Mass customization is increasingly required in the cosmetics industry to meet individual needs and expectations of fast delivery. While modular assembly supports customization in discrete manufacturing, continuous production –common for miscible products (powders, creams…)– relies on fixed batch sequences, limiting variety and responsiveness. Reverse Blending (RB), an optimization model successfully applied in the fertilizer industry, enables delayed differentiation by blending a small set of semi-finished products (i.e., CBIs) near the point of sale. This paper presents a proof-of-concept case using 100 simulated customized facial powder formulas defined by 12 dermocosmetic components. The RB model identifies only 8 common CBIs needed to reproduce all formulas without deviation. These results show that RB can simplify production while preserving full personalization, offering a smart and sustainable solution for continuous supply chains in cosmetics.