One benchmark a week, shared by everyone who takes it. That is what makes a rank or a percentile comparable between two profiles.
One attempt each. A score is a single sitting, not the best of several tries.
Questions are written from published design literature, with the source recorded for each one.