Influencer Self-Disclosure Practices on Instagram: A Multi-Country Longitudinal Study

TL;DR

This study addresses how influencer self-disclosure practices on Instagram evolve under different national regulations over a decade (2010-2020). It assembles a longitudinal dataset of $1{,}006{,}253$ posts from $400$ influencers across the US, Brazil, the Netherlands, and Germany, and analyzes disclosure strategies, monetisation timing, and engagement effects, including a semi-supervised detector for undisclosed ads using a multilingual $BERT$ model. It finds substantial cross-country differences in disclosure frequency and strategies linked to regulation, with Germany showing high disclosure visibility and Brazil showing lower adoption, and shows that disclosure does not inherently reduce engagement, though sponsorships generally engage less than non-sponsored content. The work provides practical insights for regulators and platforms to improve transparency, and offers methods for automated monitoring, data curation, and policy evaluation.

Abstract

This paper presents a longitudinal study of more than ten years of activity on Instagram consisting of over a million posts by 400 content creators from four countries: the US, Brazil, Netherlands and Germany. Our study shows differences in the professionalisation of content monetisation between countries, yet consistent patterns; significant differences in the frequency of posts yet similar user engagement trends; and significant differences in the disclosure of sponsored content in some countries, with a direct connection with national legislation. We analyse shifts in marketing strategies due to legislative and platform feature changes, focusing on how content creators adapt disclosure methods to different legal environments. We also analyse the impact of disclosures and sponsored posts on engagement and conclude that, although sponsored posts have lower engagement on average, properly disclosing ads does not reduce engagement further. Our observations stress the importance of disclosure compliance and can guide authorities in developing and monitoring them more effectively.

Figures (12)

• Figure 1: Number of posts over time per country.
• Figure 2: Percentage of disclosed posts over time per country.
• Figure 3: Fraction of features used to disclose sponsored posts across countries over time. About 80% of all disclosed posts from DE use keywords (KW), while the majority of posts from the other countries use hashtags (HT). Paid partnership feature (AD) is used for 31% of post in BR, 27% in NL, 10% in US and only 3% in DE. Posts that use more than one disclosure method are not represented. There were no disclosed posts in 2010 and only US influencers disclosed sponsored posts in 2011.
• Figure 4: Position of the first disclosure term in the caption, measured as the number of words from the beginning.
• Figure 5: Average number of words in a caption over time per country, per sponsored and non-sponsored posts.