The InsectChange database (van Klink et al. 2021) underlying the meta-analysis by van Klink et al. (2020a) compiles worldwide time series of abundance and biomass of invertebrates reported as insects and arachnids, as well as ecological data likely to influence their change. Based on a comprehensive review of the original publications and datasets, we highlight numerous issues in this database, such as errors in insect counts, sampling biases, inclusion of non-insects driving assemblage trends, omission of internal drivers investigated in the original studies and inaccurate assessment of local cropland cover. We argue that in its current state this database does not allow to study the temporal trends of insects or their drivers. Our in-depth analysis therefore calls for major changes and extreme vigilance in its use. It precisely details each problem identified and makes numerous suggestions for improvement in the supplementary information, which can be used as a basis for corrections.