If you benchmark a product and it does 13,000 req/sec (or whatever), then answer the question of why it isn't doing 130,000 req/sec. It's a way of encouraging you to debug it further.
Got it, thanks! For me the better question is often simply "Why is it X?" Knowing why the speed is X is often of much greater utility than simply knowing that X is the speed. I like to focus on determining the current single limiting factor before considering the harder question of what would be necessary to achieve 10x.
What does he mean with his repeated question "Why isn't it 10x?"