Usage-based pricing charges according to how much a customer consumes rather than how many people have access. Its central decision is the value metric: the unit you bill on. The model produces stronger expansion revenue than seat pricing when the metric genuinely tracks value, and produces unpredictable bills and customer anxiety when it does not.
When your cost or the customer's value genuinely varies with consumption. It is a poor fit where usage is flat, where the buyer cannot forecast their own consumption, or where growth in the metric does not correspond to growth in value.
What usage-based pricing is
Usage-based pricing charges according to how much a customer consumes rather than how many people have access to the product.
The model is old in utilities and telecoms. It became mainstream in software through cloud infrastructure pricing in the 2010s, and accelerated again with AI products, where the cost of serving a customer genuinely varies per request rather than being effectively zero.
Its central decision is not the price. It is the value metric: the unit you bill on. That single choice determines whether revenue grows with customer success, whether buyers can predict their bill, and what your own team ends up optimising for.
- The value metric is the decision Not the price
The unit you bill on determines expansion, predictability and what your team optimises for.
- Check value actually scales with use The precondition
Frequently assumed. If it does not hold, usage pricing adds volatility and produces no expansion.
- Five tests, run against your product Not borrowed
Most metrics in wide use fail at least one of them in some context.
- Hybrid is now the majority Base fee plus usage
Pure consumption was too volatile for both sides. A floor buys predictability.
Why it matters
The appeal is real: revenue grows as customers get more value, without a new sale and without a renegotiation. Published benchmarks put median net revenue retention for usage-based companies around 120%, against roughly 110% for seat-based, which compounds substantially over a few years.
The catch is that this only holds if the metric genuinely tracks value. Where consumption is flat, or decoupled from what the customer gets, usage pricing delivers all of the unpredictability and none of the expansion.
Finance usually finds out first, halfway through a quarter, when the forecast turns out to have been a hope with a spreadsheet around it.
- Freemium free-to-paid 2–5%
Typical free-to-paid conversion for freemium products. A quarter of products sit below 2.5%.
- Free-trial conversion 15–25%
Free-trial conversion, for comparison. A different model with a different funnel shape.
- Net revenue retention, usage-based ~120%
Median. Above 100% means the existing base grows without new sales.
- Net revenue retention, seat-based ~110%
Median. Still expanding, more slowly.
- On a hybrid model now 43%
Projected to reach 61% within a year.
- Using seats as the only metric 8%
Down from a large majority. Most still use seats as one component.
Ranges rather than single figures, because the published surveys behind them use different samples and definitions. Treat them as the shape of the market, not as targets.
Worth noticing how rarely usage-based pricing is actually recommended when something is choosing on the merits rather than following the discourse. Across 173 pricing recommendations from ShipFit’s engine, it came up five times. Five. Meanwhile the trade press has spent three years announcing that seats are dead.
Asked to price 173 different products, an AI recommended freemium-plus-tiered 73% of the time and positioned 91% of them as mid-market.
- Freemium and tiered
- 127 ideas 73.4%
- Hybrid, multiple mechanisms
- 36 ideas 20.8%
- Usage-based
- 5 ideas 2.9%
- Positioned mid-market
- 158 ideas 91.3%
Sample: n = 173 ideas that reached pricing; 98% were SaaS
What it does not say: Almost the entire sample was SaaS, which limits how far this generalises. It is the engine’s recommendation, not an observed price.
This one is unflattering to us, and we are publishing it anyway. A tool that only reports the numbers making it look good is not reporting numbers.
Choosing the metric
- It rises as the customer gets more value
Fails when: Seats, when one power user does all the work and the rest never log in.
- The customer can predict their bill
Fails when: Raw API calls, when nobody in the buying organisation knows how many they will make.
- It is countable without argument
Fails when: "Value delivered", or anything requiring a quarterly negotiation about what counts.
- Growing it is good for the customer, not just for you
Fails when: Storage, which grows because deleting things is hard rather than because they gained anything.
- It does not punish the behaviour you want
Fails when: Per-seat, on a collaboration tool whose whole value is getting more people in.
The test people skip is the fourth. Growing the metric has to be good for the customer. Storage is the classic failure: it grows because deleting things is effortful, not because anyone gained anything, so charging for it captures inertia. Customers do not object immediately, and they object eventually.
Model the bill from their side
Before committing, take three real customers and calculate what they would pay at low, expected and high usage.
If the high case produces a number that would trigger a procurement review, you have not designed a pricing model, you have scheduled a churn event. The buyer will hit it during a busy month, escalate internally, and the conversation that follows will be about your bill rather than about your value.
This is also where usage anxiety gets solved or ignored. Live spend visibility, alerts before thresholds, hard caps and a published overage rate cost very little to build. They remove the single most common reason a buyer chooses a more expensive predictable competitor.
When to use it
- Your cost of serving a customer genuinely varies with their usage.
- Customers getting more value demonstrably consume more.
- Your buyers are technical and already accustomed to metered billing.
- Seat-based pricing is capping accounts that would otherwise expand.
- Usage varies enough between customers that one price fits nobody.
- Value does not scale with consumption. Use Value-Based Pricing →
- You need a price band from buyer perception. Use Van Westendorp →
- You need distribution more than revenue right now. Use Freemium Strategy →
- Your buyers cannot forecast their own consumption. Use a hybrid model →
Against the alternatives
How much did they use?
Gives you: Revenue that expands with success, and a bill they may not predict
How many people can log in?
Gives you: Predictable revenue, and a cap on accounts that grow in value not headcount
What is the floor, and what varies above it?
Gives you: Predictability plus expansion. Now the majority model
One price for everyone?
Gives you: Easiest to sell, and leaves the most money on the table across a varied base
Usage pricing in practice: Twilio
The case that made the model fashionable, and the one condition it depended on.
Twilio · to IPO in 2016
Usage pricing where existing customers grew the account faster than new ones were signed.
Twilio charged per message and per minute. A customer began by paying almost nothing, because a developer testing an idea sends very few messages, and paid more as the thing they built succeeded.
The IPO filing reported a dollar-based net expansion rate above 150%, meaning the same cohort of customers spent more than half as much again a year later without a single upsell conversation.
The condition that makes this work is easy to state and hard to satisfy: consumption had to rise with the customer succeeding. Where it does not, usage pricing delivers all of the revenue unpredictability and none of the expansion.
- Dollar-based net expansion at IPO
- above 150%
- Entry cost for a developer
- near zero
- Upsell conversations required
- none
What it shows: The metric has to be something that goes up when the customer is winning. That is a much narrower condition than "we can meter it".
When it won’t help you
- It produces revenue you cannot forecast
Consumption drops in quiet months, over holidays and when a customer has a bad quarter. Pure usage revenue is genuinely volatile, which complicates hiring, planning and fundraising conversations.
Instead: Add a floor. A platform fee plus a usage layer is the majority model precisely because of this.
- Usage anxiety costs you deals you never hear about
A buyer who cannot predict their bill has to defend an unknown number internally. Many quietly choose a more expensive predictable competitor, and you never learn that was the reason.
Instead: Ship spend visibility, alerts and caps before you ship the pricing. They are cheap and they are the whole objection.
- It can misalign your own roadmap
Once revenue is tied to a number, that number becomes visible to everyone. If the metric rewards inefficiency, someone will eventually optimise for it, and it will not feel like a decision at the time.
Instead: Choose a metric where the customer using less is a success you can still charge for, or accept that you have built in a conflict.
- It is harder to sell in a procurement process
Enterprise buyers need a number for the budget. A variable bill requires a committed minimum, a cap, or both, which reintroduces the negotiation that usage pricing was meant to avoid.
Instead: Offer a committed-spend tier for buyers who need certainty. Most large customers will take it.
Further reading
- Value-Based Pricing. What the metric should ultimately be anchored to.
- Van Westendorp. How to run a price survey when there is no single monthly number.
- Freemium Strategy. The other model where the metric decision dominates.
- CAC / LTV ratio calculator. Whether expansion revenue is arriving fast enough to matter.
- ARR / MRR. What variable revenue does to the numbers you report.
How to apply Usage-Based Pricing
- 1
Establish whether value actually scales with consumption
This is the precondition and it is frequently assumed rather than checked. If a customer getting twice the value does not use twice as much, usage pricing will not produce expansion, it will just make the bill unpredictable for no gain.
- 2
Generate candidate metrics and test each one
Requests, records, seats, workflows, completed jobs, gigabytes. Run each against the five tests: does it rise with value, can the customer predict it, is it countable without argument, is growing it good for them, and does it avoid punishing the behaviour you want.
- 3
Model the bill from the customer's side, not yours
Take three real customers and work out what they would pay at low, expected and high usage. If the high case is a number that would trigger a procurement review, you have a churn event scheduled rather than a pricing model.
- 4
Add a floor, and probably a platform fee
Pure consumption pricing gives you revenue that can go to zero in a quiet month, which makes forecasting and fundraising hard. A base fee plus a usage layer is now the majority model, and it exists because pure usage was too volatile on both sides.
- 5
Give the customer visibility and controls
Usage anxiety kills consumption models more often than price does. Live spend visibility, alerts, caps and a predictable overage rate cost very little to build and remove the reason buyers hesitate.
- 6
Watch what the pricing does to product behaviour
You have just tied revenue to a number, and everyone in the company can now see it. If the metric rewards inefficiency, someone will eventually optimise for the metric rather than for the customer, and it will not feel like a decision when it happens.
Common mistakes
- **Assuming value scales with consumption without checking.** If a customer getting twice the value does not consume twice as much, usage pricing adds unpredictability and produces no expansion.
- **Choosing a metric the buyer cannot forecast.** Raw API calls are countable and meaningless to the person signing. A bill they cannot predict is a bill they will not approve.
- **Billing on something that grows for the wrong reason.** Storage grows because deleting is hard. Charging for it captures inertia rather than value, and customers notice eventually.
- **Going pure consumption with no floor.** Revenue that can fall to zero in a quiet month makes forecasting hard for you and makes budgeting hard for them. Hybrid exists for a reason.
- **Shipping without spend visibility.** Usage anxiety kills more consumption deals than price does, and the fix is cheap: live usage, alerts and caps.
- **Ignoring what the metric does to your own roadmap.** Tying revenue to a number makes that number visible to everyone, and a metric that rewards inefficiency eventually gets optimised for.
How ShipFit operationalizes this
ShipFit runs Usage-Based Pricing as one of the monetisation models in Stage 4 (How to Win?), and applies it in Stage 6 (How to Charge?) among the pricing models selected by product type. Stage 6 asks what the value metric is before it asks what the price is, because the metric constrains everything downstream, and where value does not scale with consumption it recommends a seat or tier model rather than defaulting to usage.
ShipFit runs 55 frameworks across 9 decision stages
Usage-Based Pricing is one tool in a bigger toolkit. The full library covers market sizing, buyer discovery, MVP scoping, pricing, and launch.
The Mom Test
Q3Rob Fitzpatrick
Validation question methodology, real interviews, not theater
Jobs-to-be-Done
Q2-Q4Clayton Christensen
Functional, social, and emotional jobs your product fulfills
7 Powers
Q4Hamilton Helmer
Strategic moats: Scale, Network, Counter-positioning, Switching, Brand, Cornered Resource, Process
Van Westendorp PSM
Q6Feature-weighted price sensitivity analysis without guessing
Blue Ocean Strategy
Q4Kim & Mauborgne
ERRC framework: Eliminate, Reduce, Raise, Create
Fake Door Testing
Q7Pre-build behavioral validation with landing pages and apology modals
+ 49 more: TAM/SAM/SOM Analysis, Porter's Five Forces, Market Timing Analysis, Unit Economics (LTV/CAC)...
Frequently asked questions
What is usage-based pricing?
How do I choose a value metric?
Is usage-based pricing better than per-seat?
What is hybrid pricing?
Why do customers dislike usage-based pricing?
Keep exploring
The 9-step playbook from market verdict to ship-ready spec.
The 7 Powers, each with its benefit and its named barrier, plus the Power Progression that decides which of them a startup can realistically build and when.
Validated learning, the build-measure-learn loop, what an MVP actually is, the three engines of growth, and the ten pivots Ries names rather than one.
Most early-stage competitive analysis is a 2x2 with your product in the top-right quadrant. The real version is harder, more boring, and tells you whether you can actually win.
Most product launches fail not because the product was bad, but because the launch was a list of channels nobody mapped to a buyer. Here is the template that fixes that.
Founder equity that won't blow up the cap table in year three.
Run six honest tests. (1) Is there a defined buyer with budget? (2) Is the pain frequent and intense, not annoying? (3) Will buyers commit money or time? (4) Do you have a defensible angle (one of the 7 Powers)? (5) Does the unit economics math work at a price they'd pay? (6) Is the timing present-tense, not 'someday'? Pass four of six and keep going. Fail four and walk away. The rest is theatre.
Fintech idea validation that tests demand, trust, and willingness to pay before you sink months into a regulated build. Forces 9 decisions. Start free.
Idea validators give you a fast score or grade. ShipFit gives you 9 forced decisions and a build-ready playbook. A number is a starting point; a plan is what you act on. Pick based on what you'll do next.
Ready to make your next product a success?
9 decisions between your idea and a product worth building.