PT ABC is a Peer To Peer (P2P) company that currently lends to micro segment with grameen model. New loan segment: warung/retailers that affiliated with PT XYZ (an Ecommerce platform) are targeted by PT ABC. With this different characteristic compare to existing segment, PT ABC has to find a new credit scoring model to mitigate the credit risk. Hence this study is developed to help PT ABC identify variables that can be used in the initial credit scoring and final scoring models. This study uses quantitative analysis with logistic regression and scorecard methodology to build credit scoring model. First model was built based on population of 876 warungs that have history of soft loan from the affiliated eCommerce PT XYZ. The final model used population of 193 debtors who apply for the Pilot Project. The pilot project was conducted in three months from December 2018 to February 2019 with the latest loan maturity in May 2019. The data used in this project is secondary data, collected from PT ABC. The number of data default although small but still meet the requirement based on theory. The result of this study is finding that variables 1) Type of warung, 2) City, 3) Duration as member of ecommerce, 4) Number & amount of Topup & Transaction are significant to the initial credit model. Meanwhile in final credit model use variable: 1) Age 2) Home Ownership, Type of warung, 4) Ratio Income/Expense, 5) Ratio Net Income/Installment, 6) Ratio Income/Limit, 7) Tenor of business. The model predictive power is also acceptable with AUC/ROC and Accuracy greater than 70%. The final scoring model is recommended to be used in the secondary pilot project with additional rule set based on data analysis. Keywords : Scorecard, Logistic Regression, Credit Model, Micro Lending, Warung