Subscription pricing charges a fixed recurring fee for access, so revenue is predictable and the offer is simple. Usage-based pricing charges for what each customer consumes, so price tracks value and cost but revenue moves month to month. Choose a subscription when usage is steady and cheap to serve; choose usage when it varies widely or costs you per action; mix them when both apply.
Key takeaways
- Subscriptions win on predictability and simplicity; usage wins on fairness, expansion and margin protection.
- Usage revenue rises and falls with your customers' activity, so plan cash flow for the dips.
- The two classic mixes are usage bands inside a subscription and a minimum commitment.
- For AI products, credits are a subscription that behaves like usage.
Years ago I was the government national sales manager for a telecommunications company.
We had a usage-based pricing model. Customers paid for the calls they made.
One month, during an election, a government customer made a crazy number of conference calls. Our revenue shot up. We celebrated.
The next month, their spending dropped off a cliff. And we panicked.
Same customer. Same contract. Same product. The only thing that changed was how much they used it. That was my first real lesson in the difference between usage revenue and subscription revenue, and it's the question I want to answer here.
What is the difference between usage-based pricing and subscription pricing?
A subscription charges a fixed fee every month or year for access. Netflix, a gym membership, Microsoft 365. You pay the same whether you use it every day or forget it exists.
Usage-based pricing charges for what you consume. Your electricity bill. A pay-as-you-go phone plan. Twilio's per-message fees. Use more, pay more.
| Subscription (fixed fee) | Usage-based (pay for what you use) | |
|---|---|---|
| Revenue predictability | High | Low to medium |
| Simplicity of the offer | High: "X a month for A, B and C" | Lower: customers must estimate usage |
| Entry price for small customers | Fixed, can be a barrier | Low: start small |
| Revenue growth from existing customers | Needs upgrades or price rises | Automatic as usage grows |
| Customers who scale down | Often cancel | Usually stay and pay less |
| Margin on heavy users | At risk if usage costs you money | Protected |
| Billing and tracking effort | Low | Higher: you need metering |
| Cash flow | Steady | Lumpy, seasonal |
| Best for | Steady, low-cost-to-serve value | Variable, cost-per-action value |
The feast-or-famine problem with usage revenue
Here's the arithmetic I used in the book. If 350,000 people pay $15 a month for a subscription, you know your revenue: $5.25 million. Nice and simple.
With usage pricing, you don't know. A big customer might cut back. A small one might take off. Holidays might slow everyone down.
Let's put numbers on it.
Three ways to bill the same 100 customers
Subscription: 100 customers at $100 a month. Revenue is $10,000 in every month of the year.
Pure usage: the same customers average $100 a month in usage, but December activity drops 40%. December revenue: $6,000.
Hybrid: a $60 base fee plus usage that averages $40. In December, usage drops 40% to $24. December revenue: $6,000 base + $2,400 usage = $8,400.
Illustrative numbers.
The hybrid kept 84% of a normal month in the quiet month. Pure usage kept 60%. For a business with fixed wages and rent, that gap is the difference between a tight month and a scary one.
Is usage-based pricing better than subscription?
Sometimes. Here's the honest scorecard.
Where usage wins. In the book I named three key benefits:
- Costs scale for customers. A small business can start cheaply and pay more as it grows.
- High customer retention. When a customer shrinks, they scale down instead of canceling.
- A wider customer base. You can serve a sole trader and a multinational on the same price list.
There's a fourth benefit in 2026 that I didn't need in 2022: margin protection. When every action costs you money (an AI model call, a text message, a gigabyte of storage), usage pricing makes sure your heaviest customers pay their way. That's why GitHub moved all Copilot plans to included credits from June 2026.
Where subscription wins.
- Predictable revenue you can plan hiring and cash flow around.
- A simple offer a customer understands at a glance.
- No bill shock. Vercel launched a Flat Rate CDN option in September 2026 after customers complained about surprise bills. Its beta users said it "eased our billing anxiety."
- Higher valuation for committed revenue. Buyers pay more for revenue that's contracted.
So the real question isn't "which is better?" It's "how much predictability do I need, and how different are my customers?"
Worked example: where does a flat fee beat a percentage?
Owner, the restaurant marketing and ordering platform that reportedly crossed $100 million ARR in 2026, publishes two plans side by side:
- Flexible: $249 a month plus 5% per online order.
- Flat Rate: $499 a month.
Where do they cross?
Break-even sales = (Flat fee − Base fee) ÷ Usage rate = ($499 − $249) ÷ 5% = $5,000 a month
A restaurant doing $3,000 a month online pays $399 on Flexible, so it saves $100. One doing $12,000 pays $849 on Flexible, so Flat Rate saves it $350.
This is clever pricing. The small restaurant gets a low-risk entry. The busy one gets a predictable bill and a reason to upgrade. And by publishing the break-even, Owner lets the customer do the maths and pick, with no salesperson needed.
You can copy this tomorrow in almost any business. Put a usage plan and a flat plan next to each other, and show where they cross.
Which model fits your numbers?
Use the free Pricing Model Picker to compare flat, tiered, usage, credit and hybrid pricing for your business.
Two clever ways to mix usage-based pricing with subscription
In the book I said it's a good idea to bundle usage with subscription to give your business more certainty. There are two classic ways to do it.
Mix #1: Usage bands inside a subscription. You charge a monthly fee, and the fee steps up with usage. When I wrote the book, Leadfeeder started at $63 a month for up to 100 leads and reached $599 a month for 3,001 to 5,000 leads. ConvertKit priced on the size of your email list. The customer gets a fixed monthly bill, and you get paid more as they grow. Most AI credit ladders work this way today.
Mix #2: A minimum commitment. The customer commits to a minimum spend, then pays for usage on top. turbopuffer, the search database that reportedly hit $100 million ARR having raised less than $1 million, has monthly minimums of $16, $256 and $4,096 by plan. Pinecone has a $50 minimum on Standard. As I wrote in the book, you can set the minimum so it claws back what you spent acquiring the customer. (More on this in the minimum commitment model.)
There's a third mix that's even older: a base fee plus a per-use rate. Zipcar charged a small monthly membership and then an hourly rate per drive. The membership paid for the relationship; the hourly rate paid for the car.
All three are versions of hybrid pricing.
What about AI products?
AI made this debate urgent, because AI products have real costs per action. ICONIQ's 2026 survey found that only 15% of consumption-priced AI companies absorb all of their inference cost. The rest pass at least some of it on.
The most common answer is credits. A customer pays a fixed monthly subscription and gets a bucket of credits to spend. To the customer, it feels like a subscription. To you, it behaves like usage, because heavy users run out and buy more. ElevenLabs, Lovable, Runway and dozens of others work this way. If you're going that route, read the credit-based pricing guide and check your margins with the AI Gross Margin Calculator.
The infrastructure has caught up, too. Stripe completed its acquisition of usage-billing company Metronome in January 2026. Patrick Collison said "the shift toward usage-based models will be a defining feature of the next decade."
Should I choose subscription, usage or both?
Score your business against this table. Count which column you land in most often.
| Question | Points to subscription | Points to usage |
|---|---|---|
| Do customers of the same size use very different amounts? | No | Yes |
| Does each unit of use cost you real money? | No | Yes |
| Can you meter usage cheaply and accurately? | No | Yes |
| Do your customers need a fixed budget? | Yes | No |
| Can your cash flow handle a 30–40% dip in a quiet month? | No | Yes |
| Do customers often shrink rather than leave? | No | Yes |
| Do you want to sell to tiny and huge customers at once? | No | Yes |
Mostly left: run a fixed subscription, perhaps with tiers. Mostly right: usage, with caps. A mix: start with a base fee plus included usage, and charge for extras.
Free Subscription Profit Review
Australian business owners can book a free 30-minute review with a BPAG Accountants adviser. An honest conversation about:
- whether a subscription model could work for your business
- how recurring revenue could cover your fixed costs each month
- how it could lift what your business is worth
No cost, no obligation. General information only until we understand your situation.
What to do next
Pull last year's numbers and look at your quietest month. If you moved to usage, what would that month have looked like? If the answer scares you, add a base fee or a minimum commitment before you add a meter.
Then work out a profitable base price with the Subscription Price Calculator, compare models with the Pricing Model Picker, and see how real companies set their meters in our usage-based pricing examples. If you're in Australia, our guide to subscription cash flow covers the accountant's view of the switch.
Frequently asked questions
Is usage-based pricing still recurring revenue?
It's recurring if customers keep using you, but it isn't contracted. Investors and buyers usually value committed subscription revenue more highly than variable usage, and some 2026 companies report 'run-rate' figures that include usage which may not repeat. If you want usage revenue to count as recurring, add a minimum commitment or an annual prepaid allowance.
What is overage pricing?
Overage is what a customer pays for usage beyond what their plan includes. Midjourney sells extra Fast GPU time at $4 an hour, and Zendesk charges $2.00 per automated resolution pay-as-you-go beyond the included pool. Price overage a little higher than the included rate so the next plan up looks like better value, but not so high that it feels like a penalty.
Do customers prefer subscriptions or pay-as-you-go?
Consumers and small businesses generally prefer a fixed price they can budget for. Developers and larger companies are comfortable with metering when they can see and cap it. Even in AI support, where outcome pricing is on offer, Decagon says most customers choose per-conversation pricing because it's easier to predict.
How do I forecast usage-based revenue?
Track usage per customer every month, group customers into cohorts by start date, and project each cohort's usage curve forward. Watch net revenue retention closely, because expansion and contraction drive most of the change. Seasonal businesses should model at least one quiet month. A base subscription or minimum commitment gives you a floor to forecast from.
Is pay-per-use better for a small local business?
Rarely on its own. Tracking and billing usage takes systems, and the cash-flow swings hurt a small business more. Most local businesses do better with a fixed membership that includes a set amount (four washes, two visits, ten classes) and a simple price for extras. That's usage-based thinking inside a subscription.
Get both books free
The Subscription Playbook (ebook + audiobook) and Feed a Starving Crowd, instantly. In Australia? Claim the free hardcover.
Robert Coorey, MBA
Robert is the author of The Subscription Playbook and Feed a Starving Crowd and a Director at BPAG Accountants in Sydney. More about Robert.
Sources
- Owner pricing (restaurant platform)
- turbopuffer pricing (checked 27 Oct 2026)
- Pinecone pricing (checked 27 Oct 2026)
- SaaStr summary of ICONIQ's 2026 State of AI report (10 Jul 2026)
- Stripe completes Metronome acquisition (14 Jan 2026)
- Vercel: Introducing Flat Rate CDN (Sep 2026)
- Decagon: Pricing AI agents
Facts and prices checked 27 October 2026. Spotted something out of date? Tell us.