Current environmental concerns have turned starch nanocrystals (SNC) into candidates of growing interest as bio‐nanofillers for nanocomposite applications. However, despite previous drastic optimization, the main drawbacks for the more extensive use of SNC remain (i) the preparation duration (5 days) and (ii) their relatively low yield (15%). Also, the final suspension of SNC actually contains both nanocrystals and microparticles. Thus, previous study optimized the SNC‐containing suspension yield, rather than the SNC yield. As an attempt to (i) further limit preparation time (to 1 day) and (ii) increase the yield of SNC assessed after filtration (as opposed to that of the heterogeneous SNC‐containing suspension), a response surface methodology (RSM) analysis has been undertaken. The modeling of size and sulfate content was not possible. On the contrary, a linear model with first‐order interactions was postulated for SNC mass yields. The model postulated to fit the SNC yield of the filtered suspension (1 µm) was more descriptive and predictive than that of the current poorly filtered suspension. Opposite effects for the same parameters have been evidenced. This confirmed the need to better isolate SNC, not only before use, but also for better modeling and optimization.