The concentration - virial mass relation is a well-defined trend that reflects the formation of structure in an expanding universe. Numerical simulations reveal a marked correlation that depends on the collapse time of dark matter haloes and their subsequent assembly history. However, observational constraints are mostly limited to the massive end via X-ray emission of the hot diffuse gas in clusters. An alternative approach, based on gravitational lensing over galaxy scales, revealed an intriguingly high concentration at Milky Way-sized haloes. This letter focuses on the robustness of these results by adopting a bootstrapping approach that combines stellar and lensing mass profiles. We also apply the identical methodology to simulated haloes from EAGLE to assess any systematic. We bypass several shortcomings of ensemble type lens reconstruction and conclude that the mismatch between observed and simulated concentration-to-virial-mass relations are robust, and need to be explained either invoking a lensing-related sample selection bias, or a careful investigation of the evolution of concentration with assembly history. For reference, at a halo mass of 1012M⊙, the concentration of observed lenses is c$_{12}\, \sim 40\ \pm$ 5, whereas simulations give c$_{12}\, \sim 15\ \pm$ 1.