Despite decades of research on school effectiveness and improvement, we continue to struggle to support school improvement at scale. I suggest this is in part due to methodological and theoretical limitations of the extant literature: While there is a growing consensus that leaders should “diagnose” school improvement needs to devise contextually appropriate improvement strategies, no empirical guidance exists to support how to make such diagnoses or which strategies to employ given contextual variation. I address this gap through a mixed methods analysis of how schools with varying improvement capacity at the onset of the COVID-19 pandemic in New York City adapted and learned in response to the pandemic’s challenges. While the COVID-19 pandemic has presented school and district leaders with a more extreme context in which to confront a central problem that has confounded educational researchers and practitioners for decades: How can school and district leaders build the improvement capacity that will enable them to continuously meet the needs of all students in all schools? There are five key findings of the present study. First, using Latent Transition Analysis (LTA), I identified six statistically significant subgroups among all New York City elementary, middle, and high schools serving students in grades 3-8 (n=1225) based on teachers’ perceptions of school improvement capacity. I further described the relationship between this typology and school contextual covariates and student outcomes, depicting the types of schools that are classified in each subgroup and the relationship between subgroup classification and academic outcomes. Second, I demonstrated how teachers’ perceptions of school capacity varied from 2017-2019 and further identified a differential relationship between principal turnover and school improvement trajectories. Third, I found strong qualitative support for the quantitative typology, describing alignment between teachers’ and leaders’ lived experiences of school improvement and change and the quantitative typology and trajectories with one key exception: those schools that experienced a leadership transition after 2019 were most likely to have experienced dramatic change. Fourth, I found teachers’ and leaders’ perceptions of challenges associated with the COVID-19 pandemic varied, in part, as a function of their improvement capacity at the onset of the pandemic. Respondents in schools from high-capacity subgroups were more likely to view pandemic challenges as easy to overcome, while respondents in schools with minimal improvement capacity were more likely to be overwhelmed by the multiple, compounding challenges they faced during the pandemic. Finally, I found key differences in the strategies schools employed to adapt and learn in response to these challenges, which again, varied based on their improvement capacity when the pandemic hit. Together, these findings provide support for a theory of differentiated school improvement.