Dynamic effects of positive
and negative online
reviews on product sales
- Analyses on tablet
product reviews from
Literature review
• Over 100 studies have investigated the effectiveness of online reviews. • Many studies have proved that both volume and valence of online reviews
have significant impacts on sales performance.
The gap!
• Limited studies have investigated the dynamic effects of online reviews have on sales and what drove the dynamic effects
• The focus of the extremity of ratings: one-star rating and five-star rating
Research questions
(1) what are the impacts of highly positive and highly negative online reviews on sales?
Conceptual Model
• The first step is to investigate
the effects that positive and negative reviews have on sales.
• The second step is to
Methodology
• A quantitative data-driven approach on data sets of tablet product
reviews collected by Wang, et al (2013) from Amazon.com starting from 1st February, 2012 to 11th July, 2012.
• Dependent variable: sale rank (Chevalier and Mayzlin 2006; Cui, et al
2012)
• Positive reviews and negative reviews: five-star rating reviews and
one-star rating reviews.
• Reviews with all the ratings: the volume of total reviews.
• The control variables included price, competitive brands dummy
Data cleaning and final variables
• Two data sets were combined, in which the 40,741 reviews were aggregated into each product based on 24 weeks.
• The (individual/cumulative) numbers of negative reviews and positive reviews were categorized for each product for 24 weeks.
Dependent variable SRt (Sales Rank)
Independent variables RPt (Ratio of positive reviews at weekt) RNt (Ratio of negative reviews at weekt) NRt (Number of total reviews at weekt)
RPCt-1 (Ratio of cumulative positive reviews at weekt-1) RNCt-1 (Ratio of cumulative negative reviews at weekt-1) CRt-1 (Cumulative number of reviews at weekt-1)
Pt (Price at weekt)
SCR(Screen size for each product)
Models
First step: to obtain the coefficients of positive and negative reviews
𝑆𝑅𝑡 = 𝛼𝑡 RP𝑡𝛽1𝑡RN 𝑡 𝛽2𝑡 𝑁𝑅𝑡𝛽3𝑡𝑅𝑃𝐶 𝑡−1 𝛽4𝑡 𝑅𝑁𝐶𝑡−1𝛽5𝑡𝐶𝑅 𝑡−1 𝛽6𝑡 𝑃𝑡𝛽7𝑡𝛽 8𝑡 𝑃𝑃𝑡 𝛽9𝑡𝑆𝑎𝑚𝑠𝑢𝑛𝑔𝛽10𝑡𝐴𝑝𝑝𝑙𝑒𝑆𝐶𝑅𝛽11𝑡 𝜀𝑡 Where,
𝛼𝑡, 𝛽𝑡 = model coefficients in week t (t=2, 3, 4,…24).
Second step: to analyze the dynamic effects of positive and negative reviews
𝛽𝑅𝑃𝐶 = 𝜇𝑅𝑃𝐶 + 𝜃𝑅𝑃𝐶𝑊𝑒𝑒𝑘𝑁𝑢𝑚 + 𝜀𝑅𝑃𝐶 and 𝛽𝑅𝑁𝐶 = 𝜇𝑅𝑁𝐶 + 𝜃𝑅𝑁𝐶𝑊𝑒𝑒𝑘𝑁𝑢𝑚 + 𝜀𝑅𝑁𝐶 𝛽𝑅𝑃 = 𝜇𝑅𝑃 + 𝜃𝑅𝑃RP∗ + 𝜀𝑅𝑃
and 𝛽𝑅𝑁 = 𝜇𝑅𝑁 + 𝜃𝑅𝑁RN∗ + 𝜀𝑅𝑁
Analyses and results
23 OLS regressions:
• Price promotion dummy variable and Apple brand dummy variable were significant in every regression.
• The other independent variables were not significant as expected in the hypotheses.
Results table
• OLS
• GLS
Analyses of dynamic effects
The coefficients of RPCt-1, RNCt-1, RPt and RNt were used as the dependent variables in the analyses.
• Time RNCt-1 (a positive relationship)
• Lag RPCt-1 RPCt-1(a positive relationship) • Means of RPt RPt (a negative relationship)
Interesting findings
• Positive reviews and Negative reviews received from the recent
weeks had positive effects on sales (Week 2, Week 4, Week 9, Week 17 and Week 18).
• Cumulative positive reviews had negative effects on sales (Week 2, Week3, Week4,and Week 24).
• The effects of negative reviews become stronger when time passes by.
• The mean of the ratios of positive reviews for each week is
negatively related to the effects of the ratios of positive reviews of the concurrent weeks
Managerial implications
Positive – five-star rating
Encourage their customers to leave five-star rating reviews as frequent as possible.
Build up the follow-up mechanisms (e-mails) to remind the customers to rate the products after purchasing.
o Too perfect reviews may give consumers doubts (creditability)
Negative – one-star rating
• Berger, et al (2010) claims that negative information can help to build consumers’ trust and make the products more real. Nothing is perfect.
Nicely explain the negative situation mentioned in the one-star rating reviews and apologize to the consumers. Take the opportunities to defend their products if the
reviews were made irrationally.
Limitation
• The dataset in the current research was limited by 24 weeks Longer time span of dataset, e.g., VAR model
• The current study was only focused on polarity of the review ratings. Investigation into two-star rating, three-star rating and four-star rating. • The price promotion and Apple brand were significant.
Moderating effects from price, or from promotion, or from the brands. • Not able to collect the accurate release date for all the product items. Investigations into the length of the product life time.
Compare the results between pre-released and after-released.
Find out what actually drives the changes of the effects of highly negative reviews received recently. Analyze the influences of those comments following
