Downstream process development for recombinant glycoproteins from yeast is cumbersome due to hyperglycosylation of target proteins. In a previous study, we purified three recombinant glycoproteins from Pichia pastoris using a simple two-step flowthrough mode approach using monolithic columns. In this study, we investigated a novel automated data science approach for identifying purification conditions for such glycoproteins using monolithic columns. We performed three sets of design of experiments in analytical scale to determine the separation efficiency of monolithic columns for three different recombinant horseradish peroxidase (HRP) isoenzymes. For ease of calculation, we introduced an arbitrary term, the relative impurity removal (IR), which is representative of the amount of impurities cleared. Both, the experimental part and the data analysis were automated and took less than 40 min for each HRP isoenzyme. We tested the identified purification conditions in laboratory scale and performed respective offline analyses to verify results from analytical scale. We found a clear correlation between the IR estimated online through our novel data-driven approach and the IR determined offline. Summarizing, we present a novel methodology, applying analytical scale advantages which can be used for fast and efficient DSP development for recombinant glycoproteins from yeast without offline analyses.