AI is now your most prolific user. In the enterprise, bots and agents are driving more software activity than people, and they’re doing it at speeds and scales that were unthinkable just a few years ago. The rules of engagement have changed, but most software billing hasn’t caught up. Enterprises are still quoting per seat, stuck on a pricing relic that was never built for this era.

Per-seat was never right for SaaS. SaaS just inherited it.

Per-seat licensing came out of the 1980s on-prem world, where every user needed a physical install, a physical license, and a fixed slice of server. The seat was a real cost unit. SaaS inherited the model not because the economics made sense, but because the buyers were the same people. CFOs and CIOs grew up on Oracle and SAP and were comfortable with seat-based pricing. It fit budget frameworks and approval processes. Per-seat was the path of least resistance. It was never a strategic choice.

SaaS delivery costs don’t scale with seats. A 500-seat instance doesn’t cost five times as much as a 100-seat instance. Marginal cost per user is nearly zero. Per-seat pricing never matched the value delivered. Power users driving outcomes and casual users who log in once a month pay the same. The model has always been disconnected from both cost and value.

SaaS survived per-seat because of three structural patches

  1. Simplicity. Per-seat is easy to quote, easy to approve, and easy to audit. Every stakeholder in an enterprise procurement cycle can understand a price-per-user model without explanation. Moving away from per-seat meant adding unwanted complexity.
  2. Predictability. For CFOs managing annual budgets, per-seat offered a known, controllable cost line. Headcount times license cost, calculable in advance, defensible in a board presentation.
  3. Governance. A seat wasn't just a billing unit. It was how enterprises tracked identity, managed access, and controlled who could touch what. As long as managing human access was a core IT concern, the seat remained a useful organizing concept.

None of these patches solved the underlying problems.

Shelfware exploded. Thirty to forty percent of seats went unused. Vendors recognized revenue on licenses that delivered no value, and churn signals disappeared in the noise. Pricing opacity turned renewals into negotiations, not partnerships. Customers had no way to know whether they were getting value relative to what they paid. And per-seat created a hard expansion ceiling: vendors had no mechanism to capture value from intensity of use, only breadth of access.

The model worked well enough to sustain a generation of SaaS businesses. It was never the right model. It was the familiar one.

Bundling was the industry's answer. It did not work.

As the disconnect between per-seat pricing and AI usage became clear, incumbents responded with bundles. Salesforce added Einstein to Sales Cloud. Microsoft included Copilot with Microsoft 365. SAP packed Joule into its platform. The promise was simple: get our AI, no extra charge, one price for everything.

The approach made commercial sense but failed structurally. As AI usage matures, the flaws were obvious:

  • Cross-subsidy. Heavy AI users are subsidized by light users. The customer running Copilot for 12 hours a day racks up massive compute costs compared to someone who drafts 1 email a week, yet both pay the same. Light users churn. Heavy users expand. The economics collapse.
  • Value misalignment. Charging for access treats every user equally, but the real value in AI lies in the outcomes delivered: tickets resolved, contracts reviewed, code written. Sophisticated buyers will ask why they pay for seats instead of results. Vendors have no good answer.
  • Competitive exposure. Competitors' pricing per outcome win the value narrative. Showing exactly what a customer paid per resolution or outcome beats the old model. The seat cannot compete.

The seat is not dead, but as the primary way to monetize AI, it is fatally flawed.

The industry's next move, per-agent pricing, repeats the mistake under a new name. An agent running twelve hours and one executing a single task are not equal. Swapping the noun does not fix the model.

What needs to happen is clear: pricing based on what the agent actually does, not just on its existence. Usage and outcomes must be metered in real time, with enough transparency that buyers see the actual value.

The infrastructure problem nobody is talking about

Changing your pricing model is a business decision. Executing it is an infrastructure problem.

Metering tokens, execution events, and outcomes in real-time, with the accuracy and auditability enterprise billing requires, is not a configuration change in your existing ERP or CRM. These systems were built for fixed entitlements, periodic reconciliation, and predictable revenue lines. They have no native capability to meter signals or settle revenue against consumption that varies by orders of magnitude between billing periods.

And the pressure is not coming from one direction. Sales motions are shifting from seat counts to consumption conversations. Customer success teams are being asked to drive expansion through usage and outcomes rather than renewals. Finance can no longer forecast linearly against headcount. Contracts are moving toward real-time value delivery rather than fixed minimums. 

Every function that touches revenue is being asked to operate in a way that the underlying infrastructure was never designed to support.

That is the real problem and not the pricing model. The foundation underneath it. Enterprises that get the pricing strategy right but leave the infrastructure unchanged will hit the same wall at a different point in the journey. The revenue stack has to be able to hold the weight of what the business is becoming, metering every signal, pricing every layer of value, reconciling in real time, and doing it with enough transparency that customers trust what they are paying for.

Per-seat pricing did not break last year. It was broken the day SaaS inherited it from on-premise software, and nobody asked whether the model made sense for a different cost structure.

AI just made sure you noticed.

About Monetize360

Monetize360 is the revenue infrastructure platform for the consumption economy, built to meter, price, and capture value across usage, token, outcome, and agent-based models without replacing the systems you already run while delivering the trust and transparency your customers expect from every invoice.

‍