Although isolating microorganisms from food microbiota may appear less challenging than from the gut or environmental sources, recovering all representative species from food remains a difficult task. Here, we showed by metagenomic analysis that several abundant species had escaped isolation in a previous study of ten cheeses, including several previously uncharacterized species. This highlights the ongoing challenge of achieving a comprehensive recovery of microbes from food. To address this gap, we designed a novel strategy integrating metagenomics-based probes targeting the species of interest, coupled with an incremental culturing approach using pooled samples. As proof of concept, we applied this strategy to two cheeses containing species that were not isolated in our previous study, with the objective of isolating all species present at levels above 2% and, in particular, potential novel food species. Through this approach, we successfully performed the targeted isolation of two Psychrobacter and two Vibrio species from the first cheese, and four Halomonas and two Pseudoalteromonas species from the second one. Notably, P. undina and V. litoralis represented, as far as we know, the first cheese isolates characterized for these species. However, we were unable to isolate a novel species of Pseudoalteromonas, with no characterized representative to date, and Marinomonas foliarum, previously isolated from marine environment. Using metagenome-assembled genomes (MAGs) and metagenomic analysis, we discussed the possible reasons for their non-recovery. Finally, this strategy offers a promising approach for isolating a set of strains representative of the microbial diversity present in food ecosystems. These isolates can serve as a basis for investigating their roles in the communities, their impact on product development, safety implications and their potential in the development of starter cultures.