A job is slower after a change. Which evidence is most useful initially?

1. Only the most recent successful dashboard screenshot
2. An assumption that larger compute always fixes it
3. Run history and execution metrics compared with a baseline
4. Only the cloud account's display name