In a candidate-led case the interviewer waits for you. After you structure, you choose what to look at, ask for the data, read it, and decide where to go next. The cycle is:
- State a hypothesis: "I suspect costs, because revenue is flat but profit fell."
- Ask for the data that tests it, not for everything.
- Read it and conclude: confirmed, or ruled out.
- Move on, out loud: "Costs are flat, so the problem is on the revenue side. Let me look at the mix."
Every few minutes, summarize where you are. The trap is fishing for data with no hypothesis, or staying on a branch the data has already closed.