To understand how practitioners operationalize evaluations of earnings quality, we obtain a proprietary dataset of 1,029 reports on aggressive reporting practices over 2003-2015 for 348 unique firms published by a research firm (RF) that sells such data to institutional clients. From these reports, we identify 121 measures of poor earnings quality under four major categories: (i) sales quality; (ii) margin quality; (iii) cash flow quality; and (iv) others. As a first-cut to short-list stocks for detailed fundamental analysis, the RF appears to screen for larger, growing firms with lower barriers to arbitrage. The firms flagged by the RF also have higher M-score, F-score, and total and abnormal accruals. The average two (251) days abnormal return after the stock is first flagged by the RF is -1.30 (-18.5) percent, and such return is incremental to the return attributable to mispricing of accruals. Modified Jones and Dechow-Dichev models of abnormal accruals do not appear to capture RF identified signals well suggesting that such models are too coarse to pick up nuanced fundamental analysis conducted by the RF. In out of sample analyses, we find that the RF signals are associated with future restatements, AAERs, and GAAP-related lawsuits after controlling for other earnings quality indicators. We develop an improved earnings quality indicator (RFSCORE) for firms in the retail, durable manufacturing, and business services sectors using the RF’s signals which are based on granular, context- and industry-specific fundamental analysis. To the Street, our paper suggests that fundamental analysis, beyond just the magnitude of accruals, can predict future stock returns. To academics, our research demonstrates that granular, context-specific analysis of public data can supplement and improve the workhorse models used to identify poor earnings quality.