Several agri-environment schemes aim to improve pollinator diversity and abundance, including the sowing of wildflower areas. These seed mixes are often either low in floral diversity and target few pollinator species (mainly social bees), or high in floral diversity but with limited evidence of the good establishment of the component species. In order to support a greater diversity of wild bees in farmland, we need more diverse seed mixes, containing species shown to support a wide diversity of insect pollinators, with the good establishment and long flowering periods. Here we trialed two typical seed mixes, a low-diversity Fabaceae-heavy mix (FAB) and a more diverse wildflower mix (WF), against two novel wildflower mixes, one based on literature sources (LT), and one based on first-hand surveys of pollinator attraction to flowers growing on a wildflower farm (WB). Both new mixes were focused on plants attractive to wild bee species. Replicated field plots were set up on two farms and monitored over three years. Our novel wildflower mixes had higher floral diversity and abundance than the FAB mix and began flowering earlier, reaching their floral peak before the FAB mix, potentially providing forage for a broader range of pollinators or those with earlier flight seasons. The high floral abundance in LT and WB was driven by annuals in the first year, and then multiple perennials in the second and third year. We identified five perennials from four families (Daucus carota, Leucanthemum vulgare, Geranium pyrenaicum, Lotus corniculatus, and Trifolium hybridum) that established well on both farms, are known to be attractive to a diversity of bee species, and thus could be considered as providing a more taxonomically diverse base for creating future mixes. However, the mixes provided few floral resources in April (needed by early-flying wild bees), and more research is required in this area.