Say "I don't know" honestly, then reason from what you do know

You will be asked something you cannot answer. Sheffield's guidance is clear that it is fine to say you do not know, and to build on what you do know rather than bluff.

A useful pattern:

1. Be honest: "I haven't measured that."
2. Reason from what you know: "Our tests used up to 200 batches per pharmacy. Queries are indexed by expiry date, so I would expect it to scale to a few thousand, but I would need to test it."
3. Say how you would find out: "I would load synthetic data at ten times the size and measure query time."

Examiners respect honest reasoning. Bluffing is usually obvious, and a confident wrong answer about your own project does more damage than "I don't know".