Recent advances in plasmonic nanoparticle synthesis have enabled extremely high per-particle surface-enhanced Raman scattering (SERS) efficiencies. This has led to the development of SERS tags for in vivo applications (e.g. tumor targeting and detection), providing high sensitivity and fingerprint-like molecular specificity. While the SERS enhancement factor is a major contributor to SERS tag performance, in practice the throughput and excitation-collection geometry of the optical system can significantly impact detectability. Test methods to objectively quantify SERS particle performance under realistic conditions are necessary to facilitate clinical translation. Towards this goal, we have developed 3D-printed phantoms with tunable, biologically-relevant optical properties. Phantoms were designed to include 1 mm-diameter channels at different depths, which can be filled with SERS tag solutions. The effects of channel depth and particle concentration on the detectability of three different SERS tags were evaluated using 785 nm laser excitation at the maximum permissible exposure for skin. Two of these tags were commercially available, featuring gold nanorods as the SERS particle, while the third tag was prepared in-house using silver-coated gold nanostars. Our findings revealed that the measured SERS intensity of tags in solution is not always a reliable predictor of detectability when applied in a turbid medium such as tissue. The phantoms developed in this work can be used to assess the suitability of specific SERS tags and instruments for their intended clinical applications and provide a means of optimizing new SERS device-tag combination products.