The field of comparative genomics is shifting toward pangenomics, aiming to alleviate the reference bias observed in reference-based approaches. High-quality genomes are needed as input for pangenomics, as explained by the principle of “garbage in, garbage out”. Errors in an assembly or annotation will lead to technical variation in a pangenome, while it is meant to reveal genuine biological variation only. Achieving the required assembly and annotation quality remains challenging, particularly in plants, given the complexity in genome size, ploidy level, and repeat content present in the plant kingdom. Nevertheless, the (technical) variation uncovered by pangenomics can, in turn, guide iterative refinements that yield high-quality genomes. The comparative approach is especially powerful to identify “abnormalities” in a set of genomes, which cannot be detected from traditional, stand-alone quality assessment. Two use cases, on roses and chili peppers, illustrate the role of pangenomics in this iterative process leading toward meaningful pangenome analyses. Thus, we argue that high-quality genomes can be achieved through pangenomics, facilitating the convergence on real genetic variation underlying complex traits and diseases, not only in plants but also in other organisms.