What is Usage-Based Pricing?
Usage-based pricing (UBP) is a model where customers pay in proportion to how much of a product they consume. Instead of a flat monthly fee, the bill fluctuates based on a specific metric: API calls, compute hours, messages sent, storage used, or tokens processed. Twilio charges per SMS sent. AWS charges per compute-second. Snowflake charges per credit of query processing time.
The model traces back to utilities like electricity and water, but it entered software in force with cloud infrastructure in the mid-2010s. According to OpenView's 2023 report, 61% of SaaS companies had adopted some form of usage-based pricing, up from 34% in 2019. By 2026, Gartner estimates 70% of businesses prefer usage-based models over per-seat pricing, driven largely by AI products where per-seat pricing doesn't reflect the value delivered.
Why Usage-Based Pricing Matters
Three forces are accelerating UBP adoption. First, AI products have broken the per-seat model. When an AI agent processes 10,000 documents for a 5-person team, charging per seat makes no sense. The value scales with usage, not headcount. Second, customers increasingly resist paying for shelfware. Seat-based contracts where only 40% of licenses are active create resentment. UBP eliminates this friction because customers only pay for what they use. Third, UBP creates a natural expansion revenue engine. As customers derive more value, they consume more, and revenue grows without a sales conversation.
The financial impact is measurable. Companies with usage-based models report median net revenue retention of 120%, compared to 110% for purely subscription-based peers. Datadog, which prices on ingested metrics and log volume, has maintained NRR above 130% for consecutive years. The model turns product adoption into revenue growth automatically.
For PMs, UBP changes what you optimize for. Instead of focusing on seat expansion or upsell triggers, you focus on reducing friction to consumption. Every barrier to usage is a barrier to revenue. This shifts roadmap priorities toward activation speed, consumption visibility, and product-led growth mechanics.
How to Use Usage-Based Pricing
1. Choose the right pricing metric. The metric should be easy for customers to understand, predictable enough for budgeting, and directly tied to the value they receive. Stripe charges per successful transaction because each transaction represents revenue for the customer. Bad metrics are ones customers can't control or predict. Charging per database query penalizes exploration.
2. Build consumption visibility into the product. Customers need real-time dashboards showing current usage, projected costs, and historical trends. AWS CloudWatch, Snowflake's usage console, and Vercel's bandwidth meters all serve this function. Without visibility, customers get surprised by bills and churn.
3. Offer spend controls and alerts. Let customers set budget caps, receive threshold notifications, and auto-pause at limits. This feels counterintuitive (you're giving customers a way to spend less), but it builds trust. Customers who feel in control of spending commit to the platform long-term.
4. Design pricing tiers with volume discounts. Most UBP models use tiered rates: the first 10,000 API calls at $0.01 each, the next 100,000 at $0.005. This rewards growth and makes the model more affordable at scale, which is essential for landing enterprise contracts.
5. Pair with a minimum commit for enterprise. Large customers need budget predictability. Offer annual minimum commitments at a discounted rate with overage pricing above the commitment. Snowflake's "capacity" contracts work this way: prepay for a credit block, consume at your own pace, and buy more if you exceed it.
Usage-Based Pricing in Practice
Snowflake decoupled storage from compute, then priced compute per credit consumed. Customers scale up for heavy queries and scale down overnight. This elastic model drove NRR above 150% at IPO because customers who started small grew consumption as they loaded more data. The product was designed so that increased usage was the natural outcome of successful adoption.
Twilio prices per message, call minute, or API request. A startup sending 100 SMS notifications per month pays cents. An enterprise sending millions pays significantly more, but the cost per message decreases with volume. This let Twilio land tiny startups and expand into enterprise without changing its pricing model.
OpenAI charges per token processed (input and output tokens separately). This model suits AI inference because the cost to serve varies directly with usage, and customer value correlates with throughput. But it also means customers actively optimize prompts to reduce token consumption, a behavior PMs must account for when designing features.
Common Pitfalls
- Choosing a vanity metric instead of a value metric. Charging per "workspace" or "project" when the real value is in the processing underneath creates misalignment. Customers game the metric (cramming work into fewer projects) rather than using the product naturally. Always tie the pricing unit to the value delivered.
- Ignoring revenue volatility. UBP introduces monthly revenue swings. A customer running a big data migration one month and nothing the next creates lumpy revenue. Build forecasting models that account for seasonality and usage patterns, and use committed contracts to smooth the curve.
- No free tier or entry point. The best UBP companies offer a generous free tier (Vercel's hobby plan, Supabase's free database) that lets developers build and test without entering a credit card. Gating access behind payment kills the bottom-up adoption that makes UBP work.
- Bill shock destroying trust. If a customer's monthly bill jumps from $500 to $5,000 because of a misconfigured script or unexpected traffic spike, they won't celebrate your revenue growth. They'll churn. Proactive alerts at 50%, 75%, and 90% of historical spending protect the relationship.
Related Concepts
Usage-based pricing is one specific approach within the broader pricing strategy framework. It pairs naturally with product-led growth because both depend on reducing friction to adoption and letting product usage drive revenue. Understanding unit economics is critical when implementing UBP. You need to know your cost to serve at every consumption level to ensure margins hold as customers scale.