I learned something hard about enterprise SaaS pricing last week that contradicts how most AI vendors build their models. It’s about why hourly billing actively prevents you from reaching enterprise scale. And it cost someone I met with millions in revenue.

An AI services vendor I spoke with operates on hourly time and materials. Their customers pay for each hour of service, typically at $150-300 per hour. Sounds reasonable. In practice, it created a ceiling. Every customer deal landed between $5k and $10k annually. Never higher. And here’s the problem: that’s far too small to justify the cost of enterprise sales.

If you’re selling to enterprises, your deal size needs to cover your customer acquisition cost. Sales teams, legal review, integrations, onboarding. For an enterprise partnership, you’re looking at $15k-50k in friction just to close. Hourly billing made it impossible.

The vendor’s customers were thriving. They got real value. But the vendor couldn’t afford to support them properly. Support tickets went unanswered. Roadmap requests piled up. The relationship eroded because the revenue couldn’t sustain the service commitment enterprise customers need.

Here’s what I realized: hourly billing is fundamentally a small business model. It works for consulting. It works when your customer is a startup with tight budgets and variable needs. It breaks the moment you need to invest in the customer relationship. Enterprise customers don’t want to feel like they’re costing you money. But under hourly billing, they do. Every question costs them. Every request costs them. So they ask less. They demand less. They extract less value.

The vendor had tried to increase usage. Upsell more hours. It never worked. Customers naturally resist paying more per hour. The conversation becomes adversarial. “Do I really need this?” versus “We really need this.”

So they switched. Fixed-fee subscription model. Same service, same hours. Different packaging. $3k per month, not $50 per hour.

The impact was immediate and massive. Deal size jumped to $30k-50k annually. Suddenly, enterprise sales made sense. A $40k annual contract justified the sales effort, the legal review, the onboarding process.

But the real win came after. With predictable revenue, the vendor could hire a dedicated support engineer. That engineer could proactively monitor customer health. They could push features. They could say yes to customizations because the margin worked.

Customers got better service. They used the product more. They expanded. That $40k deal became $75k in year two because the customer actually extracted the full value they’d paid for.

This is the enterprise SaaS law nobody talks about: your pricing model has to be large enough to support the relationship. If your deal sizes live under $15k, you can’t afford to treat customers like enterprises. But if you’ve sold to enterprises, you can’t go back. Their expectations are permanent.

I see this pattern everywhere now. AI vendors pricing per token. Per API call. Per seat. They’re building business models that can’t survive enterprise sales. They optimize for breadth—thousands of small customers—but they’ll never build durable enterprise partnerships.

The fix isn’t complicated. It’s about flipping the conversation. Instead of “How much will you use this?” ask “What business outcome are you trying to move?” Price against that outcome. Fix the price for a year. Let them use it however they need to.

The vendor I met with is now building a platform. Same AI, different packaging. They’re targeting teams instead of individuals. Annual contracts instead of hourly billing. And deal size is scaling with it.

If you’re in early-stage SaaS and pricing by consumption, ask yourself: Is this model big enough for enterprise? If the honest answer is no, fix it now. Because once you hit the ceiling, changing the model is painful. But the revenue bump on the other side is worth every bit of the friction.