Despite enormous resource investment, researchers lack consensus on seemingly simple questions about how video games affect players. Topics such as violent games and aggression, playtime and wellbeing, and more are characterized by deeply divided opinions and conflicting results even at the meta-analytic level. I argue that much of this disagreement stems from limitations of common methods, including use of problematic self-report measures, overreliance on cross-sectional research, and superficial use of theory, among others. To resolve debates more effectively, I describe six practices games research should adopt more widely: 1) strengthening the theoretical derivation chain, 2) attending to inter-individual effect size variation, 3) prioritizing longitudinal, within-person studies, 4) harnessing digital trace data, 5) controlling individual game features, and 6) adopting open research principles. For each practice, I include practical steps and resources for implementing them in one’s own work. While critical of the current state of knowledge, the paper is hopeful: with the help of these practices, the next generation of video games research can be the most informative yet.