One of the social media platforms that not only serves as a place to share stories and statuses but also as a place to sell is Facebook. The data used is a dataset from Kaggle totaling 6666 data with 10 attributes and then sampled with the Slovin technique and obtained 377 sample data which will be processed using RapidMiner software with K-Means Algorithm and then optimized with Elbow Method technique, evaluation using (Cluster Distance Performance) to find the average within centroid distance value and the Davies-Bouldin Index (DBI) value. The results obtained are, the average within centroid distance value of the 3rd clustering is proven in the cluster distance performance operator obtained ???? = 3: 200237.353, ????=5: 118343.557, ????=7: 75339.476, then the ideal of clusters in this study proven by the Elbow Method is when ???? = 5, and Davies-Bouldin Index (DBI) value which is close to zero is when ???? = 3 with a value of k = 3: 0.394. In addition, clustering based on the number of likes and comments can help sellers identify the most active group of participants and potentially become loyal customers.