AI is starting to put real pressure on the economic model most SaaS companies rely on. Costs are less predictable, usage is uneven, and the gap between what customers commit to and what they actually consume continues to widen. Together, those dynamics are forcing companies to rethink how they price, monetize, and manage AI-driven products.

That is one reason many companies are moving toward AI credits. If left unmanaged, that cost can grow faster than revenue and erode your margins.

AI is fundamentally a capital allocation decision. It behaves more like metered infrastructure than traditional software, and finance needs to govern it with that level of discipline.

To dig into these challenges, I spoke with Zuora’s Michael Mansard, Principal Director of Subscription Strategy, who has been studying AI monetization, and more recently AI credit model design by benchmarking 150+ offers the past three years, to understand where they fit and what they require from finance.

Todd: How should a CFO decide whether credits are actually the right model or just following the market?

Michael: We’re already seeing this become common. More than 30% of AI offers now include some form of credit-based monetization, particularly where usage is variable and hard to predict.

It comes down to four practical conditions that tend to show up together:

• Meaningful variability in cost: Your AI cost to serve (for example, GPU or LLM spend) swings enough that it materially impacts gross margin if you’re wrong.
• A product that delivers different types of work: Multiple agents or capabilities doing distinct jobs, not just one homogeneous task. 
• Uneven value across those outputs: Some actions are basic utility, others have 10x+ impact for the customer. 
• Customer usage that is difficult to predict: Spiky, event-driven, or still too early to confidently forecast. 

When those conditions are present, credits help absorb variability and create flexibility in how value is captured. In more stable environments, they add overhead and are harder to manage, where simpler committed or hybrid models are often a better fit.

As a rule of thumb: if you can’t clearly check “yes” on at least three of those four conditions, credits are probably an unnecessary abstraction rather than a strategic advantage.

Todd: Where does this impact finance the most? 

Michael: It shows up in how unpredictable usage flows into revenue. Credits introduce a layer between the two that must be tracked, reconciled, and recognized precisely.

You are holding deferred revenue, recognizing it as credits are consumed and estimating the portion that may go unused. At the same time, you need to connect usage events to credit burn and then to revenue in a way that is auditable and consistent. That becomes more complex when every credit model is shaped by a set of design choices, including what actions actually burn a credit, how long credits last, and what happens when balances run out.

Without a mediation layer to translate high-volume usage into an auditable credit ledger, and an order-to-cash platform that treats credits as a form of currency, the flexibility credits created on the front end can quickly turn into audit risk, margin leakage, and balance sheet complexity.

Todd: This feels bigger than pricing. Who actually owns getting this right?

Michael:It becomes a shared model. Product drives consumption, engineering determines cost, pricing packages it, and go-to-market sets expectations.

Finance is responsible for ensuring those pieces hold economically. That is why many companies are moving toward hybrid approaches, where subscriptions and credits coexist within an offer. This includes credit hybrid models where a base subscription provides stability and is bundled with credits for variable usage, like Box Enterprise plans combining subscription features with monthly allotments of AI units.

These approaches help balance predictability and flexibility, but they also increase the pressure on finance to reconcile usage, credits, and revenue precisely.

Todd: Where are companies getting this wrong today?

Michael: They underestimate how important transparency is. When customers can’t clearly understand what they purchased, how credits are consumed, or what happens near limits, confidence breaks down. This becomes even more acute with AI, where usage is dynamic, harder to predict, and less visible to the customer in real time.

The strongest implementations focus on making usage visible and predictable, with clear reporting, proactive alerts, and flexible mechanisms for adding capacity. For example, one clear indicator of this shift is the growing number of vendors offering on-demand top-up options. 

Today, that number has reached 46%, enabling customers to add capacity without friction. That combination helps customers stay engaged rather than becoming cautious or pulling back.

What This Means for Finance

As AI usage rises, so do costs. Without guardrails, expenses can outpace revenue. Credits help, but introduce complexity and new system demands.

For CFOs, the stakes are high. Poorly designed credit models don’t just create noise in the P&L, they can force restatements, damage trust with customers and investors, and slow your ability to fund the next wave of AI investment. In a model where AI is reshaping margins and business models, getting this wrong isn’t a harmless experiment. For some companies, it will determine whether they compound value or fall behind in the AI era.

Finance should impose structure early. Define when credits make sense, standardize key design attributes like granularity, rollover, and overages, and ensure your systems can connect usage, credits, and revenue with audit-grade fidelity.

For a deeper dive, including a four-factor “credit triage” and a 12-attribute blueprint, explore Michael Mansard’s full research on AI credit monetization here.

Continue the Conversation

April 28: AI Is Rewriting Revenue. Is Your Finance Team Ready?

Every AI feature you ship creates a new pricing question and a new finance problem. Usage-based models, credit wallets, outcome pricing, bundled entitlements. The monetization choices are multiplying, and they’re landing squarely on finance teams that were already stretched thin. Whether you’re already shipping AI features or still building the commercial model, this session will give you the frameworks and the tools to move faster, without the chaos.

On Demand: Supercharging Productivity with AI in Quote-to-Cash

AI is starting to show up in quote-to-cash in real ways, but finance is still held to a higher bar. Speed alone is not enough. The outputs have to be trusted. For teams looking to see where AI is actually delivering impact today, watch this video to learn how it is being embedded into core workflows, what it takes to validate results, and how to increase productivity without introducing more risk or losing control.

New AI Monetization Research: Take the Survey

Zuora, Workday, and Deloitte are conducting new research to understand how finance organizations are adapting to AI-driven revenue models, autonomous agents, governance, value-based pricing, and more. We’re seeking participation from finance and accounting leaders that will help define best practices for operating in the AI economy. Your anonymous perspective will help shape industry benchmarks and thought leadership for the next generation of finance transformation. Take the short survey here!