AI Pulse · Business Models & Winners

NielsenIQ just posted a 34% jump in revenue from its AI-native products in Q2, and over half of its top 100 clients are now using at least one of them. That’s a real number from a real incumbent, not a startup pitch deck. But the more interesting part is what CEO Jim Peck said about pricing on the company’s next product, a tool for measuring agentic commerce that’s launching later this year. He admitted they’re testing several pricing models at once, some based on consumption, some not, and they haven’t picked a winner yet.

That’s worth sitting with. NielsenIQ has been pricing data products for decades. They know how to build a rate card. The fact that they’re openly refusing to commit to one billing model before they see how customers actually use the product tells you something bigger is going on in the market. Nobody has figured out yet how you fairly charge for AI agents doing work on someone’s behalf. Charge per query and you punish efficient agents. Charge a flat fee and you eat the cost of heavy users. NIQ is basically running a live pricing experiment in public, which most companies would never admit to.

If you sell anything with an AI layer attached, or you’re about to add one, this matters more than the 34% growth number. Your customers are going to ask “how do I get billed for this” before they ask “how good is it.” Watching how a company NIQ’s size handles that question, and how honest they’re being about not knowing the answer yet, is a useful signal for your own pricing conversations. It’s also exactly the kind of usage and pricing data question we think about when we build an AI command centre for clients, so decisions like this aren’t guesswork.

Read more in the original report.