Measure AI for growth, not hours saved, and strong ROI triples

A new study from the Wharton School and GBK Collective lands on a distinction Uprovd has argued for months. Most companies - four in five - now say they see some return from AI, but barely a quarter report a return they would call strong. What separates the two groups turns out to be less about how much they spend and more about what they bother to measure.

Firms that judge AI against growth outcomes, such as new revenue, report strong ROI at roughly triple the rate of those chasing efficiency alone. Yet the metric most teams still reach for is hours saved, while net-new revenue remains the least tracked of the major measures. In other words, the companies winning on AI are the ones counting the right things.

That is the measurement gap Uprovd was built to close. Activity and time saved are easy to tally and easy to inflate; a result tied to a business outcome is harder to fake and far more convincing to a board. We connect AI usage to the outcome it was meant to produce, so a vague sense of return becomes a number a finance chief can stand behind.