SaaS Churn Rate: Logo vs Revenue, Gross vs Net Churn

The same SaaS business can honestly report a 10% churn rate and a negative one in the same quarter. It depends entirely on which of four churn definitions you pick — logo or revenue, gross or net — and each answers a different question. Churn is the single most abused number in SaaS reporting precisely because "the churn rate" is not one metric; it is a family of four, and quoting one without saying which you mean is how boards get misled and benchmarks become noise.

This guide defines the four churn measurements plainly, shows how the same customer base produces very different numbers under each, and lays out the measurement decisions — denominator, period, netting — that silently move the figure. The goal is a churn number you can defend in a board meeting and compare against a benchmark without comparing apples to oranges, the same discipline the metric-definition-drift guide applies to every core metric.

The four churn definitions, in plain terms

Churn splits along two independent axes. The first is what you count: logo churn (also called customer churn) counts cancelled accounts one at a time regardless of contract value, while revenue churn weighs each loss by the MRR or ARR it removes. As ChartMogul puts it, customer churn "counts the number of customers who have cancelled," whereas revenue churn measures the financial impact of those cancellations on your recurring revenue base (ChartMogul). The two can point in opposite directions in the same period.

The second axis is whether you net out expansion. Gross churn counts only losses — cancellations and downgrades — and is always a positive drag. Net churn subtracts expansion revenue (upsells, seat growth, usage-based increases) from those losses, so it can turn negative when a product expands faster than it leaks. Cross the two axes and you get four distinct metrics: logo churn, gross revenue churn, net revenue churn, and the retention mirrors of each. Each is legitimate; each answers a different question; none is "the" churn rate on its own.

How the same base produces four different numbers

Take a company starting the month with 500 customers and $100,000 MRR. It loses 15 customers worth $3,000 of MRR, one existing customer downgrades by $1,000, and its remaining customers expand by $6,000. Logo churn is 15 ÷ 500 = 3%. Gross revenue churn is ($3,000 + $1,000) ÷ $100,000 = 4%. Net revenue churn is ($4,000 − $6,000) ÷ $100,000 = −2%. Three of those numbers describe leakage; one is negative. All four are correct, and a reader who is not told which one you are quoting cannot know whether the business is bleeding or compounding.

This is why a churn figure without its type is close to meaningless. A competitor boasting "we cut churn to 1%" has told you nothing until you know whether that is logo or revenue, gross or net, monthly or annual. The netting axis is the most treacherous: a headline net-churn number can look excellent while gross churn is quietly terrible, because a few expanding enterprise accounts are papering over a long tail of small customers walking out the door — the exact pattern the expansion-revenue and NRR guide exists to keep honest.

The denominator decides more than you think

Once you fix the churn type, the denominator quietly sets the answer. The cleanest convention is customers (or MRR) at the start of the period, counting only losses from that starting cohort and excluding anyone acquired mid-period — otherwise new customers who had no real chance to churn dilute the rate and flatter the number. Mixing new acquisitions into the denominator is one of the most common ways churn gets understated, especially in fast-growing companies where the new-customer inflow is large relative to the base.

Growing companies face a structural bias here: rapid acquisition inflates the base and makes churn look lower than the underlying retention actually is, a distortion that reverses the moment growth slows. Some teams use an average of start and end base to smooth this, but that choice must be documented, because it produces a different number than the start-of-period convention. As one SaaS-metrics reference notes, calculating churn "correctly" hinges on defining the cohort and the window before you divide anything (SoftwareEquity). Write the denominator rule down once and apply it everywhere.

Monthly, annual, and the compounding trap

Churn compounds on a shrinking base, so monthly and annual figures do not convert linearly — and treating them as if they do is a recurring reporting error. Three percent monthly gross churn does not annualise to 36%; it compounds to 1 − 0.97^12 ≈ 30.6%, because each month's churn applies to what survived the previous month. The gap widens as the monthly rate rises, so quoting "3% monthly, roughly 36% a year" overstates annual leakage by five-plus points and quietly corrupts any forecast built on it.

Pick the cadence your business decisions run on and convert explicitly. Month-to-month products should measure monthly for fast feedback but state the compounded annual equivalent when comparing to annual-contract peers. Annual-contract businesses should measure annually but watch for intra-year timing effects the annual snapshot hides. The rule is the same one that governs retention-rate definitions: fix the window, document the conversion, and never let a monthly and an annual figure be quoted under the same label as though they were interchangeable.

Which churn metric to lead with

There is no single right churn metric — there is a right pair. Lead with gross revenue churn and logo churn together, and treat net revenue churn (or its mirror, net revenue retention) as the growth-quality metric layered on top. Gross revenue churn is your unavoidable leakage; it is what you must out-run with new sales and expansion, and it is the honest input to any lifetime-value calculation. Logo churn alongside it tells you whether the leakage is broad-based across the customer count or concentrated in a few large accounts.

Net revenue churn earns its place as the headline only when it is reported with its gross counterpart in the same view, so a reader can see how much expansion is masking. A business at −2% net and 8% gross revenue churn is not the same as one at −2% net and 3% gross, even though the headline is identical — the first is running a leaky bucket under a strong tap. Net-negative churn is the north-star for durable economics, but only when the gross number underneath it is also healthy, which ties directly to how you pick and frame a north-star metric for the whole company.

Instrument churn once, define it once

The practical fix is a written churn definition in your metrics catalog that fixes all four choices: the type (logo or revenue), the netting (gross or net of expansion), the period and its conversion rule, and the denominator (start-of-period base, new customers excluded). Those four decisions fully determine the number; any left ambiguous means different people report different churn under the same word, and every benchmarking conversation devolves into arguing about definitions instead of the business.

Instrumenting this well means your billing or subscription events — cancellations, downgrades, upgrades, reactivations — are captured cleanly enough to reconstruct each component, not just a net MRR delta. Once churn is defined and measured consistently, it becomes a trustworthy input to everything downstream: LTV, payback, and the forward projections in the predictive-churn leading-indicators guide. A churn number you cannot reconstruct from its parts is a number you cannot defend — and in SaaS, an indefensible churn rate is worse than none, because it drives real decisions with false confidence.