A year ago, if you searched for AI ROI, you’d find a handful of consulting frameworks, a few academic papers, and not much else. Search it today and you’ll find dozens of platforms claiming to solve it, most of them launched in the last twelve months, all of them using roughly the same language: track usage, attribute value, show a number that goes up. I don’t think that’s a coincidence, and I don’t think it’s a compliment to the category. It’s what happens when a real problem gets big enough that everyone wants to be standing next to the solution, whether or not they’ve actually built one.
I spend a fair amount of time looking at competitors, partly because customers ask us to explain what makes Nymbral different and partly out of plain curiosity about how crowded this is getting. The honest answer, most of the time, is that there isn’t much difference to explain, because most of what’s out there isn’t actually measuring ROI. It’s measuring activity. Logins, prompts, feature clicks, dashboards that light up green when usage is high and never ask the harder question of whether that usage translated into anything a CFO would recognize as value.
There are three questions I ask when I want to know whether a self-described AI ROI tool is the real thing or a dashboard wearing a new label. The first is scope: does it cover every vendor in the stack, or just the one it was built to sell alongside? A tool that only measures its own platform’s impact is measuring a slice of the problem and presenting it as the whole picture. The second is methodology: is the number built from an independent, published approach, or does it come from a black box the vendor won’t open up under questioning? If nobody can explain how the number was calculated, the number isn’t worth much. The third is durability: would the result survive a skeptical CFO, a curious board member, or an actual auditor asking where each input came from, or does it fall apart the moment someone pushes on it.
Most of what’s on the market right now fails at least one of those tests, and a lot of it fails all three. That’s not really an accusation. It’s a description of what happens early in any category, before the standards have hardened and before customers have learned to ask the right questions. Eventually they will, the same way finance teams learned to ask hard questions about cloud spend a decade ago and marketing attribution a decade before that. AI spend is heading toward the same reckoning, just faster, because the dollar amounts are larger and the pressure from boards is more immediate.
We built Nymbral to already be standing on the other side of that reckoning. Full coverage across every vendor and contract, a published and research-anchored methodology instead of a black box, and a score built to survive the exact kind of scrutiny most of this category is currently avoiding. Everyone can call what they built AI ROI. What matters is whether it holds up when someone with real authority in the room decides to test it.