There are several online media closed by the Ministry of Communication and Informatics for selling abortion drug. To keep PT XYZ from being shut down by the Ministry of Communicatison and Information, PT XYZ pays attention for the circulation of this abortion drug by developing the pending system. However, the pending system only waited for the title of the product using specific keywords related to the drug input by the team so that there were still abortion drug products that passed from the system because there were products that used general keywords and sometime seller play with the keywords. Therefore, this study is conducted to build text classification model derived from the existing abortion drug products in PT XYZ which will be used for the detection of future abortion drug. This study uses the CRISP-DM model for the data mining life cycle. This study compares the model with price features and without price features. The best result is Naive Bayes model with price features generates 99,34% of accuracy and 99,34% of f1-score.