Testing of enhanced engine components under real engine conditions is still important to validate the design and meet certification requirements. However, testing is expensive and time-consuming. Furthermore, there is a high risk that the numerical design differs from the experimental results. To mitigate this risk, small-scale experiments can be conducted to obtain experimental validation data in an early design stage. The present study evaluates this design strategy on the specific case of a highly aggressive inter-compressor duct (ICD). A highly modular test rig was set up at the Technical University of Berlin (TUB) as a down-scaled twin with a scaling factor of 2.55 of the German Aerospace Center (DLR) ICD rig. Test campaigns of the highly aggressive ICD were conducted on both test rigs. The ICD design is the same, but the adjacent components differ. The design process is traced backward in the present study. Starting from a detailed analysis of the full-scale test campaign, the differences between steady Reynolds-averaged Navier-Stokes (RANS) simulations and the experimental data are evaluated. The losses predicted with the RANS simulation and the Hellsten turbulence model are more than 80% higher than measured. Finally, the results of the down-scaled rig are compared with those of the full-scale rig. Although the extent of the loss regions differs, both rigs show the same loss mechanisms. Therefore, the chosen design approach is effective, particularly when the numerical modeling is subject to high uncertainties.