Software errors and incidents are inevitable in web based applications. Scalability challenges, increasing demand, and ongoing code changes can contribute to such failures. As software architectures evolve rapidly, understanding how and why incidents occur is crucial for enhancing system reliability. In this study, we introduce Carbon Health's software stack, share our microservices journey, and analyze 107 incidents. Based on these incidents, we share insights and lessons learned on microservice decomposition. Finally, we suggest that starting with monolithic modularization as an initial step toward microservice decomposition may help reduce incidents and contribute to building more resilient software.