Metabarcoding has proven to be a powerful tool to assess ecological patterns and diversity from different habitats. Terrestrial invertebrate diversity is frequently based on bulk samples, which require comparatively high sampling effort. With environmental DNA (eDNA) metabarcoding, field sampling effort can be reduced while increasing the number of recovered organism groups. However, a proof of concept is missing for several invertebrate groups, hampering the development of best-practice advice for these groups. This study aims to provide recommendations on key aspects for the processing of soil samples, from sampling effort to choice of DNA extraction method and marker genes. This study uses eDNA metabarcoding as a tool for assessing invertebrate biodiversity in soil samples, specifically comparing two DNA extraction methods (with and without a lysis step) and two genes, 18S and COI markers. The results show that the choice of marker and DNA extraction method (including a lysis step) significantly affect species detection rates and concomitantly observed invertebrate community composition. Combining methods, by using larger amounts of starting material and including a lysis step resulted in an increase of invertebrate species numbers. Together, these methods improved the detection of species with known lower population densities and allowed the assessment of temporary mesofauna. Furthermore, the choice of marker significantly influenced the diversity levels found. The 18S marker allowed the detection of a higher number of annelid and nematode OTUs, while the COI marker was more suitable for detecting changes in arthropod community structure, especially at the species level. This study makes significant advances to the field of invertebrate biodiversity assessment, particularly using metabarcoding tools by addressing several methodological considerations that are key for accurate ecological appraisals.