Examining Brand Reputation's Role in Customer Loyalty Using ML

By Soham NandiReviewed by Susha Cheriyedath, M.Sc.Jul 25 2024

In an article published in the journal Sustainability, researchers evaluated the impact of brand reputation on customer trust and loyalty, using sentiment analysis and machine learning on iPhone 11 customer reviews from the Trendyol e-commerce platform. They identified key satisfaction and dissatisfaction factors, highlighting the importance of ethical principles, social responsibility, and sustainable practices in maintaining a strong brand reputation.

Background

Reputation, encompassing the respectability and reliability of individuals or organizations, is crucial in business as it impacts customer trust, brand value, partnerships, and overall corporate health. Brand reputation, specifically, influences customer loyalty, sales, and media coverage, necessitating effective management strategies such as consistent messaging and social responsibility initiatives.

Previous studies, like those by Loureiro et al., Greyser, and Vidya et al., have explored brand reputation's influence on loyalty, crisis management, and customer sentiment using sentiment analysis and text mining. However, gaps remain in understanding brand reputation's dynamic nature and how sustainable practices influence it.

The present study addressed these gaps by analyzing iPhone 11 customer reviews from Turkey's largest e-commerce platform, Trendyol, using sentiment analysis and text mining. The results highlighted satisfaction and dissatisfaction factors, offering actionable insights for brands to improve and sustain their reputation by considering customer feedback and emphasizing ethical practices.

Research Methodology and Approach

The researchers investigated iPhone 11 customer reviews on Trendyol, Turkey's leading e-commerce platform, using sentiment analysis and machine learning to assess brand reputation. The dataset comprised of 10,143 reviews, focusing on sentiment classification into positive, neutral, or negative states. The analysis followed a four-step process, text pre-processing, feature extraction, text mining, and classification using support vector regression.

Data collection involved extracting reviews from Trendyol, a major e-commerce site established in 2010 with over 30 million active users as of 2022. The iPhone 11 was chosen due to its high review volume, enhancing the accuracy of sentiment trends. Previous research highlighted the significance of brand reputation in influencing customer loyalty and brand perception, but gaps remain in understanding the impact of sustainable practices and comprehensive sentiment analysis.

Text pre-processing involved tokenization, noise removal, normalization, stop-word removal, stemming, and lemmatization to prepare the text for analysis. Feature extraction and term weighting, particularly using n-grams and term frequency - inverse document frequency (TF-IDF), were employed to extract meaningful insights. Machine learning, specifically support vector machines, categorized sentiment, while sentiment analysis assessed emotional tone from textual data.

Brand reputation is crucial for customer trust and business success. Positive reputation enhances loyalty and market performance, while effective management requires continuous effort, ethical behavior, and strategic communication. This study filled gaps in understanding how sentiment analysis and text mining can offer deeper insights into brand reputation management, emphasizing the role of sustainable practices and customer feedback.

Findings and Analysis

Data cleaning and word-stemming were performed to prepare the data for analysis. The dataset was divided into a 70-30 split for training and testing phases. Support vector machines with a radial basis function (RBF) kernel were used for classification. Four performance metrics were evaluated, namely, precision (0.504), recall (0.963), accuracy (0.500), and F1 score (0.662). The high recall indicated the model's success in capturing positive reviews, while the lower precision suggests many false positives.

Sentiment analysis revealed that 85% of reviews were positive, indicating general customer satisfaction with the iPhone 11, particularly in terms of quality, design, and performance. Negative reviews accounted for 13%, highlighting issues with specific features, unmet expectations, and delivery problems. Neutral reviews made up 2%, reflecting mixed or undecided opinions.

Word clouds of positive, negative, and neutral comments provided visual insights. Positive comments praised the product's quality, packaging, and fast delivery. Negative comments focused on problems with product quality, shipping, price, and customer service. Neutral comments were descriptive, focusing on aspects like price, quality, and shipping without a clear positive or negative sentiment.

Conclusion

In conclusion, the researchers utilized sentiment analysis and machine learning to evaluate the impact of brand reputation on customer trust and loyalty through iPhone 11 reviews on Trendyol. Findings revealed 85% positive reviews, indicating high customer satisfaction with product quality and performance, but also highlighted 13% negative reviews addressing issues with delivery, quality, and customer service.

The analysis underscored the importance of addressing negative feedback and maintaining ethical practices to enhance brand reputation. By integrating customer feedback into strategic decision-making and emphasizing sustainability, brands can improve their reputation and foster greater customer loyalty.