Faecal metabarcoding is widely used for mammalian diet analysis. However, most extraction protocols are designed to target high molecular weight genomic DNA, making short sequences of digested DNA challenging to extract. Here, we evaluate prey DNA metrics in baleen whale scat from a phosphate buffer DNA extraction method along with two commercial extraction kits (the QIAamp Fast DNA Stool Mini Kit and the PowerSoil kit) with the following variations: 1) different durations of incubation in a phosphate buffer (1 hour and 24 hours), 2) processing of both pellet and supernatant from phosphate buffer incubation, and 3) two different concentrations of DNA binding buffer to examine prey DNA. We found that the choice of extraction protocol influenced richness, diversity and composition of eukaryotes (18S rDNA) and crustaceans (Crust16S mtDNA) detected in SRW faecal samples. The PowerSoil protocol performed well for both markers, delivering the highest target richness for 18S rDNA and highest diversity for Crust16S mtDNA, while the pellet of the phosphate buffer yielded the highest richness for Crust16S mtDNA. The phosphate buffer supernatant protocols produced the lowest results for richness and diversity. Taxonomic composition in the supernatant alone was influenced by the duration of incubation and the concentration of binding buffer and was also distinct from the corresponding pellet. Our results reinforce the importance of having clear targets prior to selecting a method for faecal metabarcoding, as the specific aims (e.g. identifying richness versus diversity) will inform the choice of extraction protocol.