This study aims to describe how the web-based interactive medium Wordwall was planned, enacted, evaluated, and refined across cycles in order to improve the Arabic writing skill of seventh-grade students, and to identify the obstacles that arose during implementation. The research employed a classroom action research design following the Kemmis and McTaggart model, conducted in two cycles at MTs Al Ukhuwah Batam with twenty-six seventh-grade students and the Arabic teacher acting as collaborator. Each cycle comprised planning, acting, observing, and reflecting. Data were gathered through classroom observation, semi-structured interviews, documentation of students’ written products, and writing tests administered before the first cycle and after the second. Qualitative data were analyzed with the interactive model of Miles and Huberman, while test scores were analyzed descriptively and through normalized gain. The findings show that the class mean rose from 31.54 in the pre-cycle stage to 88.85 after Cycle II, the score range narrowed from 10–60 to 50–100, and the normalized gain reached 0.84, which falls in the high category. Reflection identified unstable internet connectivity and unequal device ownership as the principal obstacles in Cycle I; organizing students into device-sharing groups in Cycle II reduced both. The study concludes that Wordwall functions not merely as a motivational add-on but as an instructional action that can be refined cyclically. It contributes a process-level account of gamified writing instruction in Arabic, and further research should test the model over longer interventions and with comparison classes.