Improving Ego-Cluster for Network Effect Measurement

Network effect is common in social network platforms. Many new features in social networks are designed to specifically create network effect to improve user engagement. For example, content creators tend to produce more when their articles and posts receive more positive feedback from followers. This paper discusses a new cluster-level experimentation methodology to measure the creator-side metrics in the context of A/B experiment. The methodology is designed to address the cases when the experiment randomization unit and the metric measurement unit are not the same, and it is a part of the overall strategy at LinkedIn to promote a robust creator community and ecosystem. The method is developed based on the widely-cited research at LinkedIn, but significantly improves the clustering algorithm efficiency and flexibility, leading to a stronger capability of the creator-side metrics measurement and increasing velocity for creator-related experiments.