did both with mine. tested like 30 products the first few months, killed the losers, kept 5 winners and built my whole store identity around those. the pricing rules and supplier sync made the testing phase way less painful than doing it manually. solid for either approach really