Novel, non-destructive material analysis methods provide a useful alternative to conventional chemical assay techniques, allowing rapid or real-time determination of ore grade with minimal sample preparation. Particularly when used for quantitative applications such as resource evaluation, grade control or metal accounting, it is important to establish whether new and old results can be used interchangeably.The combination of heavily skewed grade distributions and large sampling errors lead to significant biases when conventional metrics, such as correlation coefficients, analysis of variance (ANOVA), regressions and Tukey mean-difference or Bland-Altman plots are used to compare two assay methods. We show how these metrics can be modified to produce unbiased estimates of the quality of agreement between different assay results and introduce a procedural approach for validating new analysis methods.PhotonAssay is a recently introduced method for the measurement of gold in ore samples. We illustrate the application of our new metrics through the comparison of PhotonAssay and fire-assay grade results on a variety of ore types.