Usage-Based Pricing and Churn: What Happens to Retention When Revenue Is Metered
Usage-based pricing hides revenue loss instead of preventing it, since a customer can drop usage 80%+ and stay "retained" on a logo-churn count, so tracking usage trend, revenue churn rate, and contraction MRR together catches the loss weeks before it shows up in MRR.
Usage-based pricing doesn't eliminate churn. A customer can stop using your product entirely, drop their monthly charge to near zero, and never show up in your logo-churn count.
I've spent the last few years consulting on retention for SaaS, e-commerce, and subscription businesses. The most consistent pattern I see is teams optimizing save rate without asking why people leave. In a usage-based product, that leaving often happens with no cancellation at all.
Get this right and you'll know which metric to watch before the lost revenue shows up in your monthly report.
Key takeaways
- Track revenue churn rate and contraction revenue on a metered product.
- A customer can stay active while their revenue contribution falls over 80%.
- Watch usage trend first, since a declining curve predicts revenue loss weeks early.
- Landbot raised net revenue retention 26% on usage pricing, with volatile months.
Does usage-based pricing reduce churn?
Usage-based pricing doesn't reduce churn, but it changes what churn looks like by removing the clean cancellation and replacing it with slow revenue contraction. Metered billing means a customer pays for what they use, with no fixed subscription commitment.
So the question isn't whether they cancel. It's whether they keep using you.
In a flat subscription, a customer has to cancel to stop paying. With a consumption model, they can cut usage to almost nothing and stay active. Your logo-churn rate still counts them as retained, but the revenue is gone, and no cancellation report tells you it left.
Picture an AI writing tool where usage drops 99% over four months with no cancellation filed. Logo churn still lists them as retained. The account looks alive on a customer count and nearly dead on a revenue chart.
Usage pricing can cut churn at one stage of the lifecycle, though. When it lowers the barrier to start, it improves trial-to-paid conversion and keeps early customers who would have felt locked in.
Zocdoc, for one, reported a 50% drop in churn after switching to usage pricing in 2017. So usage pricing can help with early-lifecycle churn while still hiding the mid-lifecycle contraction, which is what the rest of this article covers.
Where usage-based pricing hides revenue loss (the three contraction patterns)
Three patterns drive most silent revenue loss on a consumption product, namely usage decay, drop-to-floor, and commit shortfall.
Each one shrinks an account's revenue while the customer stays on the books. It shows up as contraction monthly recurring revenue (MRR), the revenue you lose from customers who stay but pay less.
That distinction is why these losses stay hidden. A standard churn-rate formula counts customers in and customers out. A contracting customer is still in, so the formula scores the account as healthy and you never see the loss.
You end up reading a retention number that says everything is fine while the revenue per account keeps falling.
Take a transcription tool that charges $0.006 per minute, where a customer's usage falls hard between quarters with no cancellation in the record:
| Quarter | Minutes used | Monthly charge |
|---|---|---|
| Q1 | 50,000 | ~$300 |
| Q3 | 8,000 | ~$48 |
The drop from ~$300 to ~$48 flows straight into contraction MRR, and the logo stays in the retained column the whole time.
One caution before the patterns.
Not every usage dip is a churn signal. Seasonal products have real valleys, like tax prep, event tools, and campaign-driven AI use, and a quiet month can be nothing more than the calendar. The pattern that flags risk is a decline that never recovers to the prior baseline, not a single slow month.
Usage decay
Usage decay is a steady month-over-month decline in a customer's activity with no cancellation. The customer keeps the account, logs in now and then, and uses a little less each cycle. The revenue ends up a fraction of where it started.
It's the hardest pattern to catch, because no single month looks alarming. Say a 10% drop one month reads as noise. Four of them in a row is a customer on the way out. The decline only shows up in a trend line, the one view a cancellation report doesn't give you.
Drop-to-floor
Drop-to-floor is when a customer falls to your free or minimum tier and settles there. They haven't left, but they've stopped paying anything meaningful, and they tend to stay that way for a long time.
A drop-to-floor customer is worse than a clean cancellation in one way. They still use support, infrastructure, and roadmap attention, and they still count toward your active base, so they inflate your retention numbers while paying almost nothing.
A customer who cancels at least frees up the cost of serving them.
Commit shortfall
Commit shortfall is when a customer buys a usage commitment and keeps using less than they paid for. They prepaid for a volume of API calls, minutes, or credits, and they keep using less than they committed to.
The shortfall is a renewal-cancel signal you can read months ahead.
Consider a customer using 40% of a commit. They won't renew at the same level, and may not renew at all. This pattern comes with a deadline, the renewal date. That makes it the most actionable of the three, if you track the draw-down rate against the commit.
How to measure retention for a usage-based product (The UBP Retention Meter)
The UBP Retention Meter is three metrics that together catch the silent loss a logo-churn metric misses, namely usage trend, revenue churn rate, and contraction MRR.
You work through them in order, because each one answers a different question at a different time:
| Metric | What it tells you | When it tells you |
|---|---|---|
| Usage trend | A decline is coming | 30 to 60 days early |
| Revenue churn rate | How much revenue left | At period close |
| Contraction MRR | The loss came from still-active accounts | At period close |
This approach has one hard requirement.
Your billing platform has to track usage per customer. If your metering data only exists as a total-platform number, you can't segment by cohort, and you lose the early warning entirely. The whole framework rests on customer-level usage data.
Signal 1: Usage Trend
Usage Trend is the rolling direction of each customer's activity, and it's the only signal that moves before revenue does. Measure it as a 30-day rolling average per customer or cohort, then watch the slope across months.
A flat or rising trend is a healthy account. A trend declining two periods in a row, without recovering to baseline, is the earliest warning you'll get. Because it leads revenue by weeks, it decides whether you step in early or read about the loss later in a contraction report.
Signal 2: Revenue Churn Rate
Revenue churn rate is the percentage of recurring revenue lost in a period, and it's the metric that replaces logo churn on a metered product. It counts dollars instead of customers, so a customer who halves their usage shows up at half their value.
Logo churn asks how many customers left. Revenue churn asks how much revenue left.
On a consumption product those two numbers diverge hard, because most of your loss comes from shrinking accounts while few customers close. Tracking revenue churn turns the silent contraction into a number you can report.
Signal 3: Contraction MRR
Contraction MRR is the signal that names the still-active accounts behind the loss, the segment your other two signals point at but don't isolate. It catches exactly the accounts a customer count calls healthy and a revenue chart calls a problem.
That makes it the number you hand to the team that owns the save, because it points straight at accounts you can still reach. A churned customer is gone, but a contracting one is still logging in, so a usage-review call or a smaller-plan offer can still turn them around.
The full formula and the downgrade-versus-cancel distinction live in the contraction MRR breakdown.
Usage-based pricing vs subscription: which is better for retention?
Usage pricing keeps more early customers but hides their later decline, while a flat subscription shows every cancellation at once and gives up that early edge. Neither one wins in the abstract.
The right choice depends on your usage pattern and whether you can instrument contraction.
A flat subscription forces a binary choice on a fixed schedule, stay or cancel, so every churn event is visible and dated. Usage pricing spreads that same loss across months of slow decline, which makes churn harder to spot but opens real expansion upside when customers grow.
Landbot launched usage pricing for all new users and measured the results after three months. Net revenue retention rose 26%. The team also reported months with large MRR contractions from usage volatility, the exact signal a flat subscription would never produce.
The same usage pricing that lifted retention produced those swings.
The comparison breaks down for products where usage doesn't track value. A customer might lean on your AI tool in week one while learning it, then use it lightly but happily after.
The falling usage means they've mastered it, so reading the trend means knowing which signals track real value rather than raw volume.
Whichever model you run, net revenue retention works best as the top-level metric for a metered product, because it nets expansion against contraction in one number.
FAQ
How does usage-based pricing affect net revenue retention?
Usage pricing makes net revenue retention more volatile in both directions.
Does usage-based pricing work for B2C products specifically?
Usage pricing fits B2C consumption products but carries more measurement risk there than in B2B.
Can I use a cancel flow to recover a usage-based pricing customer?
A cancel flow only works on a customer who starts a cancellation, and usage decay produces no cancellation to intercept. The customer just goes quiet, so there's no cancel screen to route them through.