Metabarcoding of complex metazoan communities is increasingly being used to measure biodiversity in terrestrial, freshwater, and marine ecosystems, revolutionizing our ability to observe patterns and infer processes regarding the origin and conservation of biodiversity. A fundamentally important question is which genetic marker to amplify, and although the mitochondrial cytochrome oxidase subunit I (COI) gene is one of the more widely used markers in metabarcoding for the Metazoa, doubts have recently been raised about its suitability. We argue that (i) the extensive coverage of reference-sequence databases for COI, (ii) the variation it presents, (iii) the comparative advantages for denoising protein coding genes, and (iv) recent advances in DNA sequencing protocols argue in favour of standardising for the use of COI for metazoan community samples. We also highlight where research efforts should focus to maximise the utility of metabarcoding.