Usage-based pricing examples include Twilio (from $0.0083 per SMS), Snowflake ($2 to $4 per compute credit), Deepgram (from $0.0043 per audio minute), Fin ($0.99 per resolved support conversation), Stripe Billing (0.7% of billing volume) and Procore (priced on annual construction volume). Each charges for a unit that grows as the customer gets more value.
Key takeaways
- The best meters are units the customer already counts: messages, minutes, orders, resolutions.
- Almost every example pairs usage with something predictable: a minimum, an allowance or a flat option.
- Outcome meters (per resolution, per hire) are the newest form of usage pricing.
- The failures share one trait: customers couldn't predict the bill.
If Netflix used usage-based pricing, your bill would depend on how many shows you watched.
That's a terrifying thought for me. I know how easy it is to lose a weekend to a Netflix binge. (If Netflix billed by the episode, some months I suspect my bill would rival the mortgage.)
Netflix doesn't do it, and it's right not to. But for thousands of other businesses, charging by use is exactly the right call. The trick is picking what to count.
I've gathered 30 real examples below, checked in October 2026, and grouped them by the meter they use. Then I'll show you the patterns, the failures and what a normal business can borrow.
What are the best usage-based pricing examples?
Here are 30, grouped by the type of meter. Prices are in US dollars and are list or starting rates as published on each company's pricing page or in the dated source noted.
Communications: per message and per minute
| Company | What it meters | Rate (Oct 2026) |
|---|---|---|
| Twilio | SMS sent or received (US) | From $0.0083 per message |
| Twilio | Voice calls (US) | $0.0085 per minute to receive, $0.014 to make |
| Twilio | WhatsApp messages | From $0.005 per message |
| Twilio | Verification (one-time codes) | From $0.05 per verification |
| ElevenLabs | AI voice agent minutes | $0.08 per minute on every tier |
| Attentive | SMS marketing messages sent | Priced by messages (from the book) |
AI, speech and search: per token, minute or request
| Company | What it meters | Rate (Oct 2026) |
|---|---|---|
| OpenAI API | Tokens (GPT-6.1 Sol) | $2 per million input, $10 per million output |
| Anthropic API | Tokens (Claude Sonnet 5.5) | $2 per million input, $10 per million output |
| Deepgram | Pre-recorded transcription | From $0.0043 per audio minute |
| AssemblyAI | Pre-recorded transcription | $0.21 per audio hour (Universal-3.5 Pro) |
| Exa | Web search requests | $4 to $12 per 1,000 requests by depth |
| Firecrawl | Pages scraped | 1 credit per page; plans from $16 a month (annual) |
Data and infrastructure: per gigabyte, compute or user
| Company | What it meters | Rate (Oct 2026) |
|---|---|---|
| Snowflake | Compute credits | $2.00 (Standard) to $4.00 (Business Critical) per credit on demand |
| Snowflake | Storage | $23 per TB per month on demand |
| Databricks | Databricks Units (DBUs) | From $0.07 to $0.40 per DBU by product |
| Supabase | Monthly active users above plan | $0.00325 per MAU |
| turbopuffer | Storage, writes, queries | Up to $0.33 per GB stored, with minimums of $16, $256 or $4,096 a month |
| Pinecone | Read and write units, storage | Standard plan $50 monthly minimum |
| Vercel | Serverless compute | Active CPU from $0.128 per hour |
| Netlify | Credits for deploys, bandwidth, compute | Production deploy 15 credits; bandwidth 20 credits per GB |
Outcomes: per resolution or per result
| Company | What it meters | Rate (Oct 2026) |
|---|---|---|
| Fin (formerly Intercom) | Resolved support conversation | $0.99 per outcome |
| Zendesk | Automated resolution beyond the included pool | $1.50 committed, $2.00 pay-as-you-go |
| HubSpot | Resolved conversation (Customer Agent) | $0.50 (50 credits) |
| Salesforce | Help Agent resolution | $2 per resolution |
| Paraform | Successful hire | Per hire; third parties estimate 20–25% of first-year salary |
Business volume: per order, transaction or project size
| Company | What it meters | Rate (Oct 2026) |
|---|---|---|
| Stripe Billing | Billing volume | 0.7% of billing volume (pay as you go) |
| Stripe payments | Card charges (US) | 2.9% + 30¢ per successful charge |
| Owner | Online restaurant orders | $249 a month + 5% per order (or $499 flat) |
| Xero (Australia) | Online bill payments beyond plan | A$0.35 each |
| Procore | Annual construction volume | Custom annual fee, unlimited users |
That's 30. You'll find more in our AI pricing examples and outcome-based pricing examples, and full breakdowns in teardowns like Twilio and Snowflake.
What do the good examples have in common?
Look down those tables and four patterns jump out.
1. They count something the customer already counts. A clinic knows how many reminder texts it sends. A support manager knows how many tickets came in. A restaurant knows its online sales. Nobody needs a lesson to understand the bill.
2. The meter rises with the customer's success. More orders, more resolutions, more construction. That's the "land and expand" factor from the book: you land small and the price grows as they do.
3. Almost nobody uses pure usage. Look again. turbopuffer and Pinecone have minimums. Owner offers a flat alternative. Zendesk includes a pool of resolutions before charging. Supabase charges a platform fee and then overage. As I said in the book, bundling usage with subscription gives your business more certainty.
4. They make the bill predictable for the buyer. Supabase ships its Pro plan with a spend cap on by default. ElevenLabs charges the same per-minute agent rate on every tier, so customers can work out costs at any size.
Value Metric FinderScore candidate value metrics and pick the one that grows with your customers.Open the free tool →Worked example: what would these meters cost a small business?
Let's take a 6-practitioner allied health clinic and price three tools it might use.
One clinic, three usage bills
Appointment reminders by SMS: 2,000 texts a month × $0.0083 = $16.60 (Twilio US list rate).
Call transcription for notes: 120 hours of consults a month × $0.21 = $25.20 (AssemblyAI pre-recorded rate).
AI front-desk chat: 400 patient questions resolved a month × $0.99 = $396 (Fin's outcome rate), plus helpdesk seats.
US list rates used for illustration. Australian pricing, local carriers and GST will change the totals.
Two lessons. First, raw usage costs for messaging and transcription are tiny. If you resell them inside a service, the margin comes from your packaging, not the meter. Second, outcome pricing is priced against the cost of a human doing the job, not the cost of the technology. $0.99 a resolution is cheap next to a receptionist's time, which is exactly the point.
What should your meter be?
Score your candidate metrics in five minutes with the free Value Metric Finder.
Which usage pricing examples went wrong?
You learn as much from the failures. Every one of these is a bill-shock story.
When the customer can't predict the bill
Replit, 2025. Replit moved to "effort-based pricing" in June 2025. After its Agent 3 launched that September, one user reported spending $1,000 in a week against $180 to $200 a month before. Replit later added cheaper modes and cut cloud prices. (Read the Replit case study.)
Cursor, 2025. Cursor's Pro plan moved from 500 fast requests to "$20 of usage at API rates." Users ran out fast and the CEO apologized: "we didn't handle this pricing rollout well and we're sorry."
Netlify, 2025–2026. Netlify moved new accounts to credits. Its pricing page says that when one project exceeds limits, "all projects on your account will be paused." Some users on Reddit called the credits "abstract and unclear."
The fix in every case is the same: publish what typical tasks cost, include an allowance, cap by default and never switch a customer off because of a meter. Lovable does this well. Its pricing page lists the credit cost of example prompts ("make the button gray" 0.50 credits, "add authentication" 1.20).
How can a normal business use usage-based pricing?
You don't need to be Snowflake. Here's how the same idea looks in everyday businesses. These are illustrative, not benchmarks.
| Business | Possible usage meter | How to keep it predictable |
|---|---|---|
| Bookkeeper | Transactions reconciled per month | Bands: up to 200, 201–500, 501–1,000 |
| Accounting firm | Entities or returns lodged | Fixed monthly fee per entity, extras quoted |
| IT support / MSP | Devices managed | Per-device monthly fee with a minimum |
| Commercial cleaner | Rooms or square meters serviced | Monthly plan sized to the site |
| Marketing agency | Campaigns or content pieces delivered | Monthly retainer with an included number |
| Car wash | Washes per month | Unlimited club, or a pack of four |
| Print shop | Pages printed | Monthly allowance plus a per-page rate |
Notice that almost every row turns usage into a band or an allowance. That's deliberate. Small businesses rarely have the billing systems for second-by-second metering, and their customers want to know the price in advance. As I said in the book, the investment to track usage accurately can outweigh the benefit for some smaller businesses. Bands give you most of the upside with none of the tracking headache.
If you sell AI-powered services, read usage-based pricing for AI products and size your credits with the Credit Pack Designer.
What to do next
Pick the table above that looks most like your business. Then write down the one thing your best customer does more of every month when your product is working. That's your candidate meter.
Test it with the Value Metric Finder, check whether it should sit inside a subscription with the Pricing Model Picker, and read the full usage-based pricing guide for the four factors that make it work.
And if you want to see one company that has kept usage pricing honest for years, read how Procore did it at $1.3 billion.
Frequently asked questions
What is the most common usage metric?
For AI products, credits and tokens. For infrastructure, gigabytes, compute time and requests. For communications, messages and minutes. Among the 44 companies that reportedly crossed $100 million ARR in 2026, 18 of the 33 with visible pricing charged mainly on consumption of some kind, with credits bundled into subscription tiers the most common pattern for prosumer AI products.
Do usage-based companies show their prices publicly?
Developer-facing ones almost always do, because developers want to calculate costs before they build. Twilio, Deepgram, Exa and turbopuffer publish per-unit rates. Enterprise-focused companies often don't: Sierra, which prices per outcome, publishes no price list at all, and third parties estimate its rates.
What is a usage-based pricing model with minimum commitment?
It's usage pricing with a floor. The customer commits to a minimum spend per month or year and pays for usage above it. turbopuffer's plans have monthly minimums of $16, $256 and $4,096, and Pinecone's Standard plan has a $50 minimum. The minimum gives you predictable revenue and covers the cost of acquiring and serving the account.
Is Netflix usage-based?
No. Netflix charges a flat monthly fee by plan, regardless of how much you watch. Its 'extra member' fee ($7.99 with ads or $9.99 without, in the US) is a per-person charge, not a usage charge. Flat pricing suits Netflix because the cost of one more hour of streaming is tiny compared with the value of a predictable bill.
What's the difference between usage-based and outcome-based pricing?
Usage pricing charges for activity (a conversation handled, a minute transcribed). Outcome pricing charges only for a result (a conversation resolved, a hire made). Fin charges $0.99 per outcome and nothing if the customer asks for a human. Outcome pricing is a narrow form of usage pricing that works where the result can be counted and agreed.
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
- Twilio pricing (checked 27 Oct 2026)
- Snowflake pricing options (checked 27 Oct 2026)
- Deepgram pricing (checked 27 Oct 2026)
- Intercom (Fin) pricing (checked 27 Oct 2026)
- Stripe Billing pricing (checked 27 Oct 2026)
- Procore full-year 2025 results (12 Feb 2026)
- The Register: Replit Agent 3 pricing complaints (18 Sep 2025)
- Netlify pricing (checked 27 Oct 2026)
Facts and prices checked 27 October 2026. Spotted something out of date? Tell us.