Purpose The purpose of this paper is to investigate the relation between average ratings (viewership) and the volume and valence of electronic word of mouth (e-WOM) for early episodes of TV shows. Design/methodology/approach Linear regression was performed in which the dependent variable is average TV ratings and main independent variables are volume and valence of e-WOM. The study used a Breusch-Pagan test to detect heteroscedasticity. Accordingly, the model is analyzed using heteroscedasticity-consistent standard error estimators. Findings The results show that the volume of the early e-WOM does not significantly contribute to explaining average ratings, but the valence does. Originality/value Because the advertising revenue of television broadcasters is determined according to expected TV ratings, the average ratings should be predicted as early as possible. This study shows that analyzing early e-WOM helps predict average ratings.