Kimberlites are the primary source of economic grade diamonds. Their geologically rapid eruptions preferentially occur near or through thick and ancient continental lithosphere. Studies combining tomographic models with tectonic reconstructions and kimberlite emplacement ages and locations have revealed spatial correlations between large low shear velocity provinces in the lowermost mantle and reconstructed global kimberlite eruption locations over the last 320 Myr. These spatial correlations assume that the lowermost mantle structure has not changed over time, which is at odds with mantle flow models that show basal thermochemical structures to be mobile features shaped by cold sinking oceanic lithosphere. Here we investigate the match to the global kimberlite record of stationary seismically slow basal mantle structures (as imaged through tomographic modelling) and mobile hot basal structures (as predicted by reconstructions of mantle flow over the past billion years). We refer to these structures as "basal mantle structures"and consider their intersection with reconstructed thick or ancient lithosphere to represent areas with a high potential for past eruptions of kimberlites, and therefore areas of potential interest for diamond exploration. We use the distance between reconstructed kimberlite eruption locations and kimberlite potential maps as an indicator of model success, and we find that mobile lowermost mantle structures are as close to reconstructed kimberlites as stationary ones. Additionally, we find that mobile lowermost mantle structures better fit major kimberlitic events, such as the South African kimberlite bloom around 100 Ma. Mobile basal structures are therefore consistent with both solid Earth dynamics and with the kimberlite record.