Pick a product feature. What are the P0s, what's the data model, how do you know it works, and what do you show executives?

Because the candidate picks the subject, the scope choice is itself scored. Pick a whole product and you run out of time in the data model with nothing measured; pick something trivial and there is no range to show. The reliable failure is the fourth part: engineers describe their dashboard when asked what they would show an executive, and an executive does not want a dashboard, they want a consequence and a decision. The other tell is naming a database before naming an entity, which reads as someone who reaches for infrastructure before understanding the problem.