Not long ago, most of us assumed pricing was something you touched once every few years — like painting the office walls or changing your company logo. Maybe that’s not entirely fair, but pricing discussions often took a backseat to product development or go-to-market strategies. It was a lever to be pulled occasionally, with caution.
Today, that assumption seems almost quaint.
We’re now in a fundamentally different era of selling software. AI has altered the value equation. The conversation is no longer about who has access to your tool — it’s about what the tool actually does. And that shift is dragging billing, often kicking and screaming, into the spotlight.
Usage-based billing isn’t new, of course. But we're seeing it take on an entirely new role in the modern software stack — not as a pricing experiment or an interesting side model, but as core infrastructure. It's becoming the foundation for monetizing AI-native products, and it’s quietly re-shaping how companies operate, from engineering all the way to finance.
At Dellecod Software, we’ve felt this shift firsthand.
We’ve seen how historically brittle billing systems slow down even basic pricing experiments. At companies like Dropbox, these experiments took entire quarters to roll out — not because the ideas were complex, but because the system wasn’t built to support iteration. Add to that a lack of real-time usage data and a reliance on batch invoicing, and you had a recipe for misalignment between how customers were using products and how they were charged for them.
Today, that structure simply doesn’t hold up. AI workloads are expensive, variable by nature, and highly sensitive to usage patterns. Whether you’re calling out to an LLM or spinning up jobs across a distributed compute system, cost can spiral quickly — sometimes within hours. We’ve heard stories of startups waking up to $80,000 bills they didn’t anticipate. When your billing system isn’t built to detect anomalies or enforce limits in real time, those stories aren’t edge cases — they’re inevitable.
But usage-based billing isn’t just a technical problem. It’s an organizational one.
The practical reality is this: changing your pricing model requires changing your company. Sales compensation needs to evolve. Customer success needs to focus less on upsells and more on retention and satisfaction. Product and engineering need to be measured on the value they deliver — not just the features they release. Finance needs infrastructure that delivers clarity and accuracy in real time, not weeks after the fact.
These aren't small adjustments. They’re tectonic shifts in mindset, processes, and incentives.
What’s clear to us is that no team can do this in isolation. You can’t bolt on usage-based pricing to an existing structure and expect everything to hum. It requires top-down alignment — real alignment — and a willingness to rethink some deeply held assumptions. Who owns pricing decisions? How fast can you iterate without breaking trust? What behaviors are you incenting in sales? Who’s measuring what "value" really looks like?
We’ve found that strong leadership is non-negotiable. Someone has to own pricing — not a committee, not a council, but a single accountable owner who can cut through debates and make difficult calls quickly. The companies succeeding here structure themselves around agility. Even giants like Salesforce have overhauled their pricing models multiple times within a single year. Meanwhile, startups with nine-month pricing rollouts are being left in the dust.
The upside? When usage-based billing works, it really works.
It aligns your business with your customers. You're paid when and only when you deliver value. Sales, product, customer success, and finance start rowing in the same direction. Engineers suddenly see a direct line between infrastructure optimizations and the company's bottom line. Revenue becomes a reflection of product performance — not just contracts signed.
And because it’s so tightly coupled to value, usage-based pricing is particularly well suited to AI.
AI compounds cost and complexity. Calling an LLM is not like retrieving a row from a database — it’s expensive, adds unpredictable latency, and often involves multiple layers of infrastructure. As a result, your cost structure can change by the minute. That variability demands dynamic pricing to keep margins in check, especially when you're scaling quickly. Usage-based pricing isn't just helpful in this world — it's essential.
We’re also seeing hybrid models gain traction — mixing usage-based and seat-based elements to balance predictability with flexibility. Intercom’s AI support agents, for example, tie pricing to outcomes (resolution rates) and even include guarantees. Models like this push pricing from being merely a business model into a signal of product confidence.
The companies that navigate this well share a few traits. They’re disciplined about data. They align teams around shared metrics. They understand that pricing is a living artifact — not something etched in stone. And perhaps most importantly, they move fast. They can’t afford not to.
As we reflect on all of this, our takeaway is simple: we’re in the early innings of a new supercycle — one defined not just by new technology, but by new ways of measuring and capturing value. It’s tempting to think of pricing as a tactical matter. It’s not. It’s strategic. Possibly existential.
Every software leader today needs to consider what this shift means for their business. Not eventually — now. Because in this new era, value doesn’t live in the product alone. It lives in the system surrounding it — the way it’s packaged, delivered, billed, and understood across the company. That system needs to be built with intent.
And like most things worth building, it’s hard — but worth it.
Today, that assumption seems almost quaint.
We’re now in a fundamentally different era of selling software. AI has altered the value equation. The conversation is no longer about who has access to your tool — it’s about what the tool actually does. And that shift is dragging billing, often kicking and screaming, into the spotlight.
Usage-based billing isn’t new, of course. But we're seeing it take on an entirely new role in the modern software stack — not as a pricing experiment or an interesting side model, but as core infrastructure. It's becoming the foundation for monetizing AI-native products, and it’s quietly re-shaping how companies operate, from engineering all the way to finance.
At Dellecod Software, we’ve felt this shift firsthand.
We’ve seen how historically brittle billing systems slow down even basic pricing experiments. At companies like Dropbox, these experiments took entire quarters to roll out — not because the ideas were complex, but because the system wasn’t built to support iteration. Add to that a lack of real-time usage data and a reliance on batch invoicing, and you had a recipe for misalignment between how customers were using products and how they were charged for them.
Today, that structure simply doesn’t hold up. AI workloads are expensive, variable by nature, and highly sensitive to usage patterns. Whether you’re calling out to an LLM or spinning up jobs across a distributed compute system, cost can spiral quickly — sometimes within hours. We’ve heard stories of startups waking up to $80,000 bills they didn’t anticipate. When your billing system isn’t built to detect anomalies or enforce limits in real time, those stories aren’t edge cases — they’re inevitable.
But usage-based billing isn’t just a technical problem. It’s an organizational one.
The practical reality is this: changing your pricing model requires changing your company. Sales compensation needs to evolve. Customer success needs to focus less on upsells and more on retention and satisfaction. Product and engineering need to be measured on the value they deliver — not just the features they release. Finance needs infrastructure that delivers clarity and accuracy in real time, not weeks after the fact.
These aren't small adjustments. They’re tectonic shifts in mindset, processes, and incentives.
What’s clear to us is that no team can do this in isolation. You can’t bolt on usage-based pricing to an existing structure and expect everything to hum. It requires top-down alignment — real alignment — and a willingness to rethink some deeply held assumptions. Who owns pricing decisions? How fast can you iterate without breaking trust? What behaviors are you incenting in sales? Who’s measuring what "value" really looks like?
We’ve found that strong leadership is non-negotiable. Someone has to own pricing — not a committee, not a council, but a single accountable owner who can cut through debates and make difficult calls quickly. The companies succeeding here structure themselves around agility. Even giants like Salesforce have overhauled their pricing models multiple times within a single year. Meanwhile, startups with nine-month pricing rollouts are being left in the dust.
The upside? When usage-based billing works, it really works.
It aligns your business with your customers. You're paid when and only when you deliver value. Sales, product, customer success, and finance start rowing in the same direction. Engineers suddenly see a direct line between infrastructure optimizations and the company's bottom line. Revenue becomes a reflection of product performance — not just contracts signed.
And because it’s so tightly coupled to value, usage-based pricing is particularly well suited to AI.
AI compounds cost and complexity. Calling an LLM is not like retrieving a row from a database — it’s expensive, adds unpredictable latency, and often involves multiple layers of infrastructure. As a result, your cost structure can change by the minute. That variability demands dynamic pricing to keep margins in check, especially when you're scaling quickly. Usage-based pricing isn't just helpful in this world — it's essential.
We’re also seeing hybrid models gain traction — mixing usage-based and seat-based elements to balance predictability with flexibility. Intercom’s AI support agents, for example, tie pricing to outcomes (resolution rates) and even include guarantees. Models like this push pricing from being merely a business model into a signal of product confidence.
The companies that navigate this well share a few traits. They’re disciplined about data. They align teams around shared metrics. They understand that pricing is a living artifact — not something etched in stone. And perhaps most importantly, they move fast. They can’t afford not to.
As we reflect on all of this, our takeaway is simple: we’re in the early innings of a new supercycle — one defined not just by new technology, but by new ways of measuring and capturing value. It’s tempting to think of pricing as a tactical matter. It’s not. It’s strategic. Possibly existential.
Every software leader today needs to consider what this shift means for their business. Not eventually — now. Because in this new era, value doesn’t live in the product alone. It lives in the system surrounding it — the way it’s packaged, delivered, billed, and understood across the company. That system needs to be built with intent.
And like most things worth building, it’s hard — but worth it.