957
	PUR4	3.70	2.015	0.824

Note: SD: standard deviation, AVE: average variance extracted, CR: composite reliability, α: Cronbach’s Alpha.

.

The CR values ranged from 0.783 to 0.977, greater than the recommended threshold of 0.70 [82]. The results showed that the AVE values were greater than 0.5, all CR values were above 0.7, and all factor loadings were significant, indicating convergent validity [82]. HTMT values were found to be less than 0.9. Thus, the discriminant validity was confirmed [84] (

Table 3. Discriminant validity of the first-order CFA.

Constructs	PUR	ANT	INT	TRE	CUS	PRB	PRO	SOC	ENT	ACC	CRE	COM	SAT	REU
PUR	0.894
ANT	0.410	0.739
INT	0.608	0.496	0.831
TRE	0.535	0.348	0.617	0.828
CUS	0.722	0.499	0.677	0.696	0.815
PRB	0.722	0.457	0.626	0.770	0.702	0.841
PRO	0.648	0.374	0.696	0.669	0.677	0.709	0.806
SOC	0.357	0.197	0.209	0.141	0.343	0.352	0.270	0.893
ENT	0.547	0.311	0.445	0.404	0.557	0.550	0.484	0.736	0.799
ACC	0.587	0.439	0.601	0.550	0.654	0.680	0.622	0.290	0.519	0.871
CRE	0.507	0.245	0.587	0.667	0.636	0.654	0.625	0.138	0.430	0.711	0.813
COM	0.649	0.432	0.460	0.415	0.613	0.681	0.556	0.475	0.524	0.708	0.571	0.791
SAT	0.600	0.381	0.714	0.711	0.604	0.769	0.718	0.178	0.452	0.688	0.711	0.598	0.956
REU	0.726	0.403	0.709	0.644	0.655	0.749	0.620	0.186	0.474	0.653	0.659	0.599	0.743	0.827

Note: Square-roots of AVE are in bold, on the diagonal; Off-diagonal values represent the correlations between the latent constructs.

). The CFA also showed that all items loaded on their related factors and the measurement model showed a good fit (χ²/df = 2.023; CFI: 0.907; RMSEA: 0.070; SRMR: 0.073) [85].

In addition, Harman’s single-factor test was conducted to assess for potential common method bias [86]. The results of Harman’s single-factor test showed that the largest variance explained by an individual factor was 40.93%, which is below the 50% threshold. This clearly indicates that common method bias is not a concern in this study.

4.2. Measurement Model (Second-Order Constructs)

After conducting and validating the first-order CFA, we continued to check the proper fit of the second-order model. This study estimates second-order factor models; chatbot marketing efforts, chatbot communication quality, and motivation for chatbot usage. The substantial significant correlations (p < 0.001) between the three second-order constructs (chatbot marketing efforts (CME), chatbot communication quality (CCQ), and motivation for chatbot usage (MFCU)) and their dimensions are shown via a hierarchical construct model. Particularly, interaction (t = 13.687, p < 0.001), trendiness (t = 13.010, p < 0.001), customization (t = 11.281, p < 0.001), and problem-solving (t = 12.364, p < 0.001) have significant correlations with chatbot marketing efforts. In addition, accuracy (t = 11.642, p < 0.001), credibility (t = 9.842, p < 0.001), and communication competence (t = 11.394, p < 0.001) also have a significant correlation with chatbot communication quality. In addition, productivity (t = 9.615, p < 0.001), social and relational motivation (t = 10.303, p < 0.001), anthropomorphism (t = 5.914, p < 0.001), and entertainment (t = 10.467, p < 0.001) are found to be significant determiners of motivation for chatbot usage. Furthermore, it is recommended that the R2 of each construct’s dimensions be greater than 0.5 in order for the second-order models to be valid [82]. The R2 of the four dimensions of the CME construct found in this study ranged from 0.769 to 0.990, the R2 of the three subscales of the CCQ construct ranged from 0.605 to 0.775, and the R2 of the four dimensions of the MFUC ranged from 0.545 to 0.730. All the first-order constructs’ factor loading values and other values of chatbot marketing efforts, chatbot communication quality, and motivation for chatbot usage (CR, AVE, and Cronbach alpha values) all meet the value standard [82]. The second-order measurement model’s results, convergence, and discriminant validity are shown in

Table 4. Results of second-order factor.

Second-Order Constructs	Factor Loading	AVE	CR	α
Chatbot marketing efforts (CME)
Interaction	0.888	0.851	0.958	0.915
Trendiness	0.877
Customization	0.995
Problem-solving	0.925
Chatbot communication quality (CCQ)
Accuracy	0.881	0.674	0.861	0.858
Credibility	0.778
Communication competence	0.801
Motivation for using chatbot (MFUC)
Productivity	0.738	0.643	0.878	0.718
Social and relational motivation	0.843
Anthropomorphism	0.767
Entertainment	0.854

Note: AVE: average variance extracted, CR: composite reliability, α: Cronbach’s Alpha.

and

Table 5. Discriminant validity of the second-order CFA.

Constructs	CME	CCQ	MFCU
CME	0.992
CCQ	0.787	0.821
MFCU	0.654	0.657	0.802

Note: CME: Chatbot marketing efforts; CCQ: chatbot communication quality; MFCU: motivation for using chatbots. Square-roots of AVE are in bold, on the diagonal; Off-diagonal values represent the correlations between the latent constructs.

.

4.3. Structural Model

Structural equation modeling was used to test the proposed hypotheses. Estimated path coefficients, t-values, and the squared multiple correlations and the results of each hypothesis are presented in

Table 6. Results of hypotheses tests.

Hypothesis	Standardized β	SE	t	Result
H1: Chatbot Marketing Efforts → Chatbot Communication Quality	0.887 ***	0.075	9.803	Supported
H2: Chatbot Communication Quality → Satisfaction with chatbot usage	0.910 ***	0.129	10.242	Supported
H3: Motivations for Chatbot Usage → Satisfaction with chatbot usage	−0.033	0.062	0.682	Rejected
H4: Satisfaction with chatbot usage → Purchase Intention	0.790 ***	0.060	12.285	Supported
H5: Satisfaction with chatbot usage → Reuse Intention of Chatbots	0.939 ***	0.042	23.247	Supported

Note: *** p < 0.001.

.

According to the results, four of five hypotheses were found to be significant. More specifically, H1, which suggested that chatbot marketing efforts positively affect the chatbot communication quality, was accepted (β = 0.887, p < 0.001). H2, which suggested that chatbot communication quality has a significant impact on customer satisfaction with chatbot usage, was also accepted (β = 0.910, p < 0.001). H4, which suggests that customer satisfaction with chatbot usage positively affects the purchase intention (β = 0.790, p < 0.001) was supported. Finally, H5, which predicted that customer satisfaction with chatbot usage positively affects the reuse intention of chatbots (β = 0.939, p < 0.001) was also supported. In contrast, H3, which suggested that motivation for using chatbots has a significant impact on customer satisfaction with chatbot usage (β = −0.033, p > 0.05) was rejected.

A bootstrap analysis was applied to test the likely mediation effect [87] of chatbot communication quality on the relationship between motivations for chatbot use and customer satisfaction with chatbot usage. The tests demonstrated that the direct effect of motivations for chatbot use on customer satisfaction with chatbot usage in the presence of the chatbot communication quality (p = 0.301, 95% CI = −0.241–2.252) is not significant. Also, there is no significant indirect effect of motivations for chatbot use on customer satisfaction with chatbot usage via the intervention of chatbot communication quality (p = 0.156, 95% CI = −0.435–2.415). Thus, it is found that motivations for chatbot use and customer satisfaction with chatbot usage are not connected directly or indirectly. The results indicate that chatbot communication quality, which is triggered by chatbot marketing efforts, meets customer satisfaction with chatbot usage and enhances purchase intention and reuse of chatbots. Overall, the model explains approximately 79% of the variance in chatbot communication quality (R² = 0.788), 79% of the variance in customer satisfaction with chatbot usage (R² = 0.794), 62% of the variance in purchase intention (R² = 0.623), and 88% of the variance of reuse intention of chatbots (R² = 0.882).

5. Conclusions and Implication

5.1. Conclusions

This study examined how chatbot marketing efforts, chatbot communication quality, and motivation for using chatbots can interact with customer satisfaction with chatbot usage in online shopping and the impact of customer satisfaction with chatbot usage on online purchase intention and reuse intention of chatbots. This study provides a more comprehensive perspective by considering business- and consumer-based chatbot activities together. Specifically, it assesses how chatbot marketing efforts and chatbot communication quality from businesses (business-oriented activities) and consumer motivations for using chatbots (consumer-oriented activities) interact to influence key outcomes like customer satisfaction with chatbot usage, purchase intention, and reuse intention. This comprehensive perspective enables a better understanding of the interdependencies between chatbot strategies implemented by businesses and how these strategies align with consumers’ expectations and needs, ultimately affecting their satisfaction and behavior. In this study, factors that determine the consumers’ intention to purchase and reuse of chatbots on e-commerce websites were investigated by considering consumer motivations for chatbot usage and chatbot marketing efforts of businesses. By addressing four key research questions, several significant findings that contribute to both the theoretical understanding and practical application of chatbot technology in online shopping environments have been uncovered. This study found that chatbot marketing efforts positively affect chatbot communication quality (H1). This finding addresses our first research question and underscores the importance of strategic marketing initiatives in enhancing the communication quality of chatbot interactions. In response to the second research question, strong evidence was found that chatbot communication quality positively affects customer satisfaction with chatbot usage (H2). This highlights the critical role of chatbots’ accurate, credible, and competent features in shaping customer satisfaction with chatbot usage. Addressing our fourth research question, it is found that satisfaction with chatbot usage positively affects online purchase intention (H4) and reuse intention of chatbots (H5). Notably, the effect on reuse intention was stronger, indicating that satisfied customers are more likely to continue using chatbots for future interactions. Contrary to our expectations, our third research question yielded an unexpected result. The effect of chatbot usage motivations on customer satisfaction with chatbot usage was not statistically significant (H3). This finding challenges some existing assumptions about user motivations and their impact on satisfaction with chatbot usage. Users with different motivations to use chatbots might have higher expectations for the chatbot’s performance. These expectations may not have been fully met.

5.2. Theoretical Implications

This study makes some valuable theoretical contributions to the existing literature. First, this study combines the impact of both consumer-oriented and business-oriented chatbot activities on consumer satisfaction with chatbot usage, purchase intention and reuse intention. In this way, this study addresses gaps in the existing literature, which often treats business and consumer aspects of chatbots separately. By considering both aspects in a single model, it provides a more complete picture of chatbot effectiveness in the e-commerce environment. In addition, this study offers a comprehensive perspective on the determinants influencing the intention to purchase and reuse of chatbots, extending beyond the scope of the current literature by considering a wider range of contributing factors. The findings of this study are consistent with the previous findings. It was revealed that chatbot marketing efforts positively affect chatbot communication quality. This is partly in line with the findings of Chung et al. [35], who found that chatbot marketing efforts positively affected the communication quality components of accuracy and credibility but did not have a positive effect on communication competence. Similarly, the findings of this research extend the findings of Chung et al. [35] by revealing the impact of all dimensions, including communication competence, which is not supported by Chung et al. [35], of communication quality on customer satisfaction. Furthermore, the positive and significant direct effects of chatbot marketing efforts on dimensions of communication quality is consistent with the findings of Cheng and Jiang [36]. Moreover, our finding that chatbot communication quality positively affects satisfaction with chatbot usage is also consistent with previous studies [35,88,89,90]. Additionally, our results align with the research conducted by Chang et al. [91] and Pereira et al. [92], which demonstrated that satisfaction with the chatbot positively influences the intention to make a purchase. Lastly, our results are in line with those of Lee and Park [90], Silva et al. [93], Zhu et al. [94], and Ashfaq et al., [95] who found that satisfaction has a positive and significant effect on continuance usage intention. Our hypothesis that chatbot usage motivations positively influence customer satisfaction with chatbot usage was rejected, a result that is partially inconsistent with the findings of Rieke [76].

This study revealed that business-based chatbot activities significantly influence customer satisfaction with chatbot usage in online shopping. Conversely, the effect of consumer-based chatbot activities on satisfaction with chatbot usage was not statistically significant. That is, our findings reveal that the communication quality of chatbots has a more substantial influence than customers’ motivation to use chatbots. Consequently, the high level of customer satisfaction with chatbot usage has a greater impact on reuse intention rather than online purchase intention.

5.3. Practical Implications

Considering the findings of the research, it is possible to say that companies running in the e-commerce environment can benefit from the advantages of chatbot technology as a marketing tool to provide customer satisfaction and improve the customer experience while presenting their products and services to users or managing an order process. To benefit from chatbot technology at the highest level, companies should provide an effective communication quality and continuously improve the chatbots offered to users. Marketing managers who want to improve the quality of chatbot communication can provide better service to their customers, especially by tailoring chatbots to individual needs and by improving the problem-solving abilities of the chatbot. Leveraging historical conversation and purchase data, companies can improve user interactions through greater customization. The application of sentiment analysis and natural language processing techniques facilitates more empathetic chatbot responses as well. Furthermore, implementing post-interaction feedback mechanisms provides valuable insights into optimizing customer experiences. Also, companies should design chatbots to resolve customer complaints without the need for human support without wasting time. To do this, the data based on customers’ queries may be collected with various artificial intelligence technologies. Thus, they can train chatbots, mainly on problematic topics, and effectively solve customer problems by saving time. By doing so, online vendors can gain the advantage from the benefits of chatbot technology in the most efficient way, as customers believe that chatbots offer sincere solutions to various problems. Finally, given the stronger link between satisfaction with chatbot usage and reuse intention (compared to purchase intention), companies should view chatbots as long-term relationship-building tools rather than just immediate sales drivers.

6. Limitations and Future Research

Despite the relevance of the present findings, there are some remarkable limitations in generalizing the findings of this study. The sample size is the major limitation of the study. In addition, this study is solely built on consumer data from Türkiye by employing convenience sampling. It was carried out based on limited consumer data gathering through online questionnaires and convenience sampling. In future studies, the opinions of customers who do not prefer to use chatbots and the views of managers need to be explored and evaluated in line with these findings. Thus, it is recommended to include wider and diverse samples from different cultures, demographic profiles, and user experiences, particularly considering the varying levels of technology adoption and internet accessibility across different regions. For example, the sample of the study exhibited a skew towards younger age groups (more than 90% of the participants are between the ages of 18 and 34), potentially reflecting the commonly held perception that older individuals are often less adept at navigating digital technologies [96]. Future research could investigate the chatbot experiences of older age groups to gain a more comprehensive understanding.

In addition, future research could incorporate the perspectives of customers who do not prefer to use chatbots in an e-commerce context and gather insights from managers to gain a more holistic understanding of chatbots in e-commerce. Future studies could also extend the findings to specific types of online shopping, such as online marketplaces, social commerce, mobile apps, cross-border online shopping, or specific products. Finally, live support services provided by real customer relationship management personnel can be compared with chatbots in the context of customer satisfaction and the perceived impact on purchase intention in future studies. As chatbots are increasingly adopted in customer service due to their ability to reduce costs and provide immediate responses, live support remains a critical channel for delivering more complex and personalized interactions [97]. Further research in this domain can equip businesses with the insights necessary to effectively reconcile cost-effectiveness with customer satisfaction. Such insights will enable companies to make informed decisions regarding the integration of chatbot technologies while maintaining the human touch in customer interactions. Finally, the current state of chatbots may also lead to frustration among customers, the dissemination of incorrect information, and overlooked opportunities to find solutions to problems [98]. Future research should prioritize the enhancement of chatbots’ accuracy, contextual comprehension, and capacity to alleviate user frustration. Additionally, it could be imperative to delve into the realm of proactive problem-solving and the integration of human-in-the-loop methodologies. While this study did not directly investigate ethical issues about chatbots, future studies could further consider ethical considerations for chatbot development, such as privacy, data security, and overall trust in e-commerce websites, to augment the overall user experience and effectiveness.