The decay is mediated via the rare flavour changing neutral current transition , and constitute sensitive probes for New Physics (NP), as they are forbidden at tree-level in the SM. Virtual NP contributions can therefore have a large impact, and previous LHCb measurements of the decay have shown interesting tensions with the SM predictions at the level of . The theoretical interpretation of the anomalies is difficult due to the uncertainties in non-local SM contributions, such as charm-loops , which could mimic NP effects. This talk discusses the results from a data-driven approach to constraining the size of the charm-loops and other non-local contributions to the amplitude, in the first measurement to parameterise full dimuon invariant mass spectrum. The results are obtained using an integrated luminosity of collected by the LHCb experiment.