Environmental Impact Perspective Sustainable Online Textile Retailing: Harnessing Augmented Reality-Based Digital Twins in Bangladesh

Graphical Abstract

Click to enlarge

Highlights

• This study introduces a novel strategy for the Textile retail industry by skillfully fusing Augmented Reality and Digital twins with popular trends to raise consumer consciousness and encourage participation in sustainability.
• An advanced AR-based digital twin system is introduced, allowing online consumers to virtually experience the environmental impact of products throughout their lifecycle.
• Through applying the Linear Regression Analysis model for consumer behavior analysis to user testing, the study shows notable progress in increasing consumer awareness, swaying purchasing decisions, and promoting sustainable activities in the fast fashion industry.

Data Collection

The groundwork is laid by gathering data from various sources. Open LCA databases provide information on textile waste generated throughout the clothing life cycle. Additionally, data on environmental impact factors like water and energy consumption, chemical use, and waste littering is collected. Consumer behavior and purchase decisions are also captured through surveys, interviews, or observation. Software like Power BI might be used to organize and visualize this collected data.

Digital Twin Integration

A core aspect of this methodology is the creation of a digital twin customer. This virtual representation is built based on real-world data like a customer’s gender, size, preferences, and region. This digital twin is then integrated with an AR 3D model, allowing for a virtual try-on experience. Customers can virtually try on clothes within the digital twin system before making a purchase.

Impact Assessment

The environmental impact of this digital twin integrated fashion retail system is then assessed. This likely involves utilizing the previously collected data on environmental impact factors, allowing researchers to estimate the environmental footprint of the proposed system.

Comparison and Analysis

Finally, the research compares the environmental impact of the digital twin system with the traditional retail model. The traditional retail system’s impact isn’t explicitly measured, but is likely assumed to be higher due to physical production of clothes that might not be purchased. This comparative analysis is the heart of the research, highlighting the potential environmental benefits of the digital twin approach in fashion retail. By following these steps, the research aims to understand how a digital twin integrated system can potentially reduce the environmental footprint of the fashion retail industry.

Scope and Goal Definition

The goal of this back-end study was to conduct a life cycle assessment (LCA) of four different types of textile products in Bangladesh, specifically focusing on the environmental impacts associated with the production and use of eco-t- shirts made from organically grown cotton and processed with green dyeing. The LCA compared the environmental performance of eco-t-shirts to that of conventional T-shirts. The assessment will consider the following impact categories: global warming potential, acidification, aquatic and terrestrial eutrophication, and photochemical ozone formation. Inventory Analysis System Boundary The study was allowed to consider the entire life cycle of four individual items (T-shirt, jersey, soft shell jacket, and punjabi), including cotton cultivation and harvesting (for cotton items), fiber production, yarn manufacturing, fabric preparation, fabric processing (including bleaching and dying sub-processes), and fabrication of the final product.

Data Collection Data on the inputs and outputs associated with each life cycle stage were collected. This includes information on energy consumption, water usage, chemical inputs, and waste generation. The functional unit for the assessment was one unit of T-shirt. The data collected were analyzed using LCA methodology, following the ISO 14040:2006 and 14044:2006 standards. Inventory data was aggregated and normalized to the functional unit.

Impact Assessment and Visualization

Selection of Impact Categories The impact categories to be assessed include global warming potential (GWP), acidification (AP), aquatic eutrophication potential (AEP), terrestrial eutrophication potential (TEP), and photochemical ozone formation potential (POFP). Characterization factors were assigned to each inventory flow to quantify its impact in the specific impact categories [7].

Calculation of Impact Factors The impact scores for each impact category were calculated by multiplying the inventory flows with their respective characterization factors. The results were interpreted to assess the environmental performance of the four items compared to one another. LCA identified that soft shell jacket and jersey production had larger environmental impacts on AEP, GWP, AP, POFP and TEP, though lesser than expected, which may be due to the low uptake of pesticides and fertilizers in the cotton planting stage, resulting in a large amount remaining in the soil with 20–30% being leached into nearby water bodies or groundwater. Significant environmental impacts derived from wastewater discharge from printing and dyeing, air pollutants from fossil-fuel-based energy and transport, and large amounts of water consumption used in steam production. Examining the graphs (Figure 2), from (a) to (e), through the lens of environmental impact reveals a multifaceted picture. While aquatic eutrophication Potential (AEP) and Global Warming Potential (GWP) life cycle stages (panels a and c) generally exhibit higher values compared to Acidification Potential (AP) alone (panel e), the specific values within each stage demonstrate significant variation. For example, in panel (a) (AEP - Jacket), the “Use” stage boasts a value of 4.5, whereas it plummets to a considerably lower 2.8 during “Disposal.” This substantial drop suggests a potential link between specific life cycle stages and environmental impact, with disposal potentially having a much higher impact for jackets under the AEP analysis. Furthermore, there are variations across items even within the same AEP or GWP measurement (panel d - GWP). Here, “Jersey” during “Use” has a value of 4.2, while “Jackiet” during the same stage has a much lower value of 3.1. This difference indicates that the product type might influence the environmental impact independent of the life cycle stage under consideration.

The stage of those results in the most significant environmental impacts was printing and dyeing. It was found that AP, GWP, and POFP resulting from atmospheric pollutants such as NOx and SO2 were the key environmental impact categories in the LCA of the whole production system [16]. However, the factory noise, which had been excluded from the analysis, was much higher than acceptable for human health. Therefore, the concept of sustainable manufacturing should expand its scope to include the occupational health of employees. Furthermore, the key challenges of developing the system management concept are the availability of life cycle inventory data and the training and support process for a large user base. Green design will need broad adoption by the industry.

Click to enlarge

UI Design

Personalization with Digital Twins The application incorporated personalization with Digital Twin-generated outfits based on the user’s preferences and previous shopping behavior. Users could choose their product attribute preferences depending on sustainability metrics, price range, brand preferences, and specific product features.

Product Information The AR app had a feature to display virtual information overlays on product pages, providing users with detailed information about the product attributes they are interested in. This information included environmental impact potential, customer reviews, and any other relevant information that can aid in decision-making.

Comparative Analysis The application enabled users to compare multiple products side by side, allowing them to evaluate and contrast various attributes, such as price, quality, environmental impacts, and customer ratings. This feature helped users make informed choices based on their preferences and priorities.

Real-Time Updates The AR app provided real-time updates on product availability, pricing changes, environmental impacts, and any relevant promotions or discounts, but in this case, to get the perfect effect of environmental impact perspective effect on consumer behavior the rest of the metrics were kept standstill. This feature ensured that users have the most up-to-date information on potential environmental impacts while making their purchasing decisions.

Interactive Visualization The AR app was able to allow users to provide feedback, ratings on products they have purchased. This feedback contributed to the overall product rating, mood scores and helped future users make informed decisions.

Data Collection The groundwork is laid by gathering data from multiple sources to assess both the environmental and behavioral dimensions of the study. For the environmental assessment, data was obtained from open LCA databases, which provide detailed information on textile waste generation, energy and water consumption, chemical use, and other emissions associated with each stage of the clothing life cycle. Additionally, consumer behavior and purchase decisionmaking patterns were explored through a mixed- methods approach, including surveys, interviews, and observation. Tools such as Power BI were used to organize, analyze, and visualize the collected data. All participants were informed of their rights, and informed consent was obtained by ethical guidelines, although no formal IRB system exists in the study’s country of origin. Instagram users were selected as the target population for this study. A linear Regression model was used to prepare questionnaires.

Based on this, model the five dimensions and concepts were identified, and the questionnaires were ready basis on this concept. A sample size of 150 participants per group (AR- based and traditional) was targeted to ensure sufficient statistical power. Data collection was conducted through two main methods:

Online Survey: A structured questionnaire was developed based on the constructs derived from the Linear Regression model, including perceived eWOM, perceived information clarity, perceived user engagement, perceived ease of use, and perceived overall satisfaction. Based on the survey items provided (see Table 1), here are the corresponding questionnaire items for each construction:

Semi-Structured In-depth Interviews: In addition to the online survey, semi-structured in-depth interviews were conducted to gain a more in-depth understanding of the profiles and behaviors of AR-integrated environmental impact perspective users. These interviews were conducted through Google Meet by asking questions like those of our target group individually (see Table 2). The target group is defined as 22- to 24-year-old university-level students who are very much aware of social media and online shopping. Still, the sample size has students of different stages in their lives. The questionnaires were designed based on their propensity to purchase fashion apparel from AR-centric online and non-AR-centric. The interview duration was 10 minutes, and at the end, they also shared their interview experiences. Their interview was also recorded by obtaining their consent for further analysis.

Sampling

A purposive sampling technique was used to select participants for the online survey and the in-depth interviews. For the online survey, a convenience sampling approach was employed, targeting Instagram users who were active in online shopping. The sample size for the survey was determined based on statistical considerations to ensure an adequate representation of the target population. Participants were randomly assigned to AR-centric shopping platforms in four different stages. Table 3 reflects the demographic breakdown for a sample size of 150 people.

As we can see from the demographic table, in a sample of 150 individuals, the gender distribution is 45% male (67) and 55% female (83). Most participants are aged 25- 34 (40%, 60), followed by 18-24 (30%, 45). Education levels show that 50% (75) have undergraduate degrees, 30% (45) graduate, and 20% (30) have high school education. Control level awareness is primarily medium (50%, 75), indicating moderate awareness of sustainable online textile retailing. In short: Pre-shopping Survey: Participants completed an initial survey measuring baseline mood and demographic information.

Shopping Experience: Participants were engaged in a shopping task on the assigned platform. They were instructed to browse and make purchases as they normally would.

Mid-shopping Survey: Participants completed a short survey midway through the shopping experience to assess their current mood.

Post-shopping Survey: Upon completing the shopping task, participants completed a final survey measuring their mood, satisfaction with the shopping experience, and details of their purchases (e.g., the total amount spent, number of items bought, impulse purchases).

Data Analysis

The study can be grounded in environmental psychology, specifically the Mehrabian-Russell model, which posits that environmental stimuli affect individuals’ emotional states and influence their behaviors. In this context, AR is an enhanced stimulus in the online shopping environment. Three types of analysis will be based on three metrics.

Mood Analysis: Mood changes were analyzed using repeated-measures ANOVA.

Satisfaction Analysis: Differences in satisfaction between the two groups were assessed using independent t-tests.

Purchasing Behavior Analysis: Purchasing data will be analyzed using independent t-tests and regression analyses to identify the influence of AR on spending behavior.

Measures: The following necessary measures for the analysis were considered.

Mood: Assessed using the Positive and Negative Affect Schedule (PANAS) at three points (pre-, mid-, and post- shopping).

Customer Satisfaction: Measured using a validated customer satisfaction scale.

Purchasing Behavior: Recorded through the e-commerce platform, including total expenditure, number of items purchased, and incidence of impulse buys.

Hypotheses: Based on the experiment, three hypotheses will be developed using three types of consumer behavior analyses to conclude.

H1: AR-based online shopping led to higher levels of customer satisfaction compared to traditional online shopping.

H2: Customers’ mood improved more significantly during AR-based online shopping than during traditional online shopping.

H3: AR-based online shopping will result in higher purchasing behavior (e.g., spending more money, making more impulse purchases) compared to traditional online shopping.

Consumer Behavior Analysis

Consumer behavior was analyzed through a comprehensive study (see Table 4) involving both AR-based and non-AR-based online shopping experiences. The data was gathered using the linear regression and Mehrabian-Russell model, focusing on various aspects such as information evaluation, economic evaluation, product evaluation, affective response, risk assessment, and decision-making.

User Feedback and Insights

The results (see Table 5) indicate that users found the AR-based platform significantly more engaging and informative than traditional online shopping methods. The immersive nature of AR allowed users to better understand the environmental impact of their purchases, leading to more informed and sustainable decision-making.

The feedback gathered from user surveys and interviews highlighted the effectiveness of the AR-based system in enhancing user engagement and satisfaction. Participants appreciated the detailed information provided through AR overlays, which included potential environmental impacts, digital twin-based customization and other relevant product details Figure 3.

Click to enlarge

Mood Scores The mood scores showed a significant improvement in the AR-based shopping group compared to the traditional shopping group. The AR-based platform enhanced the shopping experience, likely due to the interactive and immersive nature of AR, which made the shopping process more engaging and enjoyable. The repeated measures ANOVA indicated a significant interaction effect between the type of shopping and the time points (p < 0.05). The mood scores were measured at three points: pre-shopping, mid-shopping, and post-shopping. The results are summarized in Table 6.

Examining the line graph (see Figure 4) titled “Mood Score Before, During and After Shopping” unveils a potential mood-boosting effect of AR-based shopping compared to traditional methods. The graph suggests a positive trend in customer mood scores across all three stages (before, during, and after) when using AR. This implies that AR might elevate a customer’s mood throughout the shopping journey from the data, we see a potential mood-boosting effect associated with AR-based shopping compared to traditional methods. Customers who used AR for shopping tended to have higher average mood scores across all three stages (pre-shopping, mid-shopping, and post-shopping) than those who shopped traditionally (see Table 1). For example, the average pre- shopping mood score for AR shoppers was 2.95, whereas for traditional shoppers, it was slightly lower at 2.91. This suggests that AR and Digital twins’ customization might elevate a customer’s mood throughout their shopping journey [17]. However, it’s important to consider some limitations. The relatively small standard deviations (around 0.7) indicate that the mood scores within each group (AR and traditional) didn’t vary much. This could involve exploring how individual personality traits or product types of influence responses to AR experiences and whether the positive mood associated with AR translates into long-term benefits for brands.

Click to enlarge

Customer Satisfaction

Customer satisfaction scores were significantly higher in the AR-based shopping group. This suggests that AR features, such as visualizing products in a real-world environment, contribute positively to customer satisfaction.

The independent t-test results confirmed the significant difference in satisfaction between the two groups (p < 0.01). Customer satisfaction was measured by post-shopping. The results are summarized in Table 8 and visualized in Figure 5, Table 7.

Click to enlarge

The graph (see Figure 6) analyzes customer satisfaction and environmental impact based on shopping experience (AR integrated vs. traditional). Customer satisfaction is higher for AR-integrated shopping than for traditional shopping. There is a positive correlation between environmental impact and customer satisfaction with AR-integrated shopping and a negative correlation between environmental impact and traditional shopping satisfaction. It’s important to note that the impact on satisfaction is likely due to other factors like Digital twin-oriented customization besides environmental impact, which is just one perspective measured in the graph. Overall, the graph suggests that AR-integrated shopping may lead to higher customer satisfaction [18] but also may have a greater environmental impact than traditional shopping. The graph doesn’t show the cause of the environmental impact, but it’s possible that AR-integrated shopping leads to more impulse purchases, which could increase packaging waste or transportation emissions.

Click to enlarge

Purchasing Behavior Participants in the AR-based shopping group spent more money, bought more items, and made more impulse purchases than those in the traditional shopping group. This indicates that AR can positively influence purchasing behavior, likely due to the enhanced product visualization and interactive experience that AR provides. The independent t-tests showed significant differences in all measured aspects of purchasing behavior (p < 0.05). Purchasing behavior was recorded regarding the total amount spent, the number of items purchased, and the incidence of impulse buys. The results are summarized in Table 8.

Diving Figures 7a-c reveals a fascinating interplay between customer engagement and traditional sales methods in AR-based shopping. Traditional shopping consistently leads to more impulse purchases across various clothing items (Punjabi, T-shirt, Sports Jersey, Soft Shell Jacket) (see Figures a,b). However, AR-based shopping emerges as the surprisingly preferred method despite its lower impact on immediate sales (see Figure b). This suggests that the current AR experience might foster customer engagement (see Figure a) without directly translating into impulse purchases. Interestingly, the effectiveness of promotions also seems to be item-specific, with variations observed across both traditional and AR channels (see Figures a,b). To unlock AR’s full potential for driving impulse purchases and potentially increasing spending, which currently leans towards traditional shopping (see Figure c), further research is crucial. Specifically, we need to explore how the user experience of AR shopping can be optimized for different product categories. Tailoring AR experiences to individual preferences and investigating product characteristics that resonate best with AR could be key to unlocking its full potential.

Click to enlarge

Funding

This study did not receive any specific grant from any agencies of the public or private sectors.

Ethics Statement

These experiments were conducted according to established ethical guidelines, and informed consent was obtained from the participants.

Ethical Guidelines

All methods were carried out in accordance with relevant guidelines and regulations.