Should AI Pilot Dashboards Show Period-by-Period Use?

A cumulative line only goes up, even when use is falling. Tesla’s Q2 filing charts paid Robotaxi miles as a running total. TechCrunch subtracted the quarters and estimated about 1.1 million paid miles in Q1 versus 700,000 in Q2—a roughly 36% drop. The same presentation can flatter an AI pilot: total tasks completed, total hours ‘saved,’ total users since launch. None tells a manager whether people used it less this month, stopped returning, or spent more time fixing results. I’d pair every lifetime total with period-by-period use, returning users, and work that had to be reopened. If the period view is falling, the headline should not still say growth. What number would you refuse to review cumulatively?