Aim/Purpose: This study empirically analyzed and examined the effectiveness of the online advocacy banners on customers’ reactions to make replacements with the similar products in their shopping carts. Background: When a product in a shopping cart is removed, it might be put back into the cart again during the same purchase or it may be bought in the future. Otherwise, it might be abandoned and replaced with a similar item based on the customer’s enquiry list or on the recommendation of banners. There is a lack of understanding of this phenomenon in the existing literature, pointing to the need for this study. Methodology: With a database from a Taiwanese e-retailer, data were the tracks of empirical webpage clickstreams. The used data for analyses were particularly that the products were purchased again or replaced with the similar ones upon the advocacy banners being shown when they were removed from customers’ shopping carts. Few pre-defined Apriori rules as well as similarity algorithm, Jaccard index, were applied to derive the effectiveness. Contribution: This study addressed a measurement challenge by leveraging the information from clickstream data – particularly clickstream data behavior. These data are most useful to observe the real-time behavior of consumers on websites and also are applied to studying click-through behavior, but not click-through rates, for web banners. The study develops a new methodology to aid advertisers in evaluating the effectiveness of their banner campaign. Findings: The recommending/advocating titles of “you probably are interested” and “the most viewed” are not significantly effective on saving back customers’ removed products or repurchasing similar items. For the banners entitled “most buy”, “the most viewed” might only show popularity of the items, but is not enough to convince them to buy. At the current stage on the host website, customers may either not trust in the host e-retailer or in such mechanism. Additionally, the advocating/recommending banners only are effective on the same customer visits and their effects fade over time. As time passes, customers’ impressions of these banners may become vague. Recommendations for Practitioners: One managerial implication is more effective adoption of advocacy/recommendation banners on e-retailing websites. Another managerial implication is the evaluation of the advocacy/recommendation banners. By using a data mining technique to find the association between removed products and restored ones in e-shoppers’ shopping carts, the approach and findings of this study, which are important for e-retailing marketers, reflect the connection between the usage of banners and the personalized purchase changes in an individual customer’s shopping cart. Recommendation for Researchers: This study addressed a new measurement which challenges to leverage the information from clickstream data instead of click-through rates – particularly retailing webpages browsing behavior. These data are most useful to observe the real-time behavior of consumers on websites and also are applied to studying click-through behavior. Impact on Society: Personalization has become an important technique that allows businesses to improve both sales and service relationships with their online customers. This personalization gives e-marketers the ability to deliver real effectiveness in the use of banners. Future Research: The effectiveness is time- and case-sensible. Business practitioners and academic researchers are encouraged to apply the mining methodology to longevity studies, specific marketing campaigns of advertising and personal recommendations, and any further recommendation algorithms.