Your Company Is Measuring AI Adoption Wrong—Here’s What To Track Instead

InnovationYour Company Is Measuring AI Adoption Wrong—Here’s What To Track InsteadByAbhishek Gandotra,Forbes Councils Member.for Forbes Technology CouncilCOUNCIL POSTExpertise from Forbes Councils members, operated under license. Opinions expressed are those of the author. | Membership (fee-based)May 20, 2026, 06:00am EDTAbhishek Gandotra is VP of Product at American Express. He is the author of The Sovereign Agent: A Manifesto for the Inference Economy. gettyTo maintain your company's edge, measure human judgment and decision-making regarding AI outputs.​Every enterprise dashboard I've seen in 2026 tracks the same AI metrics: adoption rate, time saved, tickets automated, cost per inference. These numbers are going up and to the right at almost every company that has deployed agentic AI at scale.They are also the wrong numbers.I've spent 20 years building platforms in financial services—first at Green Dot, where we powered Apple Pay Cash and served millions of accounts, and now at American Express. In both environments, I've watched organizations measure the thing that's easy to count and miss the thing that actually matters. AI adoption is no different. The metrics that feel like progress are masking a convergence problem that will show up in 18 months as a competitiveness crisis—and by then, the damage will be structural.Here's what I mean, and what to do about it.The Convergence TrapWhen an entire team adopts the same AI model with the same default settings, their output starts to look the same, converging not toward excellence, but toward the model's median. Marketing emails start sounding identical. Engineering design documents propose the same architectures. Strategic analyses reach the same conclusions. This is not a failure of the people. It is the predictable result of a system where 12 professionals are all polishing the same model's default output.From the outside, productivity is up. From the inside, differentiatio...