From the above information, it can be anticipated that the quantity and level ofrating positively affect the number of favorite marks.H3: Quantity and level of ratings positively affect
Dataset overview and Sampling technique
Dataset Description
The dataset was extracted from the Sephora Products and Skincare Reviews database, collected in 2022 from Sephora’s 8000 online website The dataset consists of 2804 observations, containing information about all beauty products from the Sephora online store, including product and brand names, prices, ingredients, ratings, and features.
The data contains 24 variables, including both numerical (loves_count, rating, reviews, etc.) and categorical variables (new, online_only, sephora_exclusive, etc.).
Table 1 Dataset range Variable Obs Mean Mode Median Std Dev Min Max Range loves_count 2804 47061.9 28818 17337 93900.13 513 1401068 1400555 rating 2804 4.23266 4 354.29 0.4150056 1.67 5 3.3333 price_usd 2804 50.6748 32 263 48.18214 3 425 422 reviews 2804 713.044 0 1 1514.398 1 21281 21280 child_count 2804 3.66476 0 0 7.743223 0 78 78 new 2804 N/A 0 0 0.2445066 N/A N/A N/A online_only 2804 N/A 0 0 0.366128 N/A N/A N/A sephora_exclusive 2804 N/A 0 0 0.457996 N/A N/A N/A out_of_stock 2804 N/A 0 0 0.1753393 N/A N/A N/A limited_edition 2804 N/A 0 0 0.1284005 N/A N/A N/A
From the table above, it can be seen that Sephora products vary in love counts, ratings and the number of reviews Their prices also range from affordable to premium This is understandable as Sephora itself is known for its versatility and prides itself on its unrivaled selection of makeup, skincare, hair, and fragrance.
Research Methodology
For this dataset, our group suggests a regression analysis to discover the relationships between the product’s popularity, which is measured by love counts, and its price, ratings, number of reviews, etc., and whether the aforementioned variables do in fact affect its love counts Through that, we would be able to discover whether to accept or reject our hypothesis.
For the regression analysis, we adopt a significance level of 0.05, meaning any variables with a p-value above 0.05 would be considered having little to no effect on the dependent variable.
Variables details
Dependent vs Independent (visualization for each)
“loves_count”: Selecting "loves_count" as the dependent variable is sound as it offers insights into user engagement and preferences It reflects the number of users who favor a product, making it a valuable indicator of consumer interest and product perception Analyzing it alongside independent variables like rating, price, and marketing attributes enables a comprehensive understanding of factors influencing product popularity
“brand_name”: In our analysis, we have categorized the "brand_name" variable, which originally encompassed 250 distinct brands, and transformed it into “brand_rank” with five meaningful categories based on their pricing and target customer segments This categorization is essential for simplifying the complexity of the data and gaining insights into the market dynamics:
Table 2 Brand classification Budget-Friendly Brands that offer cost-effective products, targeting price-sensitive consumers
Affordable Brands positioned slightly above the budget range, appealing to value-conscious shoppers
Mid-Range Brands catering to the middle-income market segment with moderately priced offerings
Premium Brands offering high-quality products at premium prices, attract discerning customers Luxury Exclusive brands with luxury products designed for a niche, affluent clientele
“rating”: The number of people who have marked this product as a favorite
Figure 1 “rating” distribution using Polynomial Transformation
“price_usd”: The price of the product in US dollars
Figure 2 “price_usd” distribution using Logarithmic Transformation
“reviews”: The number of user reviews for the product
Figure 3 “reviews” distribution using Logarithmic Transformation
“child_count”: The number of variations of the product available
Figure 4 “child_count” distribution using Logarithmic Transformation
“new” (dichotomous): Indicates whether the product is new or not 1
“online_only” (dichotomous): Indicates whether the product is only sold online or not
“sephora_exclusive” (dichotomous): Indicates whether the product is exclusive to Sephora or not
1 Yellow represents 1 (yes) and Blue represents 0 (no)
“out_of_stock” (dichotomous): Indicates whether the product is currently out of stock or not
Chart 4: “out_of_stock” distribution
“limited_edition” (dichotomous): Indicates whether the product is a limited edition or not
Descriptive Analysis
This section will not include dichotomous variables, since descriptive analysis for these variables is unnecessary for they have only two possible values, making basic counts or proportions sufficient for summarizing their distribution Complex statistics like mean or variance are not applicable Hence, this descriptive analysis provides valuable insights into the distribution and characteristics of the variables: loves_count, rating, price_usd, brand_rank, reviews, and child_count, based on a dataset of 2804 observations.
Table 3 Descriptive Statistics loves_count rating price_usd brand_rank reviews child_count
Mean: The average loves_count is approximately 4.23, indicating moderate user engagement Ratings have an average of 338.10, and the average price_usd is 1.59, suggesting a moderate price range The brand_rank averages around 0.51, and reviews have an average of 2.33.
Standard Deviation: Loves_count has a standard deviation of 0.63, indicating some variability in user engagement Ratings show significant variability with a standard deviation of 112.35 Price_usd has a standard deviation of 0.30, indicating moderate price variability.
Skewness: Skewness measures the asymmetry of the distribution After data transformation, all variables are close to zero, suggesting a relatively symmetrical distribution.
Overall, the data exhibits moderate variability, with some variables showing non-normal distributions and degrees of skewness.
Correlation Analysis
3.2.3.1 Pearson’s correlation coefficient: continuous independence v dependence
Reviews and Loves Count (0.694): The number of product reviews significantly correlates with the number of people indicating a product’s popularity, with more people expressing their love for it.
Price and Loves Count (-0.223): There is a moderate negative correlation between the price of a product and its Loves Count This implies that higher-priced products tend to have fewer favorites, indicating that price may be a factor influencing customer preferences.
Rating and Loves Count (-0.131): The weak negative correlation between product rating and Loves Count suggests lower-rated products may attract more favorites, but other factors likely contribute to product popularity.
Child Count and Loves Count (0.341): Child Count, indicating the number of related products, has a moderate positive correlation with Loves Count, suggesting that more related items lead to more favorites, potentially due to a broader product range or customer loyalty. 3.2.3.2 Point-biserial correlation: binary independence v continuous dependence
The correlation analysis for binary variables reveals that online-only and new products exhibit negative correlations with Loves Count, suggesting fewer favorites for these categories
Conversely, Sephora-exclusive products have a positive correlation with more favorites. Limited-edition status shows a weak, statistically insignificant negative correlation
The availability status (out of stock) has minimal impact.
Hypothesis Testing
Hypothesis proposed
The R-squared value of 0.637 indicates that approximately 63.7% of the variance in
"loves_count" is explained by the independent variables This suggests a reasonably good fit for the model The adjusted R-squared value, which accounts for the number of predictors, is 0.630.
It is slightly lower than the R-squared, indicating that some variables might not be adding much explanatory power
The F-statistic tests the overall significance of the model With a high F-statistic of 98.52 and a very low p-value (3.83e-111), the model is statistically significant The log-likelihood value is -228.60, representing the goodness of fit A lower log-likelihood suggests a better fit, and this value indicates a relatively good fit.
The coefficients for each independent variable represent their impact on "loves_count"
"reviews" and "child_count" have positive coefficients, suggesting that an increase in these variables is associated with an increase in "loves_count."
Conversely, "rating," "price_usd," "online_only," "new", and "out_of_stock" have negative coefficients, indicating a negative relationship with "loves_count."
The p-values associated with each coefficient determine their significance Most of the variables have very low p-values (close to zero), indicating their significance in predicting
However, "limited_edition" and "out_of_stock" have p-values above the typical significance level of 0.05, suggesting that they are not statistically significant predictors Hence we will accept the null hypothesis and reject the proposed hypotheses for these variables.
Results Interpreting
"brand_rank": This variable represents the effect of changes in the "brand_rank" predictor on the dependent variable A coefficient of 0.6202 indicates that all else being equal, a one-unit increase in "brand_rank" is associated with an increase in the dependent variable by 0.6202 units.
We accept H1: Brand names of Sephora products positively affect the number of people who rate it as their favorite
"reviews": This coefficient represents the effect of changes in the "reviews" predictor on the dependent variable A coefficient of 0.5250 indicates that all else being equal, a one-unit increase in the number of reviews is associated with an increase in the dependent variable by 0.5250 units.
We accept H2: The number of user reviews positively influences the number of people who have marked the product as a favorite.
"rating": The coefficient for "rating" indicates the effect of changes in the product’s rating on the dependent variable A small negative coefficient suggests that as the product’s rating decreases (in absolute terms), the dependent variable may increase slightly.
We reject H3: Quantity and level of ratings positively affect the number of individuals who mark a product as their favorite.
"price_usd": This coefficient represents the impact of changes in the product’s price on the dependent variable A coefficient of -0.2270 suggests that all else being equal, a one-unit increase in the price (in USD) is associated with a decrease in the dependent variable by 0.2270 units.
We reject H4: The price of Sephora’s products positively affects the number of people who rate it as their favorite
"child_count": This variable represents the effect of changes in the "child_count" predictor on the dependent variable A coefficient of 1.8346 indicates that all else being equal, a one-unit increase in "child_count" is associated with an increase in the dependent variable by 1.8346 units.
We accept H5: The number of variations of the product available positively affects the number of people who rate it as their favorite
"online_only": This coefficient represents the effect of being an "online_only" product on the dependent variable A coefficient of -0.1513 suggests that online-only products are associated with a decrease of 0.1513 units in the dependent variable.
We accept H6: Sephora products' “online only” label negatively affects the number of individuals who mark a product as their favorite.
"sephora_exclusive": The coefficient for "sephora_exclusive" indicates the effect of being a
Sephora-exclusive product on the dependent variable A coefficient of 0.1189 suggests that Sephora exclusive products are associated with an increase of 0.1189 units in the dependent variable.
We accept H7: Sephora’s exclusivity positively affects the number of individuals who mark a product as their favorite.
"new": This coefficient represents the effect of being a "new" product on the dependent variable A coefficient of -0.2518 suggests that new products are associated with a decrease of 0.2518 units in the dependent variable.
We accept H8: The newness of a product positively affects the number of people who have marked the product as a favorite.
Model Assessment
This statistic checks for autocorrelation in the residuals A value of around 2 suggests no autocorrelation, and here, it’s approximately 2.092, which is very close to the ideal value.
The Variance Inflation Factor (VIF) values suggest that multicollinearity is not a significant issue among the independent variables in the model All VIF values are close to 1, which indicates low multicollinearity This means that the independent variables are not highly correlated with each other, and they can be considered suitable for use in the linear regression model without significant multicollinearity concerns.
Standard Errors are heteroskedasticity robust (HC3) means that in the regression model, standard errors have been adjusted to account for varying error variance (heteroscedasticity) in the data HC3 robust standard errors are used to provide more reliable hypothesis testing and confidence intervals in the presence of heteroscedasticity.
Recommendations
To enhance the brand’s performance and market presence, it is imperative to focus on strengthening the brand image and reputation We recommend highlighting the brand’s unique qualities, values, and heritage to set it apart from competitors Notably, showcasing any awards, recognitions, or certifications received can bolster trust and credibility among valued customers. Brand consistency is key Ensure uniform branding across all marketing materials, from the company’s website and social media profiles to product packaging This uniformity creates a strong and easily recognizable brand identity.
Engaging with customers through active social media presence, email marketing, and other communication channels is an essential strategy Actively listen to customer feedback and respond promptly to improve their perception of the brand Collaboration with industry influencers can significantly expand the brand’s reach and authenticity
Monitoring brand rankings and assessing their correlation with sales and customer satisfaction is vital Each brand should adjust branding strategies as needed to maintain and improve its positioning in the market continually These strategies will further elevate the brand equity and strengthen its competitive advantage.
The data indicates that the number of child-related products, or the "child_count," positively impacts business performance To capitalize on this trend, Sephora should consider diversifying its product line in the child-related market segment Market research is crucial to identifying emerging trends and addressing gaps in this market effectively Sephora should aim to develop products that cater to different age groups, preferences, and the unique needs of children These products should uphold the highest standards of quality and safety, as the child market places a premium on product safety and reliability Sephora’s marketing efforts should not solely target children but also encompass parents and caregivers The benefits and features of child-related products, including their safety, educational value, and entertainment, should be highlighted. Collaborating with retailers specializing in children’s products can significantly increase product visibility and distribution, especially if these retailers have a strong presence in the market.
Customer feedback should be collected regularly, and this feedback should drive product improvements Sephora should ensure that all child-related products meet or exceed safety and compliance standards, as this is crucial for maintaining trust among parents and caregivers. Sustainability practices should be considered, including eco-friendly production and packaging, as an increasing number of parents prioritize sustainability in their purchasing decisions.Furthermore, Sephora should emphasize the educational and entertainment value of child-related products, when applicable, as this can set them apart in the market and attract parents seeking enriching experiences for their children.
To address the challenges associated with "online-only" products at Sephora, it’s crucial to enhance online product descriptions This involves investing in high-quality images, informative videos, and customer reviews, making online-only items more appealing to customers Such detailed product information ensures that customers are well-informed and can confidently favorite online-only products based on the available content.
In addition to enriching product descriptions, implementing interactive virtual try-on technology is essential This technology allows customers to virtually experience online-only products, offering features like augmented reality for trying on makeup or visualizing skincare effects This interactive experience gives customers the opportunity to engage with online-only items, increasing their confidence and likelihood to mark these products as favorites
Furthermore, exclusive online-only promotions and discounts can be a powerful strategy By offering special pricing, bundled deals, or limited-time offers solely for online products, Sephora can encourage customers to favorite these items online This approach promotes the popularity of online-only products without the need for in-store engagement.
Finally, Sephora can improve the customer experience with personalized recommendations. Implementing a recommendation system that considers a customer’s browsing and favoriting history can suggest online-only products that align with their preferences This personalized approach can increase the number of favorites for online-only items by offering tailored suggestions that resonate with individual customers.
Sephora’s “Sephora exclusive” product line has significantly contributed to increasing the number of people marking products as favorites To take advantage of that, clear promotion of
"Sephora exclusive" products is essential Sephora should actively promote its exclusive products across all its platforms, including the website, mobile app, and in-store displays Clear banners, featured sections, and special collections dedicated to these products can capture the attention of customers.
Additionally, implementing limited-time offers and special bundles can be highly effective.
By introducing exclusive discounts or bundled deals with "Sephora exclusive" items, Sephora can entice customers to add these products to their favorites For instance, a promotion like “Buy any two ‘Sephora exclusives’ and get 20% off” can create a sense of urgency and incentivize favoritism.
Highlighting customer reviews and ratings for "Sephora exclusive" products is also crucial. Genuine and positive feedback from other customers serves as social proof and enhances the credibility of these exclusive items By prominently featuring these reviews alongside the products, Sephora can build trust and encourage more customers to mark them as favorites.
Conclusion
In conclusion, the data analysis result has provided valuable insights into the factors influencing the "loves_count" of products, which reflects user engagement and product popularity
Descriptive analysis revealed key statistics that offer an overview of the dataset’s characteristics Notably, "loves_count" had an average of approximately 4.233, indicating a moderate level of user engagement and popularity The analysis also indicated the presence of variability in user engagement, particularly for "loves_count," "rating," and "price_usd."
On the other hand, the correlation analysis unveiled the relationships between different variables "Reviews" and "child_count" positively correlated with "loves_count," underlining the importance of user-generated content and product range Conversely, "price_usd" displayed a moderate negative correlation, emphasizing the influence of pricing on consumer preferences. Additionally, binary variables like "online_only," "new," and "sephora_exclusive" showed associations with "loves_count," with potential business implications.
While these insights are valuable, it’s essential to acknowledge the study’s limitations The model may not capture all relevant factors influencing "loves_count," such as marketing strategies and product quality Furthermore, the analysis assumes that the data is representative and that the identified relationships will remain stable over time, which may not always be the case due to the dynamic nature of the ecommerce industry.
For businesses, these findings have clear implications To enhance user engagement and product popularity, companies can focus on incentivizing user reviews, fostering customer engagement, and feedback mechanisms Moreover, they should carefully consider pricing strategies that align with their target market and the competitive Expanding product lines or offering related items can lead to higher "loves_count" and increased user engagement Lastly,capitalizing on the allure of exclusivity and strategically managing new product releases can further maximize user engagement and favorite counts, ultimately driving greater success and customer loyalty for businesses.
Figure 1 “rating” distribution using Polynomial Transformation 7
Figure 2 “price_usd” distribution using Logarithmic Transformation 8
Figure 3 “reviews” distribution using Logarithmic Transformation 8
Figure 4 “child_count” distribution using Logarithmic Transformation 9
Chart 4 “out_of_stock” distribution 10
Apaolaza-Ibỏủez, V., Hartmann, P., Diehl, S., & Terlutter, R (2011) Women satisfaction with cosmetic brands: The role of dissatisfaction and hedonic brand benefits African Journal of
Business Management, 5(3), 792–802 https://doi.org/10.5897/AJBM10.305
Dabrowski, D (2022) The Impact of a New Product’s Novelty and Meaningfulness on its Commercial Performance IBIMA Business Review, 1–17 https://doi.org/10.5171/2022.998442
Fakhreddin, F., & Foroudi, P (2022) The impact of market orientation on new product performance through product launch quality: A resource-based view Cogent Business &
Hamilton, R., Thompson, D., Bone, S., Chaplin, L N., Griskevicius, V., Goldsmith, K., Hill, R., John, D R., Mittal, C., O’Guinn, T., Piff, P., Roux, C., Shah, A., & Zhu, M (2019) The effects of scarcity on consumer decision journeys Journal of the Academy of Marketing Science,
Itamar Simonson, Jonah Berger, Michaela Draganska (2006) The Influence Of Product
Variety On Brand Perception And Choice Stanford Graduate School of Business. https://www.gsb.stanford.edu/faculty-research/working-papers/influence-product-variety-brand- perception-choice
Mudambi, S., & Schuff, D (2010) Research Note: What Makes a Helpful Online Review?
A Study of Customer Reviews on Amazon.com MIS Quarterly, (1), 185.34 https://doi.org/10.2307/20721420
Park, C., & Lee, T M (2009) Antecedents of Online Reviews’ Usage and Purchase Influence: An Empirical Comparison of U.S and Korean Consumers Journal of Interactive
Marketing 23, (4), 332–340 https://doi.org/10.1016/j.intmar.2009.07.001
Shi, X., Li, F., & Chumnumpan, P (2020) The use of product scarcity in marketing.
European Journal of Marketing, 54(2), 380–418 https://doi.org/10.1108/ejm-04-2018-0285
Subaebasni Subaebasni (2018) Effect of Brand Image, the Quality and Price on Customer
Satisfaction and Implications for Customer Loyalty PT Strait Liner Express in Jakarta -
ProQuest Www.proquest.com. https://www.proquest.com/openview/c0704c4b3f9948e2f1b192e184227dc7/1?pq- origsite=gscholar&cbl6339
Verhagen, T., & Van Dolen, W (2011) The influence of online store beliefs on consumer online impulse buying: A model and empirical application Information & Management 48, (8),320–327 https://doi.org/10.1016/j.im.2011.08.001