Since not all influencers properly disclosure whether a post is sponsored, and only a few countries have a disclosure policy, we use a machine learning-based approach that can find out if a certain post is sponsored or not. We identify such posts as Likely Sponsored.
Likely Sponsored posts include:
- Ambiguous disclosures like #thanks, #collab, #sp, #spon, #partner, #ambassador, etc.
- All types of disclosure in other languages, e.g. #sponsorisé, #produitoffert, #partnerschaft, #kooperation, #partenariat, etc.
- Non-hashtag disclosure, e.g. “Thanks to @brand for sponsoring”, “Shoutout to @brand for gifting me this awesome makeup palette.“, etc.
Our current accuracy rate is 95%. However, this accuracy is always changing when the machine learning model is updated. It is not a static algorithm, yet we are regularly training the system with paid posts so that it can learn how to classify new content.
To learn about how we detect undoubtedly paid content please check the article about Sponsored posts.