Dry mangosteen leaves are one of the raw materials used to produce marker ink. However, research using this free and abundant resource is rather limited. The less efficient one-factor-at-a-time (OFAT) approach was mostly used in past studies on plant-based marker ink. The use of statistical analysis and the regression coefficient model (mathematical model) was considered essential in predicting the best combination of factors in formulating mangosteen leaf-based marker ink. Ideally, ink should have maximum color lightness, minimum viscosity, and fast-drying speed. The objective of this study to study the effect of glycerol and carboxymethyl cellulose (CMC) on the color lightness and viscosity of mangosteen-leaves-based marker ink. The viscosity, color lightness, and drying properties of the ink were tested, the significant effect of glycerol and CMC (responses) on ink properties was identified and the prediction model on the optimum value of the responses was developed by using response surface methodology (RSM). The microstructure of mangosteen leaves was analyzed to study the surface morphology and cell structure during dye extraction. A low amount of glycerol used was found to increase the value of color lightness. A decrease in CMC amounts resulted in low viscosity of marker ink. The optimum formulation for the ink can be achieved when the weight percents of glycerol, benzalkonium chloride, ferrous sulphate, and CMC are set at 5, 5, 1, and 3, respectively. SEM micrographs showed the greatest amount of cell wall structure collapse on samples boiled with the lowest amount of glycerol.