What usage-based pricing means
Usage-based pricing means the customer pays in proportion to how much they use your product, measured in whatever unit reflects value: API calls, seats active in a given month, GB stored, or workflows run.
It works when your product's value scales directly with consumption. A customer running 10x the workflows through your platform is getting 10x the value, so it feels fair for them to pay 10x more. This is the core reason usage-based pricing has grown from roughly 27% of SaaS companies in 2023 to about 38% today, according to SaaS benchmarking research from OpenView and High Alpha. AI-native products are accelerating this shift specifically, since inference costs scale with usage and founders want revenue to scale the same way.
The tradeoff is predictability. Usage-based-only companies see 25 to 40% more month-to-month revenue volatility than subscription peers, which makes forecasting harder and can spook investors who are used to clean MRR charts.
What subscription pricing means
Subscription pricing means the customer pays a fixed amount per billing period, usually monthly or annually, regardless of how much or how little they use the product that month.
Put simply, you are billed for access, not consumption. This works best when your product delivers a consistent baseline of value every period, when usage is hard to meter cleanly, or when your buyer's finance team needs a predictable line item to approve the purchase.
Subscription pricing also fits products with a learning curve, where usage is low in month one and climbs as the customer ramps up. Charging by usage in that window would undercharge you during onboarding, exactly when your support costs are highest.
The mistake most early-stage founders make
Most founders pick a pricing model by copying whoever they admire, not by looking at their own cost structure. If a competitor uses usage-based pricing, they copy it. If Slack uses per-seat, they copy that instead.
The actual question is narrower: does your marginal cost per unit of usage go up in a way your customer can see and understand? If a customer running your product harder makes your infrastructure bill go up in a way you can point to, usage-based pricing lets you pass that cost through fairly. If your costs stay flat no matter how hard a customer uses the product, a subscription is simpler for everyone and you keep 100% of any upside from heavy users.
The second mistake is picking one model and assuming it is permanent. Pricing is not a launch decision, it is a live system you revisit as your cost structure and customer base change. This is the same instinct that trips founders up when they skip straight to a go-to-market plan before they know who they are actually selling to.
Usage-based pricing vs subscription pricing: the decision framework
Six questions decide which model fits your product. For each signal below, the model listed first is where the evidence points.
- Cost structure. Usage-based fits if marginal cost rises with usage (compute, storage, API calls). Subscription fits if marginal cost stays roughly flat regardless of usage.
- Value delivery. Usage-based fits if value scales directly with volume used. Subscription fits if value is delivered as ongoing access, not volume.
- Customer finance process. Usage-based fits if your buyer accepts variable spend for variable value. Subscription fits if your buyer needs a fixed number to get budget approved.
- Product maturity. Usage-based fits if usage is easy to meter cleanly and customers trust the meter. Subscription fits if usage is hard to define or customers distrust metering.
- Revenue predictability needs. Usage-based fits if you can tolerate 25 to 40% more month-to-month volatility. Subscription fits if you need clean, forecastable MRR for planning or fundraising.
- Onboarding curve. Both models face slow early usage, but usage-based makes low early bills feel fair, while subscription lets a flat fee fund your onboarding cost regardless of ramp speed.
Most companies do not land purely on one side. Roughly half of SaaS companies today combine a flat subscription base with a usage-based component on top, usually a base fee that covers the floor cost of serving a customer, with metered charges once usage passes a threshold, a pattern documented in recent SaaS benchmarking research from High Alpha. This hybrid model exists because it solves the actual tension: subscriptions fund your fixed costs, usage lets you capture upside from your heaviest users without appearing to punish growth. It is worth revisiting once you have nailed down your ideal customer profile, since who you are pricing for shapes which side of this table you land on.
Real examples of each model working
Twilio built its entire pricing model on usage: pay per SMS or API call. This works because Twilio's own infrastructure cost is directly tied to message volume, so passing that cost through transparently was always going to be fair to the customer and profitable for Twilio.
Basecamp has run flat subscription pricing for years specifically because project management software delivers consistent value regardless of how many tasks a team creates that month. Metering task creation would add billing complexity with no real fairness benefit to the customer.
Snowflake and most modern data warehousing tools default to usage-based compute pricing because query volume is both the actual cost driver and the actual value delivered. A team running more analysis is both consuming more resources and getting more value, so the two move together cleanly. Revenue volatility from this kind of consumption pricing is real. Orb's research on usage-based revenue puts it 25 to 40% higher month-to-month than subscription peers. Founders in this category accept that tradeoff only because consumption and value are genuinely tied together, so the volatility is explainable rather than random.
The 30-day move before you commit
Do not redesign your entire pricing page before you have data. Instead, instrument usage tracking for your 10 to 20 most active customers this month, even if you are still charging flat subscription fees. Log what "usage" would even mean for your product: seats, API calls, records processed, whatever unit maps to value.
At the end of 30 days, look at the spread. If your heaviest users are using the product 5 to 10x more than your lightest ones and your costs scale with that usage, you have a real signal to introduce a usage-based component. If usage is roughly flat across your base regardless of who is a power user, subscription pricing is already the right model and metering would only add friction.
Frequently asked questions
Is usage-based pricing better than subscription pricing for SaaS?
Neither is better by default. Usage-based pricing is better when your costs and customer value both scale with consumption. Subscription pricing is better when value is delivered as consistent access and your costs stay flat regardless of usage.
Why are more SaaS companies switching to usage-based pricing?
Usage-based pricing adoption grew from about 27% of SaaS companies in 2023 to roughly 38% today, largely driven by AI features where inference costs scale directly with usage, making flat-fee pricing a losing structure for the vendor.
What is hybrid pricing in SaaS?
Hybrid pricing combines a flat subscription base, which covers your fixed cost of serving a customer, with a usage-based component on top that captures additional value from heavier users. Close to half of SaaS companies now use some version of this model.
Does usage-based pricing hurt revenue predictability?
Yes, measurably. Companies running pure usage-based models see 25 to 40% more month-to-month revenue volatility than subscription peers, which matters for forecasting and can complicate fundraising conversations if you cannot explain the swing.
Should an early-stage startup start with usage-based or subscription pricing?
Start with whichever model matches your actual cost structure today, not the one you expect to need at scale. Most early-stage products are simpler to price on a flat subscription first, then introduce usage-based components once you have real data on how customer value and your costs scale together.
Can you switch pricing models after launch?
Yes, and most companies do. Pricing is a system you revisit as your cost structure and customer base evolve, not a one-time decision. Grandfather existing customers on their current terms when you switch to avoid unnecessary churn.
Pricing is one of the few decisions that touches your cost structure, your fundraising story, and your customer relationships all at once. Get the model right for where your business actually is today, instrument the data to know when it changes, and treat the switch as routine maintenance, not a crisis. For more frameworks like this one, see the rest of our founder-facing GTM and pricing playbooks.