We describe the effect of social media advertising content on customer engagement using data from Facebook. We content-code 106,316 Facebook messages across 782 companies, using a combination of Amazon Mechanical Turk and natural language processing algorithms. We use this data set to study the association of various kinds of social media marketing content with user engagement-defined as Likes, comments, shares, and click-throughs-with the messages. We find that inclusion of widely used content related to brand personality-like humor and emotion-is associated with higher levels of consumer engagement (Likes, comments, shares) with a message. We find that directly informative content-like mentions of price and deals-is associated with lower levels of engagement when included in messages in isolation, but higher engagement levels when provided in combination with brand personality-related attributes. Also, certain directly informative content, such as deals and promotions, drive consumers' path to conversion (click-throughs). These results persist after incorporating corrections for the nonrandom targeting of Facebook's EdgeRank (News Feed) algorithm and so reflect more closely user reaction to content than Facebook's behavioral targeting. Our results suggest that there are benefits to content engineering that combines informative characteristics that help in obtaining immediate leads (via improved click-throughs) with brand personality-related content that helps in maintaining future reach and branding on the social media site (via improved engagement). These results inform content design strategies. Separately, the methodology we apply to content-code text is useful for future studies utilizing unstructured data such as advertising content or product reviews. Stanford GSB AbstractWe investigate the effect of social media content on customer engagement using a large-scale field study on Facebook. We content-code more than 100,000 unique messages across 800 companies engaging with users on Facebook using a combination of Amazon Mechanical Turk and state-of-the-art Natural Language Processing algorithms. We use this large-scale database of advertising attributes to test the effect of ad content on subsequent user engagement − defined as Likes and comments − with the messages. We develop methods to account for potential selection biases that arise from Facebook's filtering algorithm, EdgeRank, that assigns posts non-randomly to users. We find that inclusion of persuasive content − like emotional and philanthropic content − increases engagement with a message. We find that informative content − like mentions of prices, availability and product features − reduce engagement when included in messages in isolation, but increase engagement when provided in combination with persuasive attributes. Persuasive content thus seems to be the key to effective engagement. Our results inform advertising design in social media, and the methodology we develop to content-code large-scale textual data provides a framework for f...

We describe the effect of social media advertising content on customer engagement using data from Facebook. We content-code 106,316 Facebook messages across 782 companies, using a combination of Amazon Mechanical Turk and natural language processing algorithms. We use this data set to study the association of various kinds of social media marketing content with user engagement-defined as Likes, comments, shares, and click-throughs-with the messages. We find that inclusion of widely used content related to brand personality-like humor and emotion-is associated with higher levels of consumer engagement (Likes, comments, shares) with a message. We find that directly informative content-like mentions of price and deals-is associated with lower levels of engagement when included in messages in isolation, but higher engagement levels when provided in combination with brand personality-related attributes. Also, certain directly informative content, such as deals and promotions, drive consumers' path to conversion (click-throughs). These results persist after incorporating corrections for the nonrandom targeting of Facebook's EdgeRank (News Feed) algorithm and so reflect more closely user reaction to content than Facebook's behavioral targeting. Our results suggest that there are benefits to content engineering that combines informative characteristics that help in obtaining immediate leads (via improved click-throughs) with brand personality-related content that helps in maintaining future reach and branding on the social media site (via improved engagement). These results inform content design strategies. Separately, the methodology we apply to content-code text is useful for future studies utilizing unstructured data such as advertising content or product reviews.

Personalization is becoming ubiquitous on the World Wide Web. Such systems use statistical techniques to infer a customer's preferences and recommend content best suited to him (e.g., "Customers who liked this also liked…"). A debate has emerged as to whether personalization has drawbacks. By making the Web hyperspecific to our interests, does it fragment Internet users, reducing shared experiences and narrowing media consumption? We study whether personalization is in fact fragmenting the online population. Surprisingly, it does not appear to do so in our study. Personalization appears to be a tool that helps users widen their interests, which in turn creates commonality with others. This increase in commonality occurs for two reasons, which we term volume and product-mix effects. The volume effect is that consumers simply consume more after personalized recommendations, increasing the chance of having more items in common. The product-mix effect is that, conditional on volume, consumers buy a more similar mix of products after recommendations. AbstractPersonalization is becoming ubiquitous on the World Wide Web. Such systems use statistical techniques to infer a customer's preferences and recommend content best suited to him (e.g., "Customers who liked this also liked…"). A debate has emerged as to whether personalization has drawbacks. By making the web hyper-specific to our interests, does it fragment internet users, reducing shared experiences and narrowing media consumption? We study whether personalization is in fact fragmenting the online population. Surprisingly, it does not appear to do so in our study. Personalization appears to be a tool that helps users widen their interests, which in turn creates commonality with others. This increase in commonality occurs for two reasons, which we term volume and product mix effects. The volume effect is that consumers simply consume more after personalized recommendations, increasing the chance of having more items in common. The product mix effect is that, conditional on volume, consumers buy a more similar mix of products after recommendations.

How consumers use review content has remained opaque due to the unstructured nature of text and the lack of review-reading behavior data. The authors overcome this challenge by applying deep learning–based natural language processing on data that tracks individual-level review reading, searching, and purchasing behaviors on an e-commerce site to investigate how consumers use review content. They extract quality and price content from 500,000 reviews of 600 product categories and achieve two objectives. First, the authors describe consumers’ review-content-reading behaviors. Although consumers do not read review content all the time, they do rely on it for products that are expensive or of uncertain quality. Second, the authors quantify the causal impact of read-review content on sales by using supervised deep learning to tag six theory-driven content dimensions and applying a regression discontinuity in time design. They find that aesthetics and price content significantly increase conversion across almost all product categories. Review content has a higher impact on sales when the average rating is higher, ratings variance is lower, the market is more competitive or immature, or brand information is not accessible. A counterfactual simulation suggests that reordering reviews based on content can have the same effect as a 1.6% price cut for boosting conversion.