Was gold evidence retrieved?
        │
        ├── No
        │     ↓
        │   Investigate dense retrieval
        │
        └── Yes
              ↓
        What displaced it?
              ↓
        What type of passage was it?
              ↓
        What were the scores?
              ↓
        What intervention could address it?
              ↓
        Re-run the benchmark

        __
        __