If you think monetization strategy is about making a number on a pricing page go up or down, you might be missing the bigger picture.
I recently built a dataset in Claude that tracks pricing changes across twenty-two AI services from fourteen companies. There have been around fifty pricing changes in the last eighteen months. That’s not surprising. Everyone is running discovery in real time.
But here’s what jumped out: the majority of those changes never moved a published price at all. They changed what the published price buys.
Magicians have a term for it: misdirection. You don’t hide the trick, you simply direct attention somewhere else while it happens in plain sight.
AI pricing strategy has its own version. Everyone sees the numbers on the pricing page. Meanwhile the real move is happening one layer over, in the package, or what that number actually buys.
These companies have stopped competing on the number, and started competing on the terms attached to the number.
Here are three examples of what that looks like.
Google raised its price and nobody noticed. In March 2025, Google stopped selling Gemini as a Workspace add-on. It had been $20 to $30 per user per month. Overnight, it dropped to zero. What’s more, a Workspace Business Standard customer paying $32 for Workspace plus Gemini saw their monthly bill drop to $14. So did Google just decide to give away Gemini for free? Nope. All the Workspace customers who didn’t want Gemini saw their monthly price jump from $12 to $14 and they got it anyway. AI for everyone (for a slight, Netflix-like monthly price bump).
Zoom used the package to redefine the customer. AI Companion has been included at no additional cost on paid Zoom plans since 2023. What changed in April 2026 is that Zoom began selling AI Companion on its own ($10 a month at launch) to free-tier users with no Workplace license at all. The AI is now the front door, not the upsell. Zoom will sell you the assistant before it sells you the platform.
GitHub changed everything and touched nothing. On June 1, Copilot moved every paid plan from premium request quotas to token-metered AI credits. In other words, they moved to usage-based billing. But every published price stayed identical. Pro, still $10. Pro+, still $39. What changed is what the money buys. GitHub had been eating the difference between what a heavy user consumed and what they paid. It didn’t make sense to charge the same for a quick chat and a multi-hour agentic session. Same pricing page, materially different product.
None of this is settled science. As I discussed in my piece about credits, no one knows what a unit of AI work is worth — not GitHub, not Google, not any of us. But when a market hasn’t converged on its unit of value, letting customers self-select into the model that fits how they buy is the fastest way to find out.
Pay attention to the package first. The price will follow.
