In this study, a novel smart surface-morphing technique is devised that dynamically optimises roughness parameter on a sphere with varying flow conditions to minimise drag. A comprehensive series of experiments are first performed to systematically study the effect of dimple depth ratios in the range of 0 ≤ k/d ≤ 2 × 10−2 across a Reynolds number range of 6 × 104 ≤ Re ≤ 1.3 × 105. It is observed that k/d significantly affects both the onset of the drag crisis and the minimum achievable drag. For a constant Re, drag monotonically reduces as k/d increases. However, there is a critical threshold beyond which drag starts to increase. Particle image velocimetry (PIV) reveals a delay in flow separation on the sphere’s surface with increasing k/d, causing the flow separation angle to shift downstream. This results in a smaller wake size and reduced drag. However, when k/d exceeds the critical threshold, flow separation moves upstream, causing an increase in drag. Using the experimental data, a predictive model is developed relating optimal k/d to Re for minimising drag. This control model is then implemented to demonstrate closed-loop drag control of a sphere. The results demonstrate up to a 50 % reduction in drag compared with a smooth sphere, across all Reynolds numbers tested.