Metabarcoding of DNA extracted from environmental or bulk specimen samples is increasingly used to profile biota in basic and applied biodiversity research because of its targeted nature that allows sequencing of genetic markers from many samples in parallel. To achieve this, PCR amplification is carried out with primers designed to target a taxonomically informative marker within a taxonomic group, and sample-specific nucleotide identifiers are added to the amplicons prior to sequencing. The latter enables assignment of the sequences back to the samples they originated from. Nucleotide identifiers can be added during the metabarcoding PCR and during "library preparation", that is, when amplicons are prepared for sequencing. Different strategies to achieve this labelling exist. All have advantages, challenges and limitations, some of which can lead to misleading results, and in the worst case compromise the fidelity of the metabarcoding data. Given the range of questions addressed using metabarcoding, ensuring that data generation is robust and fit for the chosen purpose is critically important for practitioners seeking to employ metabarcoding for biodiversity assessments. Here, we present an overview of the three main workflows for sample-specific labelling and library preparation in metabarcoding studies on Illumina sequencing platforms; one-step PCR, two-step PCR, and tagged PCR. Further, we distill the key considerations for researchers seeking to select an appropriate metabarcoding strategy for their specific study. Ultimately, by gaining insights into the consequences of different metabarcoding workflows, we hope to further consolidate the power of metabarcoding as a tool to assess biodiversity across a range of applications.