CohortMargin

Usage-Based SaaS Margin by Cohort

Your average customer is profitable. Run this per decile and find the ones that aren't.

Customers in this decile

Monthly plan fee ($)

Units included in plan

Units actually used

Overage price per unit ($)

Your cost per unit ($)

Fixed support/infra per customer ($)

Margin per customer

$79

Run this once per decile of your customer base, from lightest to heaviest usage. A blended average across all customers hides the loss-making tail — and that tail is usually your happiest, least likely to churn segment.

Margin %: 33.1%
Revenue per customer: $239
Cohort margin / month: $7,900

What you entered

Customers in this decile: 100Monthly plan fee ($): 99Units included in plan: 10000Units actually used: 45000Overage price per unit ($): 0.004Your cost per unit ($): 0.003Fixed support/infra per customer ($): 25

1
Usage above the included allowance
max(0, 45,000 − 10,000)= 35,000 units
2
Revenue per customer
$99 plan + 35,000 × $0.00= $239
3
Cost to serve per customer
45,000 × $0.00 + $25 fixed= $160
4
Margin per customer
$239 − $160= $79 (33.1%)
5
Cohort total (100 customers)
$79 × 100= $7,900

Result

Margin per customer: $79

This decile brings in $239 and costs $160 to serve — $79 margin per customer (33.1%), or $7,900/month across 100 customers.

Margin per customer by usage decile on a $99 plan with 10,000 units included

Every row uses the same plan: $99 a month including 10,000 units, overage billed at $0.004 per unit, a serving cost of $0.003 per unit, and $25 of fixed support and infrastructure per customer per month. Only monthly usage changes, so each row stands in for one decile of a customer base sorted from lightest to heaviest.

Units used per month	Revenue per customer	Cost to serve	Margin per customer	Margin %
2,000	$99	$31	$68	68.7%
5,000	$99	$40	$59	59.6%
10,000	$99	$55	$44	44.4%
20,000	$139	$85	$54	38.8%
45,000	$239	$160	$79	33.1%
80,000	$379	$265	$114	30.1%
150,000	$659	$475	$184	27.9%
300,000	$1,259	$925	$334	26.5%

This plan survives its own heavy users because the $0.004 overage rate sits above the $0.003 unit cost, so every extra unit still contributes. Even so the margin percentage compresses from 68.7 percent to 26.5 percent across the deciles, converging on the 25 percent spread between the overage rate and unit cost. Drop the overage rate to $0.002, below cost, and the same customers invert: the 45,000-unit decile falls to $9 of margin per customer, the 150,000-unit decile loses $96, and the 300,000-unit decile loses $246 each, or $24,600 a month across 100 accounts. That is the tail a blended average hides, and it is usually your least churn-prone segment, so the fix is repricing rather than removal. These are direct unit economics only, excluding customer acquisition cost, sales commission and support headcount beyond the fixed allocation. Illustrative only; run your own deciles with your real usage distribution and cost per unit.

A usage-based SaaS margin calculator computes profitability for one slice of your customer base at a time, so you can find the segments that lose money instead of averaging them away. Usage-based and hybrid pricing models — a monthly plan with an included allowance plus metered overage — create a fundamental problem that seat-based pricing does not: customers with identical bills can impose wildly different costs. A blended, whole-book margin figure conceals that completely, which is how companies discover a loss-making tail only when a single enterprise account's infrastructure bill lands.

The model prices one cohort at a time. Revenue per customer is the plan fee plus any metered overage, calculated as usage above the included allowance multiplied by the overage rate. Cost per customer is the usage actually served multiplied by your unit cost, plus a fixed allocation for support and infrastructure that every account consumes regardless of usage. The difference is margin per customer, reported in dollars and as a percentage, then multiplied across the cohort to give its total monthly contribution. Two structural traps show up immediately when you run real numbers through it. The first is a generous included allowance: if customers routinely use most of their allowance and the plan fee barely covers that usage, the plan is priced for light users and subsidizes heavy ones. The second is an overage rate set too close to — or below — your unit cost, which means every additional unit sold makes the account worse.

The method the source research recommends is to run this by decile rather than in aggregate: sort your customers by usage, split them into ten groups, and compute margin for each. The pattern that emerges is usually a healthy middle, a high-margin group of light users who barely touch their allowance, and a deeply negative top decile. That top decile is also, uncomfortably, the segment getting the most value and least likely to churn — so the fix is rarely to remove them. It is to reprice: raise the overage rate above unit cost, reduce the included allowance on new contracts, introduce a volume tier that reflects real cost, or add a committed-use discount that trades certainty for a rate you can actually serve. Use this to find which decile breaks and by how much before designing that change. It models direct unit economics only and excludes acquisition cost, so pair it with payback analysis for a full picture.

Averages hide the tail that hurts

A blended margin across all customers can look healthy while the top usage decile is deeply negative, because thousands of light users mask a few dozen heavy ones. Sort by usage, split into deciles, and compute each separately — the shape of that curve tells you whether your pricing survives your own product's success.

Two structural traps in usage pricing

First, an included allowance so generous that the plan fee doesn't cover typical usage — the plan is then priced for light users and subsidizes heavy ones. Second, an overage rate at or below your unit cost, which means every extra unit sold actively loses money. Either one turns growth into a margin problem instead of a margin engine.

Frequently asked questions

100 customers on a $99 plan with 10,000 units included, using 45,000 units, overage at $0.004 and cost at $0.003 with $25 fixed — what's the margin?

Overage is 35,000 units × $0.004 = $140, so revenue is $239. Cost is 45,000 × $0.003 + $25 = $160. That's $79 margin per customer (33%), or $7,900/month across the cohort — healthy, because the overage rate sits above unit cost.

Why should I calculate margin by decile instead of overall?

Because usage distributions in SaaS are heavily skewed — a small share of customers generate most of the consumption. An overall average can show a comfortable margin while your top decile loses money on every account. Deciles expose where the curve crosses zero, which is exactly where your pricing needs to change.

My heaviest users are unprofitable — should I fire them?

Usually not. Heavy users get the most value, are least likely to churn, and often drive references. The fix is repricing: raise the overage rate above unit cost, trim the included allowance on new contracts, add volume tiers that reflect real cost, or offer a committed-use discount. Model the change here before proposing it.

How is this different from the e-commerce contribution margin calculator?

That one prices a single physical order — fees, shipping, ads, returns. This one prices a recurring software customer whose cost scales with metered usage, and reports margin for a whole cohort. Same discipline of counting real cost to serve, applied to subscription rather than transactional economics.