A SaaS company can grow annual recurring revenue by 40% and still watch gross margin fall from 78% to 61% in the same year. That is not a growth problem, it is a billing model problem. Usage-based pricing does not erode margin on its own, it exposes cost structures that flat-rate pricing was quietly hiding for years.
What is the Concept
Usage-based billing (UBB), also called consumption or metered pricing, charges customers based on what they actually use, such as API calls, compute minutes, storage gigabytes, or AI tokens processed. Gross margin in this model is calculated as revenue minus cost of goods sold (COGS), where COGS now scales dynamically with customer behaviour instead of staying fixed like it did under seat-based pricing.
The core issue is that most finance teams still model COGS as a static percentage of revenue, carried over from flat-rate days. In a metered model, COGS is a moving target driven by cloud compute pricing, third-party API costs, support ticket volume tied to heavy users, and infrastructure redundancy. When these variables are not tracked per unit of usage, margin leaks accumulate invisibly until a quarterly finance review reveals the damage.
Why It Matters in Canada (2025–2026 Context)
Canadian SaaS companies face a currency layer that most US-based playbooks ignore. AWS, Azure, and Google Cloud bill in USD even when workloads run in the Canada Central (Montreal) or Canada East regions, so a weakening loonie directly compresses gross margin on every metered customer, regardless of usage volume. A 5% CAD depreciation against USD can quietly erase a full margin point across an entire usage-billed customer base without a single operational change.
Toronto, Vancouver, Waterloo, and Montreal have become dense clusters of usage-priced SaaS and fintech products, from developer tools to vertical platforms in insurance and logistics. As Canadian scale-ups like Shopify, Clio, and Coveo have shown, the shift toward consumption pricing is now the default expectation from enterprise buyers, not a differentiator. That means Canadian founders are adopting UBB under competitive pressure, often before their finance stack can attribute COGS at the customer or feature level, which is exactly when margin erosion goes undetected the longest.
How AI Is Changing This
AI features have introduced a second, more volatile layer of usage cost on top of traditional cloud infrastructure: large language model inference. Every AI-powered feature billed as part of a subscription now carries a token cost that fluctuates with model provider pricing, prompt length, and customer behaviour. Unlike storage or compute, LLM API pricing has changed multiple times within a single fiscal year, and Canadian companies paying in USD absorb both the pricing volatility and the exchange rate volatility simultaneously.
The upside is that AI also makes real-time margin visibility possible for the first time. Canadian SaaS finance teams are starting to deploy anomaly-detection models that flag customers whose usage-to-cost ratio deviates from their pricing tier in near real time, rather than discovering the erosion at month-end close. This turns margin management from a reactive accounting exercise into a proactive, per-customer operating discipline.
Real-World Examples
Consider a Toronto-based API infrastructure company charging per API call. Its published gross margin target was 75%, but a cohort of enterprise customers running high-frequency automated workflows generated support tickets and rate-limit engineering work at a cost that flat per-call pricing never accounted for. Within two quarters, that cohort's true margin had dropped to 52%, dragging the blended company margin down by several points before the finance team built usage-tier-specific COGS tracking.
A Vancouver fintech offering usage-priced fraud detection faced a different leak: free-tier and sandbox API usage from prospective enterprise customers, intended for evaluation, was consuming compute at production-grade cost with zero revenue attached. Because sandbox usage was not capped or cost-isolated, it was quietly diluting company-wide gross margin by an estimated 3 to 4 percentage points, a pattern common among Canadian B2B SaaS firms competing for enterprise deals through generous trial usage.
Practical Insights / Actions
Canadian SaaS leaders can apply what we call the Margin Leak Ladder, a five-rung diagnostic for usage-based pricing: Rung 1, infrastructure volatility (cloud and LLM pricing changes plus CAD/USD exposure); Rung 2, support cost creep (ticket volume that scales with heavy usage but isn't billed for); Rung 3, free-tier overhang (evaluation or sandbox usage running at production cost); Rung 4, multi-cloud egress (data transfer fees between regions or providers, often invisible in COGS reporting); and Rung 5, metering blind spots (features bundled into a flat tier that should be metered separately). Auditing each rung quarterly, per customer segment, surfaces margin leaks months before they show up in blended P&L numbers.
Three specific actions compound quickly: first, attribute COGS at the feature and customer-tier level rather than company-wide, so a single heavy-usage enterprise account can't mask erosion in the broader base; second, hedge or contract USD cloud spend where CAD revenue dominates, since currency mismatch is a solvable operational risk, not a market inevitability; third, cap or cost-isolate free and trial usage so evaluation spend never silently competes with paying customer margin. None of these require rebuilding the pricing model, only rebuilding how cost is measured against it.
Future Outlook
Through 2026 and beyond, expect Canadian SaaS finance teams to adopt margin-aware pricing engines that adjust rate cards or usage caps automatically when a customer's cost-to-serve exceeds a defined threshold, rather than waiting for annual repricing cycles. AI cost governance, treating LLM and compute spend as a first-class line item with its own forecasting model, will move from a nice-to-have to a board-level reporting requirement, especially for companies raising growth capital where investors now scrutinize gross margin durability under usage pricing, not just top-line growth.
Founders who treat gross margin as a per-customer, per-feature metric rather than a single company-wide number will have a structural advantage in fundraising, pricing negotiations, and enterprise renewals, because they can defend unit economics with evidence instead of averages.
Conclusion
Usage-based billing does not inherently break gross margin; undifferentiated cost attribution does. Canadian SaaS companies that isolate infrastructure volatility, support cost creep, free-tier overhang, multi-cloud egress, and metering blind spots protect margin while still capturing the growth benefits of consumption pricing. If your finance stack still reports one blended margin number across your entire usage-billed customer base, that number is hiding more than it's revealing. RP SoftTech works with Canadian SaaS teams to build per-customer COGS attribution and margin monitoring into their billing architecture, request an audit of your usage-based pricing model to find where your margin is actually leaking.