In this paper, we investigate whether the U.S. broiler companies used public information sharing mechanisms to collude to reduce output and fix prices. Utilizing natural language processing methods, we converted quarterly earnings call transcripts of publicly traded broiler companies into structured data. We identified 6 keywords used as signals: cut, balance, constrain, discipline, reduction and adjustment. Our results show statistically significant and mildly elastic negative relationship between the keywords signals and three different broiler production precursors. For example, a 1% increase in aggregated signal is associated with 1.24% reduction in the broiler-type eggs placed in incubators. The results are corroborated when using LASSO estimation to identify the most important elements in the vector of collusion signals. Our findings can be interpreted as adding to the already large body of prosecutorial evidence that in the early 2010s broiler companies colluded to reduce production and fix prices.