What is usage-based billing and how does it work?

Key takeaways

• According to recent industry data, 85% of surveyed software companies have now adopted some form of usage-based pricing.¹

• Usage-based billing charges customers based on what they actually consume, from API calls and tokens to data storage and completed transactions, metered in real time and invoiced on those values, making it a natural fit for AI-driven products where delivery costs fluctuate with usage.

• The top usage-based billing platforms are Airwallex Usage-based Billing, Stripe Billing, Orb, and Oracle NetSuite. Airwallex Billing pulls ahead because it pairs metering with a built-in multi-currency treasury network, so software companies can collect consumption revenue in 130+ currencies without eating hidden FX markups.⁵

AI agents and autonomous workflows have broken the old seat-based pricing playbook. Enterprise buyers are auditing their software stacks and cutting idle licenses, and vendors are responding by tying price to value instead of headcount. This guide walks through what usage-based billing is, how it works, and how to pick a platform that can scale it globally.

What is usage-based billing?

Usage-based billing charges customers for what they actually use, not what they might use. Vendors track specific activity metrics, API calls, processed tokens, data storage, completed transactions, and invoice against those numbers instead of collecting a flat monthly or annual fee. Revenue only grows when the customer gets more value, which is a healthier deal for both sides than a flat access fee.

It's spreading fast across developer tools, cloud infrastructure, and generative AI products, where service delivery costs scale with customer activity. Because the COGS for cloud resources and model inference moves with consumption, this pricing approach protects gross margins. It shifts the commercial model from paying for static access to paying for continuous value.

Usage-based billing vs. subscription billing

Unlike recurring payments, which charge a flat, predictable fee on a set schedule regardless of usage, usage-based billing lets costs move with consumption. That lowers the barrier for new customers, who can start small and pay more only once they're getting value. The table below breaks down the core differences between the two models:

This comparison shows why static billing structures are proving insufficient as software ecosystems grow more AI-driven and API-dependent.

Why software companies are dropping per-seat pricing

SaaS companies ran on per-seat subscriptions for over a decade, but that playbook is breaking down. Buyers are fatigued by subscriptions and no longer want to pay for licenses nobody's using. The rise of AI agents and the explosion of API-driven workflows have compounded the problem by severing the link between human seat counts and software value.

AI agents don't sit in seats

Traditional seat-based pricing assumes software value maps to the number of human users, but when an AI agent can perform the work of ten humans without logging in, per-seat pricing becomes commercially unviable. If a customer deploys an autonomous workflow processing millions of tasks without human intervention, the vendor can't monetize that through user licenses. The revenue model breaks.

AI products also carry significant variable computing costs, particularly large language model (LLM) inference.² A single power user running intensive agentic workflows can consume more in model API fees than the monthly subscription price. Usage-based pricing fixes this: it ties revenue to the compute and tokens actually consumed, so gross margins hold steady no matter how heavy a user gets.

Model costs also move fast. Between November 2021 and November 2024, the cost of running a model dropped 1,000-fold: from $60 per million tokens for GPT-3 down to $0.06 per million tokens for Llama 3.2 3B. Yet frontier models like OpenAI's o1 launched at $60 per million tokens, the same price GPT-3 charged three years earlier. Premium AI applications carry heavy inference costs, and flat-rate pricing means margin compression every time a heavy user scales up.

Customers prefer paying for what they actually use

Enterprise buyers are pushing for greater flexibility in how they consume software.⁴ When finance teams use SaaS spend management software to audit their software estates, they often uncover substantial waste. The average business manages roughly 305 SaaS applications in 2026, and a lot of that spend is on software nobody's actually using.

Usage-based billing cuts the shelfware problem. When a customer's usage fee drops during a slow quarter, they don't cancel. They scale back and stay, and when activity picks up they're comfortable paying more because the spend tracks directly with business expansion.

How usage-based billing works, step by step

Building a consumption-based monetization pipeline takes product engineering, data infrastructure, and finance working together. Flat-rate subscriptions run on a simple recurring invoice template; usage billing depends on real-time data flows where precision matters at every single stage. The following steps trace the lifecycle from the initial user action to payment collection.

1. Track what customers are doing

The process begins with metering: continuously tracking specific customer actions within the software. Every billable event, an API call, a database query, an AI completion, triggers the application to emit a usage event: event ID, customer ID, timestamp, event name, and quantity consumed. That metering infrastructure has to handle high-velocity ingestion with at-least-once delivery, because a dropped event turns into a billing dispute.

2. Clean up the raw data

Once ingested, usage events pass through a mediation layer that cleans and standardizes the raw data. Raw event streams are messy and full of duplicates, so the system needs deduplication and idempotency checks, or you'll end up billing the same event twice. The mediation process also handles late-arriving events, aligning them with the correct billing cycle based on the configured timezone and late-event policies.

3. Apply your pricing rules

After usage data is cleaned and consolidated, the rating engine applies the active pricing rules tied to each customer's contract. This is where raw consumption units become actual monetary values, with support for flat per-unit rates, graduated volume discounts, or tiered thresholds. The engine also factors in promotional coupons, contract discounts, and minimum commitment thresholds before it finalizes the charge.

4. Generate the invoice

Finally, rated charges get rolled up into one itemized invoice that breaks down the usage charge and any taxes owed. Pair that invoice with a detailed usage dashboard and customers can verify the numbers themselves, which cuts down on payment friction. Once finalized, the system triggers a payment pull or dispatches the invoice for collections.

Common usage-based pricing models explained

There's no single formula for consumption monetization. Vendors pick from a handful of pricing archetypes based on product complexity, infrastructure costs, and how customers actually want to buy. Here are the structures you'll see most often.

Pay-as-you-go (flat per-unit pricing)

Pay-as-you-go (PAYG) charges a fixed rate per unit, no minimums, no monthly caps. Think $0.01 per API request: developer-led startups like it because they can test and integrate without any upfront financial commitment. The tradeoff for vendors is volatile monthly revenue.

Tiered and volume-based pricing

Volume-based pricing drops the unit price across every unit consumed once you cross a threshold. A cloud storage provider might charge $0.10 per GB below 1 TB, then $0.08 per GB across the entire balance once you exceed it. Tiered pricing works differently: the lower rate applies only to units inside that specific bucket, not retroactively to everything you've already used.

Prepaid credits and usage wallets

Prepaid credits require customers to buy consumption credits upfront, then draw down the balance in real time as they use the product. Most platforms build in automatic top-ups, charging a card the moment the balance drops below a set threshold. It's a good fit for high-compute sectors like generative AI, where it keeps processing costs in check and cuts non-payment risk.

Hybrid pricing: A base fee plus overage charges

Hybrid pricing pairs a fixed platform fee with a usage-based overlay for anything beyond an included allowance. A marketing platform might charge $500 a month covering 10,000 emails, then $0.02 for every email after that. Vendors keep a predictable baseline, and customers get room to scale without renegotiating a contract.

Benefits of usage-based billing

A usage-based billing model pays off across the entire customer lifecycle. Tie cost to the value a customer actually gets, and you unlock growth levers a rigid subscription can't touch. These benefits span acquisition efficiency, expansion revenue, margin health, and churn reduction.

When incentives line up like this, vendor and customer are pulling in the same direction, and that cuts commercial friction over the long run.

The challenges of running a consumption billing model

Usage billing isn't free of tradeoffs. Moving off flat-rate subscriptions touches finance, product engineering, and sales all at once, and without the right guardrails a flexible pricing strategy turns into a liability fast. Expect revenue volatility, more customer support tickets, and new accounting compliance headaches.

Revenue can be hard to forecast month to month

When bills track variable consumption, MRR swings with seasonal demand, macro shifts, and whatever's happening inside your customers' own teams. That makes cash flow forecasting and budget allocation a lot harder for FP&A. It can also complicate investor relations, since markets historically favor the clean predictability of traditional subscription models.

Surprise bills drive customer churn

Without real transparency, consumption pricing leads straight to bill shock. A runaway script, a misconfigured API loop, or an unexpected usage spike, and a customer can end up staring at a massive invoice at month-end. Real-time usage dashboards and proactive threshold alerts are the fix, and companies that skip them pay for it in churn.

Revenue recognition gets complicated fast

ASC 606 requires software companies to map transaction prices to distinct performance obligations. That's simple enough for flat subscriptions, but it gets complicated fast with tiered usage, prepaid credits, rollover limits, or hybrid bundles. Reconciling all of that by hand in spreadsheets is slow and error-prone, which is why you need a billing platform with revenue recognition logic built in.

How to choose a usage-based billing platform

The billing infrastructure you pick determines whether you can actually execute your monetization strategy at scale. Legacy engines built for subscriptions weren't designed for the high-throughput, real-time demands of consumption data, and it shows. Evaluate platforms on ingestion speed, rating accuracy, and how deeply they integrate with the payments layer.

When evaluating candidates, look closely at the underlying data architecture. Platforms using a raw event architecture preserve a complete audit trail and allow retroactive pricing simulations; systems that store only aggregated data can't do either. Finance leaders should also check for multi-currency invoicing, automatic tax compliance, revenue recognition workflows, and developer-friendly APIs, so engineering can iterate on pricing without writing custom code every time.

Why businesses should choose Airwallex for usage-based billing

Global software companies invoicing in multiple currencies need billing and payments in one system, not stitched together through third-party gateways. Airwallex Billing is built for exactly that.

Airwallex Billing is the only platform that natively pairs a usage-based billing engine with a proprietary global payments network. That means you invoice, collect, and settle in 130+ currencies from the same dashboard, without paying FX markups or managing sync between two vendor systems. For global software companies billing European clients, Airwallex's EUR to USD converter is also useful for cross-currency pricing planning.

Key benefits

• Billing and payments on the same infrastructure, no third-party gateway needed

• Invoice and collect in 130+ currencies

• Settle like-for-like in 20+ currencies without SWIFT markups eating into revenue

• One dashboard covers invoicing, subscriptions, and usage-based billing

How to implement a usage-based billing platform

Rolling out a usage-based billing platform takes a cross-functional team: product engineering, data infrastructure, and finance, all working from the same page. With Airwallex Billing, teams configure and test metering logic in one unified dashboard, which shortens the path to go-to-market. A clean implementation starts with mapping customer consumption patterns to the right value metric.

Engineers must then instrument the software to emit clean usage events via API, ensuring each payload contains the required customer and quantity metadata. Once events are flowing, finance teams configure meters in the billing platform to define aggregation rules. From there, the billing platform connects to the global payments layer and your ERP, automating invoicing, tax calculations, and ledger reconciliation in real time.

Why your billing stack and payments layer should be the same system

The traditional SaaS monetization stack is fragmented by design: separate vendors for metering, subscription billing, and payment processing, each running its own API and its own data model. Every time a billing engine hands a payment request to an external gateway, a failed API call or a declined payment turns into a reconciliation problem somebody has to clean up manually. Running billing and payments on the same infrastructure means invoice status updates automatically as payments settle, with no external sync required.

The cost argument is equally strong, especially for companies with international customers. Traditional payment processors charge FX markups when billing in multiple currencies, and those fees add up fast as you scale across regions. Using Airwallex Billing, which operates a proprietary global payment network, you can invoice and collect in 130+ local currencies and settle like-for-like into corporate accounts, with payment management built directly into the workflow.

How to forecast revenue when billing fluctuates month-to-month

Forecasting revenue under usage-based billing comes down to three moves. MRR alone won't cut it once consumption starts shifting month to month, so FP&A teams need to stack commitment contracts, product analytics, and cohort data on top of each other. Here's the approach that works.

1. Anchor with commitment contracts. Offer enterprise customers committed-use discounts in exchange for a baseline usage floor paid upfront. This gives you a predictable revenue anchor, while any consumption above that floor is billed as a variable usage charge, which is where your upside lives.

2. Track consumption signals in real time. Monitor live indicators, including active compute minutes, API request volume, and active accounts, weeks before the billing cycle closes. These signals let you build forecasts that reflect what's actually happening, not what happened last quarter.

3. Run historical cohort analysis. Layering product-usage data with cohort-based patterns, specifically how usage evolves across months 1, 3, 6, and 12 of the customer lifecycle, can reduce monthly forecasting variance to under 20%.

Usage-based billing best practices

The companies that scale consumption billing successfully share a few common practices. Here's what they actually do:

• Twilio gives customers live usage dashboards and proactive alerts at 50%, 75%, and 90% of their expected monthly limit. That visibility alone prevents most billing disputes.

• Snowflake lets clients set automatic spend caps and circuit breakers that pause service the moment a budget threshold hits. Give customers control over their own costs, and they're far less likely to churn when the bill runs high.

• Datadog's graduated pricing rewards growing customers with volume discounts, making it commercially painful to switch to a competitor as usage increases.

• Documenting all rated invoices and contract modifications ensures full ASC 606 compliance and makes audits significantly less painful.

Frequently asked questions about usage-based billing

What is the difference between metered billing and usage-based billing?

Metered billing is the operational process: tracking precise consumption events and calculating charges off those raw units. Usage-based billing is the broader strategy, any pricing model where charges move with product interaction. Put simply, metered billing is the technical pipeline that counts API calls or storage gigabytes, while usage-based billing covers how those units get packaged, tiered, or discounted across customer segments and contracts.

Is consumption-based billing the same as usage-based billing?

Yes, in practice the two terms get used interchangeably across the software industry, both describing pricing that varies with usage. Some analysts do draw a finer line: consumption billing ties specifically to infrastructure resource depletion, server runtime, data storage, model tokens, while usage-based billing can also cover action-based metrics like page views or completed tasks. For practical purposes, treat them as synonymous.

How do companies prevent customer bill shock with consumption pricing?

Companies prevent bill shock with real-time consumption dashboards and automated spend-limit alerts, so customers have full visibility into charges as they accrue throughout the billing cycle. Modern platforms also let customers set hard spending caps or circuit breakers that pause service once a budget threshold is crossed. Notifications at 75% and 90% of expected usage have become standard practice, and they cut month-end disputes dramatically.

How does usage-based billing affect ASC 606 revenue recognition?

Usage-based billing complicates ASC 606 compliance because transaction prices fluctuate, so revenue can only be recognized as performance obligations are satisfied, not upfront. Flat subscriptions recognize revenue straight-line over time; usage billing instead requires mapping rated consumption to the right accounting period. That's why companies need a billing platform with advanced revenue management built in, one that generates ASC 606 schedules automatically.

Can I combine flat subscription fees with usage-based billing?

Yes, combining a flat subscription with usage-based overages is called hybrid pricing, and it's become the dominant model for SaaS companies that want predictable cash flow without giving up monetization flexibility. The flat base fee covers core platform access and a built-in consumption allowance; anything beyond that threshold bills as a variable charge. That way vendors capture revenue from heavy users while light users stay on a predictable tier.

What is the biggest challenge of tracking usage data at scale?

The biggest challenge is accuracy: catching and preventing duplicate charges across millions of real-time events. One dropped packet or duplicated stream is enough to trigger a billing dispute, customer churn, or lost revenue. Modern platforms solve it with a raw event architecture that runs real-time deduplication and handles late-arriving events systematically.

Why are software companies moving away from seat-based subscription models?

Software companies are dropping seat-based models because AI agents and automated workflows broke the core assumption behind them, that software value tracks human user counts. One autonomous agent can churn through a massive workload without a single human logging in, which makes per-seat pricing commercially unworkable. Usage-based models keep pricing tied to actual consumption, protecting margins while giving customers a structure that's fair and transparent.

Sources

1. https://metronome.com/state-of-usage-based-pricing-2025

2. https://www.advisable.com/insights/the-saas-pricing-shift-why-usage-based-models-are-beating-subscriptions-in-2026

3. https://highalpha.com/saas-benchmarks

4. https://www.netsuite.com/portal/resource/articles/accounting/usage-based-billing.shtml

5. https://www.airwallex.com/us/billing