The interfacial bond between repair mortars and concrete is often considered the weak link in concrete repair. Achieving material performance compatibility and interfacial bonding is essential to ensure a monolithic behaviour of the repaired structure. Vital to this is adequate surface preparation particularly by, among others, surface roughening. However, conventional concrete removal techniques produce large roughness variations within a single repair zone, complicating objective comparison between various repair mortars and bond strengths. Therefore, this research proposes a reproducible methodology to create a controlled concrete surface roughness, which is then validated against an on-site removal method. In this respect, four artificial roughness types are mathematically designed and applied to a fresh poured MC(0.4) and C(0.4) concrete with three consisting of varying groove configurations and one consisting of localised round chipping. Their roughness was quantified using mathematical calculations, the sand patch test and 3D scanning software. Afterwards, four commercial repair mortars were applied to the prepared substrates, resulting in 16 test combinations. Bond strength is determined using the standardised pull-off and slant-shear tests, while internal bond quality is assessed via Ultrasonic Pulse Velocity (UPV) measurements. From these tests, it was found that 3D scanning provided the most accurate assessment, although slight overestimations occurred due to edge effects. The sand patch method illustrated some usage limitations for very rough and inclined surfaces. Slant-shear tests demonstrated that rougher surfaces with a favourable groove orientation enhanced mechanical interlocking, resulting in enhanced bond strength. Pull-off tests consistently reveal interfacial failure, with no correlation to surface roughness. UPV measurements correlated with pull-off strength, with lower pulse velocities associated with lower bond quality.