AI products generate thousands of usage events per customer, and each one represents activity that has to be priced, tracked and recognized as revenue. Metering captures the data, billing turns it into invoices and revenue recognition makes it auditable. But with an increase in AI services and products, it’s becoming crucial to tie your service delivery with billing and rev rec. Revenera reports that 80% of technology companies already offer AI-enabled products or features, but 70% say delivery costs are eroding profitability.
When metering and billing live in separate systems, the gap between what the customer used and what the general ledger shows turns into a reconciliation exercise every financial close. Getting usage based billing right means deciding where that work happens, and the answer is inside the enterprise resource planning (ERP) system, not next to it.
Key highlights:
- AI usage billing means charging customers based on tokens, API calls, compute hours or other AI-related services.
- Hybrid pricing – a base subscription plus usage overage – is now the dominant model for AI companies, which means billing systems have to handle fixed and variable charges on one invoice.
- Metering platforms solve data capture well, but invoicing, revenue recognition and cash collection still have to run through the ERP to be trustworthy at close.
- ZoneBilling connects usage data to invoicing, revenue recognition and payment collection inside NetSuite, so finance owns the full billing cycle on one record.
What is AI usage billing?
AI usage billing means charging customers based on their consumption of AI powered services, measured in tokens, API calls, compute time, agent actions or resolved outcomes. It works differently from traditional subscription billing in a few important ways:
- Usage volumes are unpredictable, so a customer’s bill can swing widely month to month.
- Metering often spans multiple dimensions at once, like tokens and API calls and storage, rather than a single seat count.
- Unit costs vary because different AI models cost different amounts to run.
- Margin exposure grows with usage, since inference cost rises right alongside the revenue it generates.
Common AI billing models
Most AI companies use one of five pricing approaches, often blending more than one.
- Token-based pricing: Customers pay per input or output token, common for LLM APIs. This model requires metering at the model level and tracking cost per token so finance can watch margin.
- Credit-based pricing: Customers buy credit pools that get consumed by usage. Credits abstract raw tokens into a simpler unit customers can understand, and this has become the dominant model for AI products with several usage dimensions.
- Outcome-based pricing: Customers pay for completed actions, like a resolved support ticket, a generated report or a matched invoice. This pricing model carries higher value but is harder to meter, since the business has to define exactly what counts as a completed action.
- Hybrid pricing: A base subscription plus usage overage. Pure pay-as-you-go pricing is less common than it was in 2024, and hybrid is now the model most AI companies default to. A typical structure looks like $500 a month for 100,000 tokens, then $0.005 for each additional token.
- Tiered pricing: Usage bands with a lower unit price at higher volumes. Tiers encourage adoption and make bills more predictable for customers who scale up.
Whichever model a business lands on, metering is only the first piece of the puzzle. Developer-first platforms like Orb, Metronome, Stripe Billing and Solvimon capture usage events well.
What metering can’t do alone is turn usage into a trustworthy financial record. Invoicing has to match usage to contract terms and tax rules, revenue recognition has to keep pace with ASC 606 and IFRS 15, and cash application and reconciliation both have to tie back to the same customer record. When those steps run in separate systems, each handoff is a step that can slow down reconciliation tasks and disrupt month-end close.
What to look for in AI usage billing software
Evaluating a usage billing solution comes down to seven questions worth asking before you buy.
- Metering flexibility: Can the system meter tokens, API calls, compute time, agent actions and custom dimensions? Agentic AI adds new complexity here, since autonomous agents that span multiple models and tools generate billing events that are harder to attribute to a single usage metric. Multi dimensional metering is table stakes now, not a nice to have.
- Hybrid billing support: Can the system handle subscription, usage and credits on the same invoice without custom engineering work?
- ERP integration depth: Does metered usage flow to the general ledger (GL) automatically, or does finance have to export and reconcile it by hand?
- Revenue recognition: Does the system handle ASC 606 and IFRS 15 for consumption based revenue on its own, or does that require a separate tool?
- Invoice accuracy: Does the invoice reflect contract terms, credits, prepayments and tax, or does it just show raw usage?
- Margin visibility: Can the team see cost to serve per customer alongside billed revenue? Token costs vary by model, so billing without cost attribution leaves the team billing blind.
- Scale without engineering dependency: Can finance configure pricing changes, like new tiers, new dimensions or new models, or does every change require an engineering ticket?
How to implement usage-based billing for AI services
Getting usage-based billing right starts with clarity about what usage actually means for your product. For AI services, that could be tokens processed, model type, API calls, seat count or generated outputs, and each metric changes how revenue gets calculated, priced and recognized. Ambiguity here makes billing harder to scale and even harder to audit, while a defined metric mapped directly to the contract gives finance a clear, auditable path from usage to revenue.
1. Define your usage metrics
Start by naming the exact metric that drives your pricing, whether that’s tokens, API calls, model type, seat count or generated outputs. Each choice changes how revenue gets calculated and recognized down the line, so this decision has to happen before pricing or contracts get built, not after.
2. Apply usage-based pricing logic
Once the metric is defined, decide how it turns into a price. That might mean tiers, floors or ceilings, prepaid or pooled consumption, or overages layered on top of a base plan. The logic needs to match how customers actually buy, not just how the system happens to track usage.
3. Build contract-level billing setups
Pricing logic rarely stays uniform across every customer. Trial terms, customer-specific discounts and negotiated pricing all have to layer onto the base model without turning into one-off exceptions engineering has to maintain by hand.
4. Create a clear handoff from usage tracking to billing
Usage data only becomes revenue once it’s mapped to charge logic, and that handoff needs to be explicit rather than assumed. Metrics have to connect directly to billing rules inside NetSuite, so a spike in usage turns into an invoice line item automatically instead of waiting on someone to notice it.
5. Run pricing, rating and billing logic inside NetSuite
The last step is where most of the value shows up. When pricing, rating and billing all run in NetSuite, a rate change or a new tier flows straight through quote to cash without a parallel update in a second system. Finance can make that change in an afternoon instead of waiting on an engineering sprint.
3 risks of managing AI billing outside NetSuite
Billing outside the ERP feels manageable at low volume, and then it doesn’t. What starts as a workable stopgap – a spreadsheet here, a manual export there – scales into real risk as usage volume and AI-powered revenue grow.
Disconnected revenue data and audit risk
When AI charges live in one system and financial records live in another, deferred revenue tracking and audit readiness turn into manual processes. AI credits and tokens are often prepaid but consumed later, so finance has to track expiration policies and roll unused credits forward by hand instead of letting the system handle it.
That gap compounds under ASC 606 and IFRS 15, which require revenue to be recognized as the performance obligation is satisfied. When billing and revenue data live outside the ERP, maintaining that kind of compliance becomes far harder to prove, since the records an auditor needs to see are split across two systems instead of one.
Duplicate work and reconciliation drag
When usage is billed in a separate tool, finance has to re-enter transactions into NetSuite or extract and reconcile them in Excel every month, and that lost time compounds as usage volume grows. Some finance teams end up dedicating multiple people just to move billing data between a customer relationship management (CRM) platform or quoting tool and NetSuite, downloading CSVs, double checking contracts and reformatting data by hand before it ever reaches the ledger.
What starts as “good enough” scales into hidden costs. Subscription changes, proration and cross-sells or upsells slip through the cracks, and bottlenecks turn into recurring escalations rather than one-off exceptions.
Scattered reporting and guesswork forecasting
Finance can’t pull real-time insight when AI revenue is buried in an external billing tool, so financial forecasts start running on guesswork instead of live data. Multiple systems create multiple versions of the truth, and finance ends up reconciling numbers instead of reporting on them.
The more AI-powered revenue a business generates, the bigger each of these risks becomes, and the harder it gets to catch a discrepancy before it reaches close.
What can you do with AI service billing data?
When AI billing is centralized, structured and connected, it becomes the foundation for smarter decisions. Finance can use it to track key metrics in real time, spot growth opportunities early and adjust faster. With usage, pricing and contract data captured at the source and kept aligned inside NetSuite, finance gets a direct line to the metrics that matter, from forecasted revenue to account health, without waiting on an export or an end of month report.
Transform usage data into predictive forecasts
With AI services, usage patterns spike and dip, contracts shift mid-term and one missed assumption can throw off a quarter. Manual forecasting in a spreadsheet rarely keeps up with how fast things change.
An automated billing system built inside NetSuite gives finance the structure and timing to forecast with confidence. Usage activity flows in as it happens, charge records update instantly and teams can run usage scenarios, preview future charges and view forecasted revenue from within the subscription record itself.
That makes it possible to:
- Generate prospective charges based on usage and existing contract terms
- Preview projected charges and revenue directly in NetSuite subscription records
- Reduce manual work and improve forecast accuracy
Identify revenue opportunities and churn signals
The same data that powers billing also reveals what’s growing and what’s at risk. When usage, contract and sales data live in one system, finance can track expansion trends and catch churn signals early instead of stitching together dashboards from multiple tools.
Keeping billing and revenue logic in sync as accounts change means upsells, coterms and renewals get tracked without rebooking contracts, and metrics like annual recurring revenue (ARR) and usage by customer are easy to find. Decisions come faster and with more clarity.
That makes it possible to:
- Monitor ARR, monthly recurring revenue (MRR) and usage at the customer level
- Track mid-contract changes without rebooking or manual updates
- Spot churn indicators using structured billing data
Improve transparency with real-time dashboards
Forecasting and revenue analysis fall apart when data is fragmented. Multiple systems create multiple versions of the truth, and finance ends up reconciling instead of reporting, which slows everything down and adds risk.
Embedding billing, revenue recognition and reporting inside NetSuite keeps everything aligned. Finance gets dashboards that match the general ledger, real time answers to critical questions and structured data that flows directly into BI tools like Power BI for deeper analysis.
That makes it possible to:
- Report on billed, deferred and recognized revenue from inside NetSuite
- Maintain a complete audit trail from contract to cash
- Build dashboards that reflect official financials
Use billing insights to refine pricing strategies
AI pricing doesn’t stand still. Models evolve, new features launch and customer needs shift, so finance needs a way to update pricing without manual work or compliance risk.
Applying contract level discounts, adjusting usage tiers or scheduling rate changes should flow directly into billing and revenue schedules. When that runs inside the ERP, updates stay compliant with ASC 606 and billing stays aligned across every account.
That makes it possible to:
- Roll out AI-specific pricing models like token or seat-based billing
- Test and launch new rates or discounts with minimal friction
- Adjust pricing mid-contract without manual intervention
How ZoneBilling handles AI usage billing inside NetSuite
ZoneBilling connects usage data to invoicing, revenue recognition and payment collection on one record inside NetSuite, so finance never has to reconcile usage against the ledger by hand. It supports:
- Multi-dimensional usage rating: Tokens, API calls, compute time and custom metrics all rate against configurable pricing rules.
- Hybrid invoicing: Subscription and usage charges combine on one invoice, with credits and prepayments applied automatically.
- Automated revenue recognition: ZoneBilling wraps around NetSuite Advanced Revenue Management (ARM), so metered consumption flows to revenue schedules without manual journal entries.
- Connected payment collection: ZonePayments applies incoming payments directly to the customer record, so cash application doesn't depend on a separate system.
