First, I carefully analyzed the requirements to understand the main bottleneck, which was the slow query execution due to the size and complexity of the dataset. To address this, I decided to implement a combination of indexing and caching to improve the efficiency of data access.