Paper Number
2309
Paper Type
short
Description
Virtual influencers have started to amass significant followings on online social networks and to collaborate with popular brands. However, the content characteristics driving the engagement with virtual influencer posts are still unexplored. This study addresses this knowledge gap by analyzing a unique dataset comprising over 25,000 posts from 171 influencers gathered using Google Cloud Vision and Instagram Graph API data. We conduct hierarchical negative binomial regression models for both likes and comments and find that posts displaying emotions such as anger and joy result in substantial increases in engagement metrics. Additionally, interactive posts using hashtags or tags also generate more engagement. Moreover, we present our remaining work which includes incorporating text characteristics (e.g., sentiment and emotional congruence) and manually coded content types (informative, entertaining, remunerative, and relational formats) into the model as well as post-hoc analyses and robustness checks to further disentangle the effects identified.
Recommended Citation
Schwehm, Kevin; Schwehm, Steven; and Prigge, Jana-Kristin, "Driving Engagement with Virtual Influencer Content – Integrating Computer Vision, Text Analysis and Manual Coding" (2023). ICIS 2023 Proceedings. 13.
https://aisel.aisnet.org/icis2023/socmedia_digcollab/socmedia_digcollab/13
Driving Engagement with Virtual Influencer Content – Integrating Computer Vision, Text Analysis and Manual Coding
Virtual influencers have started to amass significant followings on online social networks and to collaborate with popular brands. However, the content characteristics driving the engagement with virtual influencer posts are still unexplored. This study addresses this knowledge gap by analyzing a unique dataset comprising over 25,000 posts from 171 influencers gathered using Google Cloud Vision and Instagram Graph API data. We conduct hierarchical negative binomial regression models for both likes and comments and find that posts displaying emotions such as anger and joy result in substantial increases in engagement metrics. Additionally, interactive posts using hashtags or tags also generate more engagement. Moreover, we present our remaining work which includes incorporating text characteristics (e.g., sentiment and emotional congruence) and manually coded content types (informative, entertaining, remunerative, and relational formats) into the model as well as post-hoc analyses and robustness checks to further disentangle the effects identified.
When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.
Comments
15-SocialMedia