Decoding influencer marketing effectiveness on instagram: Insights from image, text, and influencer features

Abstract

Influencer marketing has become a crucial strategy for modern brands. By collaborating with influencers to publish sponsored posts on social media platforms, brands can leverage influencer popularity and follower engagement to enhance brand exposure and attract potential consumers. This study aims to predict the popularity of sponsored posts on Instagram and analyze the factors influencing their effectiveness. To achieve this, four distinct feature sets were extracted from sponsored posts: image visual features, image topic features, text topic features, and influencer features. These features were used individually and in combination as predictive variables to develop models for predicting post popularity using various methods. Experimental results demonstrate that the predictive models achieve strong performance, with the best results obtained when incorporating all four feature sets, highlighting the importance of considering multiple factors in evaluating sponsored post effectiveness. Furthermore, this study employs the Taguchi experiments to analyze the relative contribution of the four feature sets to the post popularity and utilizes odds ratio analysis from logistic regression to provide detailed insights into the impact of individual features. By examining the influence of visual, textual, and influencer-related factors, this study offers valuable guidance for brands in selecting influencers and optimizing post content, providing deeper insights into influencer marketing strategies.