<p>Non-profiled side-channel attacks (NP-SCA) don’t require redundant parameter tuning and modeling processes. However, they require higher-quality traces and longer attack times. In this paper, an evaluation metric AR-SNR is proposed to effectively evaluate the attack performance of traces in NP-SCA without executing attacks. It will help reduce the time consumption of attack failure caused by the traces themselves and then improve the efficiency of the entire attack process. The metric is designed based on the statistical characteristics of the power consumption traces and the signal-to-noise ratio (SNR) in electronic devices. It assesses the potential capability to successfully recover keys by analyzing the degree of difference in the correlation between each sampling point of the traces and sensitive intermediate values. We verified this metric on the unprotected CW-Lite dataset and the masked ASCAD dataset. The experimental results show that it can reflect the attack performance of the traces in real attacks.</p>