Incentives Encourage Overconsumption

Imagine paying a contractor for every nail hammered instead of for the renovation, or a rideshare for every turn it takes instead of whether it gets you there, or a salesperson for the number of emails they sent instead of the revenue they brought in. In no time, you'd be spending a lot of money without getting results. Sounds illogical, right?

Yet this is precisely how a lot of technology gets priced today. Swap "nails" for tokens, "turns" for credits, or "emails sent" for compute hours, and the logic barely changes. You're billed for activity, not outcomes.

Nobody sets out to buy more queries or more tokens or more pipeline runs. What they're actually after is better decisions, made faster, with less effort. But the pricing models built around data and AI rarely track that. They track the motion instead, and motion is easy to inflate.

Most software pricing looks rational, and paying for what you use makes sense until you look a little closer. Then you notice the vendor's revenue grows exactly when you use more. Unfortunately, that's how most pricing models are designed, and they're built to keep consumption going up.

Current players and the pricing models they have

Across the technology stack, pricing typically lines up with usage, compute, or users. Cloud infrastructure bills by compute, storage, or data processed. Data platforms charge by credits or compute time. AI platforms meter tokens and model calls.

These models exist because they're convenient... for the vendor. Usage is measurable, it scales cleanly, and it's easy to bill against. But the less talked-about reason is that usage-based pricing makes great bait. Entry prices are kept low so organizations can get in the door cheaply, build workflows around the platform, and become dependent on it. Once that dependency is set, the vendor pretty much controls the pricing journey from there. Customers rarely have a clear way to forecast how their consumption will grow over time, which leaves the cost trajectory largely out of their hands.

In every category, the vendor benefits when the customer does more. And in big enterprises, workloads are hard to track, let alone figuring out where the redundancies are.

Consider a simple example. A report is only needed once a month, but it's being processed weekly out of habit or over-caution. The obvious recommendation is to run it less often, which is the efficient, correct answer for the customer. It's also the answer that shrinks the vendor's consumption-based revenue. So under this pricing logic, what's best for the customer is directly at odds with what's best for the vendor.

This is the core problem. Current pricing is designed to optimize for "do more," not "achieve more." And it doesn't stop at misaligned incentives, because many vendors have made it structurally difficult to even draw the line between usage and outcome. If a customer can't easily see how their consumption maps to actual business results, they can't really push back on being encouraged to do more of it.

A better pricing model that puts customers first

Fixing this doesn't require abandoning usage-based billing altogether. What it requires is changing what the system is optimizing for, starting with outcomes over outputs. A system genuinely built around customer success should be willing to correlate usage and compute with results and then recommend, "You don't need to run this as often."

That only works when incentives are truly aligned. Vendors need to be able to guide customers toward optimal consumption. Right now that kind of guidance is economically irrational for most vendors to offer, and I don't think it should be. That's the entire ethos behind The Modern Data Company. In my next note, I'll dig into why more organizations can benefit from right-to-left thinking.