Efficient compound selection remains a key challenge in drug discovery today. The goal is to identify developable drug candidates early in the screening process while simultaneously flagging compounds with off-target effects indicative of liabilities or alternate indications. This goal overlaps but is distinct from the goal of toxicogenomics which is focused primarily on identifying toxicity signatures of lead candidates in key tissues. We propose a framework where global changes in gene expression levels in response to compounds can be used as an objective metric for early compound prioritization. We call this metric the Relative Transcription Index (RTI). RTI is a measure of the relative activity of compounds as ascertained by their effects on transcription at a genome-wide level. Compounds with a low RTI affect the expression of only a few genes whereas compounds with a high RTI affect the expression of a large number of genes. This information is useful for differentiating compounds that, based on phenotypic assays alone, may appear to be equally efficacious. Since compounds with high RTI are more likely to display off-target effects, the RTI metric, if implemented early in the screening process, can become a valuable tool for compound selection. The utility of the RTI metric is demonstrated by its application to two different gene expression datasets - one involving modulators of the liver X receptor (LXR) and the other concerning antibacterial compounds belonging to diverse mechanistic classes.