I wouldn't jump straight into a refactor. First, I'd spend time understanding where the pain points actually are. I'd review the data model, record types and relationships, process models, interfaces, integrations, security, and the overall application structure. I'd also look at Health Check results, production issues, application performance, and feedback from both the development team and end users to understand what's creating the biggest challenges.

Once I have a clear picture, I'd focus on the changes that provide the most value. That might mean removing duplicated logic, simplifying complex process models, improving record and query design, reducing unnecessary database calls or process instances, and moving business configuration out of hardcoded logic where it makes sense. I'd also work with business stakeholders to understand which pain points have the greatest impact so the highest-value improvements are addressed first.

I also wouldn't try to redesign everything at once. I'd take an incremental approach, validate improvements along the way, and make sure the team is aligned on design standards going forward. The goal is to improve the application's long-term maintainability, performance, and scalability while minimizing risk to ongoing development and business operations.