Usage-Based SaaS Business Model: How It Makes Money
Direct answer: Usage-based SaaS charges customers for what they actually consume rather than a fixed seat or license fee. Revenue expands when customers use the product more and contracts when they use it less, making NRR the primary indicator of whether the model is generating compounding returns. Key costs are cloud infrastructure proportional to usage plus the sales, support, and R&D required to drive product adoption.
Revenue structure
Where a seat-based SaaS company earns a predictable fixed fee per user per period, a usage-based company earns revenue in proportion to consumption. The consumption metric varies by product: Snowflake charges on compute credits and storage; Datadog charges on hosts monitored and log volumes ingested; Cloudflare charges on bandwidth, requests, and compute units; Twilio charges per message or per call minute.
The pricing mechanism fundamentally changes the revenue recognition pattern. A seat-based company recognizes its contracted revenue ratably over the contract term; a usage-based company recognizes revenue when the usage event occurs. This makes reported revenue a closer real-time indicator of customer activity, but also more volatile. Monthly revenue can fluctuate materially based on customer workloads, which makes year-over-year comparisons more informative than sequential quarter-to-quarter figures for evaluating the underlying trend.
Why usage-based NRR can be very high or very low
Net revenue retention in usage-based models reflects product adoption depth rather than contract renewals. When a customer adopts a usage-based platform more deeply into their operations, usage grows naturally without a formal expansion sale. A data analytics company that processes twice as many queries in year two as year one doubles its Snowflake spending without anyone renewing a contract or negotiating additional seats.
This mechanism enables NRR to reach levels uncommon in seat-based models: Snowflake reported NRR above 130% during periods of strong customer expansion, reflecting customers deeply integrating the data cloud into their operations. Datadog has sustained NRR above 120% across multiple years as customers expand from one monitoring product to multiple observability modules.
The same mechanism creates downside risk. When customers reduce usage, whether due to optimization, workload migration, or economic pressure, revenue contracts without a formal churn event. Twilio's NRR fell materially in 2022 and 2023 as customers optimized messaging volumes and usage growth slowed from pandemic-era peaks, demonstrating that usage-based NRR can turn negative quickly in a usage downturn even with customers remaining on the platform.
Cost structure
Usage-based SaaS companies carry infrastructure costs that scale with consumption, unlike seat-based companies where the marginal cost of serving one more user is very low. A company like Snowflake must pay cloud infrastructure providers (AWS, Azure, Google Cloud) for the compute and storage its customers consume. Gross margins for usage-based businesses are typically lower than seat-based SaaS at similar scale: 60% to 75% is common versus 75% to 85% or higher for pure subscription businesses, reflecting the variable infrastructure cost embedded in cost of revenue.
Sales and marketing for usage-based businesses often takes a product-led growth approach: a low-friction free tier or trial allows customers to start using the product with minimal upfront commitment, and revenue grows as those customers adopt the product more broadly. This can improve CAC payback relative to a direct enterprise sales motion, though large enterprise contracts still typically require a field sales team and a formal procurement process.
Failure modes
Revenue volatility is the defining risk. Usage can slow in any quarter due to seasonal workloads, customer budget cycles, economic conditions, or product-specific factors like shifts in messaging volume or query patterns. A company that has grown accustomed to high NRR driven by usage expansion is particularly exposed when usage plateaus or contracts, because the model relies on existing customers continuing to grow their consumption rather than renewing a fixed fee.
Infrastructure cost exposure is the second risk. If customers shift usage toward the most compute-intensive (and most expensive to serve) features, or if cloud infrastructure costs increase, gross margins compress. Usage-based companies must price their consumption tiers carefully to maintain margin as the product mix evolves.
Representative companies
Snowflake (SNOW) reported product revenue of $3.2 billion for fiscal year 2025 (ending January 2026), growing 28% year over year, with a net revenue retention rate of 127% as of that report. Datadog (DDOG) reported revenue of $2.7 billion for fiscal year 2024, with NRR consistently above 120%. Twilio (TWLO) demonstrated the risk side: revenue growth decelerated sharply in 2022 and 2023 as communication volumes normalized post-pandemic, and NRR fell below 105%. These figures are from company 10-K and annual report filings for the periods stated; current figures should be verified in the most recent report.
Related models
- Enterprise SaaS: fixed subscription pricing, more predictable revenue, NRR driven by contract expansion
- AI Model API Provider: token-priced API access, a usage-based model applied to AI inference
- Infrastructure-as-a-Service: consumption-priced cloud compute and storage at hyperscale
Frequently Asked Questions
How does usage-based SaaS pricing work?
Usage-based SaaS charges customers in proportion to what they consume: data processed, API calls made, messages sent, compute-hours used, or similar product-specific metrics. Customers with low usage pay little; customers with high usage pay more. This aligns pricing with the value customers receive and lowers the barrier to starting, since a customer does not need to commit to a fixed annual fee before understanding their actual usage. The downside for the vendor is revenue unpredictability, since usage can fluctuate with customer workloads, economic conditions, or decisions to optimize consumption.
What is the difference between ARR and revenue for usage-based companies?
ARR (annual recurring revenue) for usage-based companies is typically estimated based on current consumption rates annualized, rather than contractual committed revenue. Unlike seat-based SaaS where the contract defines the future revenue, a usage-based company's future revenue depends on whether customers maintain or grow their consumption. Some usage-based companies have committed-spend contracts or minimum commitments that provide a revenue floor; others rely entirely on consumption-based billing. Investors should distinguish between contractually committed ARR and estimated ARR based on run-rate usage when evaluating revenue visibility.
Why can net revenue retention exceed 150% in usage-based SaaS?
NRR measures revenue from existing customers in the current period versus the prior period. In usage-based models, a customer who doubled their data warehouse queries, API calls, or compute usage would double their spending without the vendor needing to negotiate a new contract or expand a seat count. Because usage can grow rapidly as customers adopt a product more deeply into their workflows, NRR in usage-based models can reach 130% to 160% or higher in growth phases. However, the same mechanism works in reverse: a customer who reduces usage lowers revenue without a formal cancellation.
What are the main risks specific to usage-based SaaS?
The key risks specific to usage-based SaaS are revenue volatility (consumption can slow in economic downturns or when customers optimize spend), infrastructure cost exposure (the vendor bears the marginal cost of serving each unit of usage, so margin compresses if consumption grows in cheaper-to-serve categories), and pricing pressure as customers optimize consumption or negotiate volume discounts. Twilio experienced significant NRR pressure in 2022 and 2023 when customer communications volumes slowed after pandemic-era peaks, illustrating how quickly usage-based revenue can decline relative to the prior period baseline.
How do usage-based companies report revenue when usage is not contracted?
Usage-based companies typically report revenue on a billing-period basis: customers receive an invoice for the usage they consumed in the prior month or quarter. This is recognized as revenue when the obligation is satisfied (the usage has been delivered). Some usage-based companies have hybrid models that include a minimum committed spend plus consumption above that floor, which creates a combination of contracted revenue (recognized ratably) and variable revenue (recognized on usage). Remaining Performance Obligations (RPO) disclosed under ASC 606 capture the committed portion but not the variable portion, so RPO is less comprehensive for purely usage-based businesses than for seat-based SaaS.