1. Introduction
The expansion of intelligent gadgets and advancements in gesticulation recognition and motion capture procedures have allowed retailers to offer creative solutions to improve consumers’ experiences with their products [
1].
Beauty tech, which utilizes Artificial Intelligence (AI) and AR technologies in industry, aims to personalize and customize the customer experience. It includes features such as coloring cosmetics simulation and personalized recommendations. International brands like L’Oréal have embraced these to gain market share [
2].
Virtual reality (VR) and AR have gained a significant position in online retail, changing individuals’ purchasing habits [
3,
4]. Sellers have incorporated immersive technology components into their e-commerce websites and have developed mobile apps which allow clients to virtually test products such as makeup, sunglasses, furniture, and clothing, bridging, thus, the gap between pre- and post-consumption experiences [
3,
4].
AR co-occurs with the natural world via digital content. VR is based on avatars and refers to real content over the user’s virtual background [
1,
3,
5].
On the one hand, VR is appropriate for cases where user body presentation is not crucial, such as gaming or flying simulators. On the other hand, AR is more suitable for cases where self-representation is essential, like trying on body-involving products. Therefore, AR is the landmark of our study because it merges virtual objects/products with the client’s natural world, allowing them to interact accurately with the products [
1].
AR in e-commerce has proliferated, with the AR market estimated to be USD 88.4 billion by 2026 (see in [
6]). AR emotional apps activate a wow effect, delivering surprise [
7].
Even though the adoption of AR in e-commerce is still limited, it has become a significant component for leading retailers such as IKEA, Sephora, L’Oréal, and Ray-Ban [
3,
4,
8]. AR improves users’ perceptions and influences e-commerce behaviors by delivering an authentic product experience. The development of AR services for mobile devices has improved online shopping experiences, making them more accessible to a broader audience [
8]. Multiple industries have already adopted AR to boost customer experience satisfaction and make products available 24/7.
Although AR has only recently gained popularity and accessibility due to the appearance of smartphones equipped with the necessary hardware [
8,
9], AR applications are currently being developed within various industries and have the potential to substantially transform traditional retail and marketing activities [
7,
9,
10]. AR is not just a functional technology but also a persuasive one, creating value through enjoyment, efficiency, visual pleasure, and playfulness [
9].
AR technology will soon significantly reconfigure shopping and experiences in e-commerce worldwide [
9]. AR has the potential to revolutionize e-commerce by enhancing and personalizing the shopping experience. Companies like Google, Apple, Facebook, Alibaba, Microsoft, HTC, Sony, and Samsung have invested heavily in AR technology [
11]. AR has evolved into a superior technology for successful businesses, and it has become indispensable for people’s everyday life in commerce routines [
1]. Therefore, the possibility of AR integrating digital details into the real world has been challenging for both academia and industry in trying to discover and predict its effect on customers’ perceptions, adoption intentions, and use [
9]. Major global retailers have already started implementing these tools for online and offline shopping experiences, but these systems are still in the earlier phases of adoption [
12]. While there is growing interest in AR within the retail and e-commerce industry, there is still a need for a better understanding of its value for consumers and brands to encourage further investment [
13].
This paper aims to study if the use of AR apps in online women’s makeup stores (with products like lipstick, lip liner, eyeshadow, eyeliner, mascara, foundation, powder, and hair color) stimulates the desire to buy makeup products (and how it does) and improves the e-shopping process.
This paper uses the term AR app when referring to applications incorporating AR technologies for virtual makeup and hair coloring. Also, the term makeup refers to makeup and hair coloring.
Ref. [
14] is one of the most remarkable recent reviews of AR in e-retailing. According to [
14], the AR literature uses 30 different classical and modern theories. The most employed classical theoretical frameworks for explaining the drivers behind the adoption of AR in e-commerce are SOR theory, the technology acceptance model (TAM), the situated cognition theory (related to consumer psychology in an AR environment), and the uses and gratifications theory. Besides many other factors, the key factors driving the consumer adoption of AR in e-commerce and online shopping are related to technology attributes, utility, hedonism, immersion, and social factors [
14]. For future studies related to social factors, ref. [
14] proposes the co-creating effect, social sharing, and social value. Despite increasing curiosity, examination of challenges to AR adoption, like practical and psychological obstacles, is limited, impacting consumer adoption of AR in e-shopping. Also, research on social aspects of AR could be more extensive [
14]. Starting from the previous research based on the SOR paradigm [
1,
3] and models such as the Unified Theory of Acceptance and Use of Technology (UTAUT) [
9], our study aims to identify the determinants of users’ intentions to adopt AR apps in e-commerce and stimulate users’ intention to buy products. In this paper, we consider the following factors: social value, perceived hedonistic value, fit confidence, perceived utilitarian value, immersion, and innovativeness. Fit confidence is one sub-dimension of perceived benefits that can better explain the intention to adopt virtual try-on apps (VTOs) [
5].
AR apps and e-shopping have distinct fields, but their intersection creates a unique e-commerce experience. While the primary goal of e-shopping is product purchase, AR apps are tools meant to enhance this experience. Past studies on this intersection have been more tech-adoption-focused than commerce-focused. In this study, we also add emphasis to the components that stimulate the purchases side. In any e-commerce context, if site visits are without sales, the problem may be related to the products rather than the AR technology itself. Besides the constructs from the SOR, our research introduced confidence, innovativeness, and social value as influential factors.
Consequently, we intend to provide pertinent answers to the following research questions:
RQ1: What factors influence the intention to use AR apps in e-shopping?
RQ2: What are the relationships between the determinants mentioned in RQ1?
This paper is structured in six sections, as follows.
Section 2 provides a general background based on a consistently updated literature review.
Section 3 describes the methodology and data collection.
Section 4 presents the results.
Section 5 presents discussions.
Section 6 presents the conclusions.
3. Methodology and Data Collection
The SOR framework, often used in retail studies, examines how environmental cues influence people’s thoughts and feelings, leading to certain behaviors like purchasing. This has been observed in diverse shopping settings, from physical stores to online platforms. With the shift towards experiential retail, researchers are applying the SOR model to understand how such retail elements affect consumer behavior. Notably, the SOR model is particularly effective in understanding consumer behavior in AR and VR environments. While the early literature mostly discussed AR’s potential for enhancing user experiences, the current emphasis is on understanding the impact of environmental factors in AR on those experiences. The SOR framework is endorsed as an apt tool for explaining the initial adoption behaviors of emerging technologies like AR and VR [
12].
We used an online survey with constructs used in the SOR framework (continued usage intention, perceived hedonistic value, perceived utility value, immersion) and fit confidence (from [
5]), social value (from [
4]), and innovativeness (from [
9]). For CUI, we used seven items: three from [
3], two from [
12], and one each from [
5,
7] concerning continued use of the AR app and the purchase of products. For SV, we used five items: three from [
4] and two from [
13] concerning improving personal social image (using makeup products selected with AR apps) and sharing obtained image content (obtained with the AR app) on social media. For FIT, we used five items from [
5] concerning makeup fit, both real (after shopping) and virtual (during shopping), after usage of the AR app. For PUV, we used five items: two from [
12], two from [
66], and one from [
3] concerning if AR app usage for makeup shopping is practical. For IMM, we used six items: three from [
8] and one each from [
1,
7,
67] concerning the immersion effects of AR apps on customers. For INT, we used three items from [
9] concerning the client’s familiarization with new technologies.
The analysis in [
14] of the respondent profiles in the published articles revealed that studies have primarily examined the AR usage behavior of young populations aged 16 to 35. A few studies represented only female users in their samples as they tested AR for categories related to makeup and women’s clothing [
14].
Regarding data collection, throughout February 2023 we promoted the use of AR apps among our students from the Faculty of Economics and Business Administration, West University of Timișoara, but also to our friends, and we asked them to encourage their friends to use such applications as follows:
- -
We gave an introductory presentation on virtual try-on tools (website or mobile apps) for makeup, glasses, clothing, and furniture.
- -
We presented the pictures obtained from [
68] (we uploaded a photo from [
68] and used lipsticks, eye shadows, and hair colors).
- -
We invited our students and friends to use (with a live camera or photo) an AR app for a period of time [
68], or other AR apps for makeup and hair coloring.
At the end of May 2023, we returned with an online questionnaire and asked them (students and our friends) to complete it only if they used AR apps. The questionnaire from this study included items from previous research (see
Appendix A) using 5-point Likert scales. This study obtained 394 valid responses from 427 Romanian women aged between 18 and 53 (see
Table 1).
According to [
2], the countries where consumers are the most familiar with smart mirrors are China, South Kora, the United States, the United Kingdom, France, and Germany, with percentages between 24% and 52% of the female residential online population. Previous research papers [
1,
13] and statistical studies report results from the makeup market for an exclusively female population. In [
3], the sample is 20% male and 80% female. With a confidence interval of 95%, considering a population proportion of 24% (for the case of Romania being less than 24%) and a precision level of 5%, we obtained that 280 was the minimum value for a representative sample.
In the AR in e-retailing literature, the most widely used technique is structural equation modelling (SEM) (n = 26), comprising the partial least squares method (PLS) (n = 17) and the covariance-based approach (CB) (n = 9) [
14].
Composite-based models provide excellent outcomes in PLS-SEM, better, for example, than Covariance-Based Structural Equation Modeling (CB-SEM) [
69]. We used the PLS-SEM method in Smart PLS 4 software for the model analysis. The measurement properties analysis assessed the reliability and validity of the measurement model.
4. Results
In PLS-SEM, outer loadings refer to the relationships between items and their factor. Cross loadings refer to the relationships between observed items and factors other than the ones they measure. In
Figure 2 and
Table 2, we can see the values for outer loadings (INT4 is 0.685, all the other values are between 0.714 and 0.909—the items’ reliability), and in
Table 2 the values for cross loading.
We examined the reliability with Cronbach’s alpha and composite reliability, and all constructs showed high internal consistency with values between 0.844 and 0.933. The average variance extracted (AVE) was used to assess convergent validity, and all constructs exceeded the minimum threshold of 0.5, indicating good convergent validity (
Table 2).
We conducted a discriminant validity assessment by comparing the AVE values to the squared correlations (
Table 3).
In
Table 4, the heterotrait–monotrait ratio (HTMT) matrix values are less than 0.850, confirming the discriminant validity [
70].
Usual assessment measures include R2, the determination coefficient; the variance inflation factor (VIF); the statistical significance and relevance of path coefficients; and the predictive relevance (Q2).
The VIFs are between 1.522 and 4.808—obtained through multicollinearity examination. The R
2 value is 0.743 for the CUI; it is 0.702 for PUV; it is 0.619 for IMM; it is 0.587 for SV; it is 0.570 for PHV; and it is 0.406 for FIT (
Figure 2).
SV has the most considerable direct influence on CUI (0.320), followed by FIT (0.256), PUV (0.247), and IMM (0.180). FIT has the most considerable total (direct and indirect) influence on CUI (0.643), followed by INT (0.565), IMM (0.458), SV (0.320), PHV (0.297), and PUV (0.247). IMM has the most considerable influence on SV (0.766), followed by FIT (0.469), INT (0.414), and PHV (0.378). FIT has the most considerable influence on IMM (0.612), followed by INT (0.540) and PHV (0.493). INT has the most considerable influence on PUV (0.698), followed by FIT (0.514), PHV (0.353), and IMM (0.135). INT has a positive influence on FIT (0.637). INT has the most considerable influence on PHV (0.638), followed by FIT (0.525) (see
Table 5).
With 5000 iterations, we performed in SmartPLS a bootstrapping technique to explore the R
2 statistical implication. All hypotheses are supported—the values in
Table 6 are statistically significant. In
Table 6, we present the path coefficients (all are greater than 0.100, which means that they are significant), t statistics (all values are greater than 1.96),
p values (all values are less than 0.05), and the remarks on each hypothesis.
In
Table 7, we deliver the indirect effects of the factors. In
Table 7, we present the path coefficients (all are greater than 0.100, which means that they are significant), t statistics (all values are greater than 1.96), and
p values (all values are less than 0.05). These values from
Table 7 show that the indirect effects are significant.
In
Table 8, we can see the data for the total effects. In
Table 8, we present the path coefficients (all are greater than 0.100, which means that they are significant), t statistics (all values are greater than 1.96), and
p values (all values are less than 0.05). These values from
Table 8 show that total effects are significant.
In
Table 9, we present the effect size (f
2), which refers to the magnitude or strength of the relationships between variables (≥0.02 is small; ≥0.15 is medium; ≥0.35 is large) [
71].
SRMR (Standardized Root Mean Square Residual) evaluates the goodness of fit between the model and the observed data; its value is 0.079. Q
2 is high for PUV (0.473), FIT (0.400), and PHV (0.401); moderate for IMM (0.240) and CUI (0.212); and low for SV (0.101), showing the high predictive relevance of the model. The goodness of fit (GoF) is the geometric mean of the average communality and the average R
2. With the AVE values from
Table 2, the average communality is 0.685, and the average R
2 is 0.605. The GoF is 0.643, which exceeds the cutoff value of 0.36 for a large effect size and can be considered satisfactory. With the data and observations from this section, we can confirm the utility of the model in
Figure 1. Descriptive statistics of the indicators are in
Appendix B.
The current research aimed to understand how consumers experience AR apps and what factors influence their intention to reuse them in e-shopping. The findings of the current research indicate the following:
All the examined variables significantly affect (directly or indirectly) the continued usage intention of AR apps in e-commerce.
The current research found that FIT is the most significant variable positively affecting the CUI (the path coefficient is 0.643, see
Table 4). FIT influences CUI directly and indirectly, with the indirect impact (0.387) being more substantial than the direct impact (0.256). FIT (see
Table 4 and
Figure 2) also influences IMM (0.612), PHV (0.525), PUV (0.469), and SV (0.469).
Thus, we can conclude that it is essential that the makeup selected by the AR application suits the clients well; the AR application helps to choose shades that suit customers; the customers are satisfied with the photos obtained within the AR application; and customers are satisfied with the products offered by the vendor, which they can try virtually in the AR application.
The results show that it takes more than just the fact that the AR app helps users find the right products. Among AR app users, those who feel immersed and delighted when using AR apps, those who find usefulness in using makeup, as well as those who have positive feedback from acquaintances, friends, or family (by using makeup products) will be more tempted to continue using the application and buy makeup products.
The result obtained regarding FIT also suggests that if a customer does not find what seems fit, she will not be interested in continuing to use the app or buy products, even if the AR app is technologically excellent. For example, let us say that a customer wants to buy lipstick, and only light shades are suitable for this customer, and the online store sells only lipsticks in dark shades. In this case, the customer is not motivated to return to the website or try the AR app because she will never buy lipstick in dark shades.
INT is the next factor influencing CUI (the path coefficient is 0.565, see
Table 4). It only indirectly influences the CUI. INT influences all variables: PUV (0.698), FIT (0.637), PHV (0.638), CUI (0.565), IMM (0.54), and SV (0.414). Even if it is about e-commerce, the AR app is central to our study. This result shows us that it is essential for customers to be passionate about IT technologies and to be up to date with the latest developments in the field. Some women may want to wear makeup and buy makeup online. However, the fact that they feel far from the technology or using the AR app will keep them away from using it and maybe even the online store more generally.
INT does not directly influence CUI. If someone is passionate about IT and the latest developments in the field, they may still want to avoid wearing or buying makeup. The results from
Figure 2 and
Table 4 tell us (about the women that are the potential customers who will continue to use AR apps and will buy makeup products) that they are passionate about using AR apps, found suitable products, feel immersed and excited, find usefulness in using makeup, and receive positive feedback from people around them.
IMM influences CUI (0.458 in
Table 4) directly and indirectly. IMM strongly directly influences SV (0.766 in
Table 4). The indirect effect of IMM is greater than the direct effect on CUI. Some people may feel immersed using AR apps. Our results show that they will become buyers and use the AR app again, especially if they receive positive feedback from those around them. Some people may be impressed by the AR app; however, in the absence of social feedback, these people may eventually become disinterested in both makeup and using the AR app.
The next factor influencing CUI is SV (0.3.20 in
Table 4). SV is the factor with the more substantial direct influence on CUI. The people who share the images obtained from the AR app and share their experiences with those around them are among the people interested in using the AR app in the context of e-shopping. According to the Unified Theory of Acceptance and Use of Technology (UTAUT), social influence directly affects behavioral intention, shaping potential users’ attitudes towards technology [
8].
PHV indirectly influences CUI (0.297 in
Table 4). Even if using the AR app delights some people and makes them feel excited or enjoy themselves, the AR app user becomes a makeup product buyer only if they feel immersed in the application and perceive the utilitarian value. Also, PHV indirectly influences SV (0.378 in
Table 4).
PUV indirectly influences CUI (0.247 in
Table 4). This result tells us that people who can combine various products and easily choose what suits them better, who feel that the AR app is necessary or practical and makes shopping effective, will become buyers of makeup products.
5. Discussions
As we saw in the literature review, the SOR paradigm is widespread in research papers on adopting AR apps.
In addition to the constructs already established in SOR (CUI, IMM, PUV, PHV), using other research works in the field, we added three other constructs, SV, FIT, and INT, out of the desire to understand, in the makeup buying process on online,
- -
Customers want to adopt AR technologies;
- -
How this technology determines the purchase of products (or improves the purchase process).
From the presented results, we noticed that if with the AR app customers find the right products more efficiently and feel that using them improves their social image, they are more tempted to adopt AR technology and be buyers. This is the part that best reflects the contributions of our study. This model can provide comparable results depending on the type of commerce (for example, fashion, watches, or jewellery). However, we expect that other e-commerce domains are replacing SV or FIT. For example, if we are buying furniture, SV may not necessarily be an influencing factor. Other constructions aimed at that type of trade may be more suitable. For example, comfort might be better than SV when discussing furniture. Even so, our model suggests that adding constructs from established trade models will improve models of technology adaptation by providing answers to purchasing problems—which is the primary objective of trade activity.
The results of INT show us that ignorance and misunderstanding of technological functionalities could keep people who are interested in buying products (in the case of our study, makeup products) away from e-commerce. In other words, many women may want to buy makeup, but they do not access such applications due to unfamiliarity with technology or AR technology.
Based on the SOR paradigm, ref. [
3] studies how experiential AR apps affect buyers’ experiential value while improving CUI. The results indicate that the characteristics of an experiential AR app have a more significant positive effect on hedonistic value than utilitarian value; in turn, only hedonistic value positively impacts continued usage intention. Perceived customer support positively moderates the effect of hedonistic value on CUI. The model in
Figure 1 uses some items of the constructs presented in [
3]: PUV, PHV, and CUI. The differences between our results and those of [
3] are as follows:
We found that PUV influences CUI. In [
3], all the responders were buyers of makeup products who used the AR app. For our respondents, such an application proves useful for some (who want to buy makeup), but only for some. So, on a more diversified framework of respondents, we can conclude that PUV influences CUI.
Ref. [
3] used experiential AR app features to see how they influence CUI. We replaced these factors with FIT to see the intention to continue using the application to buy products. Ref. [
3] was focused on usage of the AR app by people already buying products. We moved the study towards shopping, and from here we conclude that an AR application can be excellent from a technological point of view; however, if people do not find the right products, they will not make purchases. In [
11], both PHV and PUV positively influence CUI.
Refs. [
1,
8] use IMM and find that this construct has a positive influence on AR app adoption. Ref. [
48] find that it influences enjoyment but does not affect AR app reuse intention.
Built on human value orientation theory [
72] and consumption value theory [
53], the conclusions of the analysis in [
4] indicate that usability value, playful value, and visual magnet value positively impact client fulfilment and that social value does not influence customer satisfaction. Even though they did not obtain a positive result regarding social value, the authors in [
4] recommended it for future studies, and they were right. Our study shows that not only does SV influence CUI, but among the factors that influence it directly, it is the one with the most significant impact. The different results between our study and those of [
4] could be due to several reasons. We studied makeup. In [
4], the study covered different types of products. SV is essential when discussing makeup; it may not influence CUI when discussing other types of products. Cultural differences could be another reason. Future studies will be able to analyze these aspects.
Inspired by [
5] (for fashion), we introduced FIT (for makeup). Like [
5], our results show that FIT is a construct to consider in studies of AR apps in e-commerce.
From the beginning of the paper, we flagged [
14] as an exceptional study in the field. However, neither this paper nor the many he referred to considered FIT or a comparable construct. Thus, the work in [
5] shows us that the specialized literature is still richer than it seems at first glance and may contain other excellent results, but which may not have come to light for various reasons.
Once again, the message of our work is that the studies in this field focus much more on trade, not only on technology, and we also recommend the inclusion of constructs that have been successful in traditional trade models.
Ref. [
9] used INT and found that INT does not affect behavioral intention (in a study on adopting AR apps in shopping malls), and only affects enjoyment. Also, in our case, INT influences PHV but indirectly influences CUI. The difference in results may come from the fact that in our study, the user, if he buys products, is obliged to use the technology, which is not mandatory in [
9], related to the use of AR apps in shopping malls.
6. Conclusions
AR has become a superior technology for successful e-commerce. This study attempted to understand the factors influencing the intention of using AR apps in makeup e-shopping and purchasing products.
It offers some results that can be useful both for practitioners and future research. The present study, using CUI, IMM, PHV, and PU—constructs used in SOR—demonstrates that FIT, INT, and SV are vital variables influencing CUI when using AR apps in makeup e-shopping.
6.1. Theoretical and Managerial Implications
This study identifies a gap in the research of AR apps in e-commerce: topics on commerce remain unexplored. The study also uses rarely used variables (FIT, SV, and INT), forming an integrated structure including both popular and other variables. It guides researchers in choosing pivotal factors for further investigation and using this framework for hypothesis development and empirical validation.
This research contributes to developing a theoretical framework to examine the impact factors on the continued usage intention of AR apps in makeup e-shopping. Researchers could use the proposed conceptual framework to investigate the topic further and apply it to other areas using AR apps in e-commerce (for example, in fashion).
From a managerial viewpoint, AR-based apps create new possibilities and permit organizations to maintain or enhance their trade competitiveness. In e-tailing, AR offers cost-effective options for classic product trials, like free return approaches.
By allowing consumers to visualize products in a more personal context, AR apps can increase their confidence in purchasing decisions. Factors like FIT and INT can influence, engage, and attract customers. Hence, investing in such technologies can offer a more exciting experience to users. This research holds several implications for retail brands, emphasizing the importance of understanding the critical determinants of CUI. Retailers should publicize information on the use of their AR app (on their websites and social media), help with tips (e.g., video, posts, or chat) for potential customers to find the right makeup, and arrange themselves so that they are accepted and appreciated by those around them to improve the way they are perceived and create good impressions to encourage customers to share the beautiful pictures obtained with AR applications on social networks with their friends and acquaintances.
6.2. Limitations and Future Research
In this paper, we talked about virtual try-on for makeup and hair coloring. In AR apps with other types of products, we can expect the results to be significantly different, as our study group included only women. The respondent groups must include men for other products (e.g., apparel, sunglasses, and furniture).
The study was conducted on Romanian consumers, limiting the generalizability of the results due to cultural differences. Future cross-cultural research could offer additional insights.
AR generates more positive experiences (novelty, surprise, inspiration) and fewer negative consequences (information overload, distraction, privacy risks). Information overload can generate distractions, decreasing purchase intentions via AR apps [
7]. Technology anxiety can lead to risk-avoidance behavior [
12]. A limitation of this paper is that it does not detect adverse effects and how they affect CUI, e.g., perceived intrusiveness [
13,
73], perceived social risk, financial risk, physical risk, performance risk, and psychological risk [
74].
Our study did not include any effects on brand attitude. Interactivity and vividness affect immersion and brand loyalty [
74]. Future research could examine how AR app try-on experiences affect online customers’ purchasing decisions or willingness to pay for unknown brands or different product categories.
In a watch e-commerce context, the findings showed that individuals without prior AR experience were more likely to experience immersion and a subsequent feeling of ownership, suggesting that familiarity with new stimuli reduces the novelty and, thus, the motivation to be mentally immersed [
8]. For us, the context is makeup e-commerce, and we have seen that INT has a considerable impact on IMM. Here, we notice some differences in results that could come from several causes: either there are differences between the types of commerce (we buy some products much less often than makeup), or, after extended use of AR apps, they are not so captivating anymore. Future studies could investigate if prior AR experiences reduce the motivation to be mentally immersed in the case of specific types of products (e.g., watches in [
8]) or if it generalizes.
Despite the limits discussed above, the current study is one more step towards a better understanding of the factors influencing consumers’ behavior towards AR apps in e-commerce.