The team at Netlify aimed to optimize the structure of their existing system by reducing the work involved in deploying files. They identified performance issues with their current system, which was causing problems with large deploys and concurrent operations on MongoDB. To address this, they introduced a new approach that separates the concerns of tracking deploy state and resolving origin requests. The solution involves creating a new collection to store metadata for each file, deferring the creation of the final tree until the deploy is done, and batching file diff operations. This resulted in a 2x increase in deploy speed, known as "nitro deploys." The team faced challenges during the rollout, including edge cases that required careful handling, but ultimately successfully deployed the feature at 100% without issues. The new approach has improved testing and observability, allowing for better handling of massive scale and concurrency.