SaaS churn, how to measure it and build your own baseline

What SaaS churn actually measures

Churn is the share of something you had at the start of a period that you no longer have at the end. The something matters. It can be customers, it can be recurring revenue, and recurring revenue can be counted before or after expansion. Each choice produces a different number from the same data.

• Logo churn counts accounts. If 10 of 200 accounts leave, logo churn is 5 percent, whatever those accounts paid.
• Gross revenue churn counts recurring revenue lost through cancellations and downgrades, with no credit for expansion.
• Net revenue churn subtracts expansion from existing accounts, so it can be negative when upsell outweighs loss.

All three are useful. None of them is the churn rate. A team that reports one of them without saying which leaves every reader to guess.

Why published churn figures disagree

Published SaaS churn figures differ for structural reasons, not because some are right and others are wrong. Before you compare yourself with any outside number, check how it was built.

Segment and contract value

A product sold to small businesses on a card, month to month, loses accounts for reasons that have nothing to do with the product: the business closes, the owner cuts costs, the card expires. A product sold to large companies on annual contracts has procurement, onboarding and switching costs that slow churn down. Mixing the two in one average produces a number that describes neither.

Logo churn versus revenue churn

If small accounts leave more often than large ones, which is common, logo churn will be higher than revenue churn. A figure that does not say which one it is cannot be compared with yours.

Monthly versus annual

A monthly rate and an annual rate are not related by multiplying by twelve, because each month starts from a different base. Annual contracts also churn in lumps, at renewal dates, so a monthly figure for an annual business swings from month to month.

Definitions of a churned account

Some teams count a subscription as churned the day the customer clicks cancel. Others count it at the end of the paid period. Some count failed payment terminations, others leave them out, and some never count paused accounts as churned at all. Each choice moves the result.

Who is in the data

Figures built from surveys depend on who answered. Figures built from one vendor's customer base depend on what kind of companies use that vendor. Neither describes the market as a whole.

How to build your own baseline from billing data

Your own billing history is the only source where you control every definition. Building a baseline takes a few steps, and once they exist you can rerun them every month.

• Fix the period. A calendar month is easiest. Use the same one every time.
• Normalize terms. Count annual and quarterly plans at their monthly equivalent, so a 1,200 USD yearly plan is 100 USD of MRR each month of the term.
• Define churn events. Decide whether churn happens at cancel or at the end of the paid period, and whether a failed payment termination counts. Write the rule down.
• Separate involuntary churn. Tag subscriptions that ended after a failed payment, so you can see the share that billing work can remove.
• Treat pauses and comps explicitly. A paused account bills nothing, so either remove it from MRR while paused or keep it in its own bucket.
• Segment. Split by plan, by billing interval and by account size band, because their churn behaves differently.

Then calculate logo churn, gross revenue churn and net revenue churn for every month you have. Twelve to twenty four months of history is enough to see seasonality, the effect of price changes and the months with large annual renewals.

Gross and net revenue churn side by side

Gross revenue churn = (Churned MRR + Contraction MRR) / Starting MRR

Net revenue churn = (Churned MRR + Contraction MRR minus Expansion MRR) / Starting MRR

For example, a company starts a month with 200,000 USD of MRR. Cancellations remove 6,000 USD, downgrades remove 2,000 USD and upgrades from existing accounts add 5,000 USD.

A net figure of 1.5 percent looks calm. The gross figure says that 4 percent of existing revenue walked out in one month, and expansion covered a little more than half of it. Report both. The full derivation, including net revenue retention, is in the guide on the net revenue churn formula.

The cohort method

Monthly rates mix accounts of every age. A cohort view follows a group of accounts that started in the same month and tracks what share of their revenue remains over time. It answers questions a monthly rate cannot, such as whether accounts acquired after a pricing change stay longer, or whether most churn happens in the first three months.

Build it as a table. Each row is a start month. Each column is months since start. Each cell is the cohort's current MRR divided by its starting MRR.

For example, a cohort starts with 10,000 USD of MRR. After three months it holds 8,500 USD, after six months 8,000 USD and after twelve months 8,400 USD because surviving accounts expanded.

The shape matters more than the last number. A steep drop in the first months points to onboarding or to cards that fail at the first renewal. A curve that flattens and then rises means the accounts that stay grow, and the early loss is where the work is. Compare cohorts with each other, not with an outside figure.

When logo churn and revenue churn tell different stories

For example, 300 accounts start the month. 15 leave, a logo churn of 5 percent. If those 15 paid 50 USD each, they took 750 USD. If the base is 60,000 USD of MRR, revenue churn from cancellations is 1.25 percent. The business is losing many small accounts and few dollars. The response is different from a business that loses one large account and a lot of dollars. Neither situation shows up if you read only one number.

Logo churn is the better early signal of product fit among small accounts. Revenue churn is the better measure of financial health. Track both, and segment both by plan.

What moves your baseline, and what only looks like it

Once a baseline exists, changes become visible. Some are real, and some come from data problems.

• Real improvements come from retry rules by decline code, pre-expiry card emails, a cancel flow with pause and lower plan options, and onboarding that gets accounts to value quickly.
• False improvements come from upgrades that were recorded but never billed, which inflate expansion, and from paused or comped accounts that still count as active MRR.
• False deterioration comes from annual renewals bunching in one month, and from refunds booked as churn when the subscription did not end.

Involuntary churn deserves its own line in the baseline, because it can be reduced with billing configuration alone. The guide to involuntary churn shows how to measure it, and failed payment recovery explains the retry schedule. For the voluntary side, see how to reduce churn with cancel flows and save offers.

A reporting routine that holds up

• Report logo churn, gross revenue churn and net revenue churn every month, with the same definitions.
• Split gross revenue churn into voluntary and involuntary.
• Show a trailing three month average next to the monthly figure.
• Keep a cohort table by start month and update it monthly.
• Reconcile expansion with paid invoices, so unbilled upgrades do not count.

A subscription analytics view that reads the billing system directly does most of this without spreadsheets, and a billing leakage review catches the data problems that make a baseline lie. Over time, your own trend becomes the only benchmark that matters: whether this quarter's cohorts keep more revenue than last year's.

Find the part of your churn that billing can fix

RevenueLeak connects read-only to your billing system, separates churn after failed payments from deliberate cancellations, and shows in USD per month how much comes from failed renewals, expiring cards, unbilled proration, forever coupons and the other leak categories, with the accounts behind each figure.