Churn rate: formulas, 2026 benchmarks and a worked example

In October 2025 MotorK, the software company for car makers and dealers founded in Milan in 2011, published nine-month results with an awkward line in them. The €5.6 million of new committed annual recurring revenue signed since January ‘was almost entirely offset by churn in the retail segment and phasing-out non-core legacy retail customers’. Part of it was deliberate. Still, the total shrank. On 30 June 2026 committed ARR was €33.1 million, down 7.5% in twelve months.

MotorK calls it churn. Your churn rate is the share of customers you lose each month, or the share of recurring revenue. Below you’ll get the formulas for customer churn, revenue churn and net revenue retention. Then the reason 2% a month isn’t 24% a year, one month of MRR worked through in euros, and the benchmarks investors are quoting in 2026. If the jargon is new, start from the beginner’s map, from zero to a first round.

What is churn rate? Customer churn vs revenue churn

Churn comes in two versions, and they rarely agree.

Customer churn, also called logo churn, counts heads. ChartMogul divides the paying customers who cancelled during the month by the paying customers you had on day one, with two adjustments: a customer who signed up and cancelled within the month doesn’t count, and neither does one who left and came back. Free trials and freemium users stay out.

Revenue churn counts money, usually MRR, the monthly recurring revenue (ARR is the same figure times twelve). ChartMogul’s illustration: three customers pay $20, $30 and $50 a month. Now the $50 customer cancels. You’ve lost one customer in three, and half the revenue.

• Customer churn rate. Count the customers who left during the month, then divide by the number you had on the 1st.
• Gross MRR churn rate. The same in money: churned MRR plus contraction MRR, over the MRR you opened the month with.
• Net MRR churn rate. Knock expansion and reactivation MRR off that numerator before dividing. When upsells beat losses, it drops below zero.

Contraction is a customer who stays but pays less, after a downgrade or fewer seats; expansion is the reverse. Andreessen Horowitz prefers the gross number: ‘Gross churn estimates the actual loss to the business’, while net churn ‘understates the losses (as it blends upsells with absolute churn)’. Report both.

Net revenue retention: the number investors ask for

Net revenue retention measures the opposite: what you keep. Pick out the customers who paid you a year ago and forget anyone who signed since. What does that group pay now, against what it paid then? SaaS Capital divides December MRR from last December’s customers by total MRR last December, upsells and price increases included.

Gross revenue retention (GRR) runs the same sum with each customer capped at last year’s payment, so it can’t pass 100%. NRR can. In ChartMogul’s example four customers paid a combined 770 a year ago and pay 800 today: NRR of about 103.9%. Churn, in SaaS Capital’s shorthand, is 1 minus gross retention.

Why the fuss? Because NRR above 100% means your existing base grows on its own, before sales signs anyone new. In SaaS Capital’s 2025 data, companies with NRR between 110% and 120% grew at a median 30% a year. Below 90%, the median was 15%.

How to calculate churn rate: a worked MRR example

One invented month makes the arithmetic concrete. Picture a Bari startup with booking software for gyms and yoga studios. It starts March with 400 paying customers, who bring in €40,000 of MRR between them: €100 each, on average.

MRR movement in March	Amount	What happened
Starting MRR, 1 March	€40,000	400 paying customers
+ New MRR	€5,000	40 new customers
+ Expansion MRR	€1,200	30 studios add a second room or SMS reminders
− Contraction MRR	€400	Downgrades to the cheaper plan
− Churned MRR	€1,200	12 customers cancel
Ending MRR, 31 March	€44,600	428 paying customers

Four rates come out of that table:

• Customer churn: 12 of the 400 customers left, so 3.0%.
• Gross MRR churn is 4.0%. Cancellations took €1,200 and downgrades €400, so €1,600 of the €40,000 went.
• Net MRR churn takes off the €1,200 of expansion, leaving €400 lost: 1.0%.
• Monthly NRR is 99.0%. The customers you had on 1 March pay €39,600 between them by the 31st, against €40,000.

MRR grew 11.5% in a month, and plenty of board decks would stop there. Look at the churn line instead. Twelve times the monthly rate won’t give you annual churn. Each month you lose a share of a base that has already shrunk, so Wall Street Prep writes annual churn = 1 − (1 − monthly churn)^12. Lose 3% of customers a month and after a year 30.6% are gone, not 36%. At 2% it’s 21.5%, David Skok’s ‘about 22%’.

Repeat March for a year and gross revenue retention lands near 61% (0.96 to the 12th power), with NRR around 89%, far below the medians in the next section. Strong new sales can hide that for a long time. Scale it up and you get MotorK’s sentence from 2025.

SaaS churn benchmarks for 2026: what the data says

Read the footnotes first. A benchmark describes the companies in its sample, not yours. Two respected datasets from 2025 land almost 20 points apart on median B2B net revenue retention.

Source and date	Sample	What it found
SaaS Capital, September 2025	1,000+ private B2B SaaS companies above $1M ARR	Median NRR 101%, GRR 91%
SaaS Capital, same survey, by contract size	Annual contract value under $12k vs over $250k	NRR 98% vs 106%; GRR 90% vs 95%
SaaS Capital, April 2026	Bootstrapped companies with $3M to $20M ARR	Median NRR 103%, GRR 91%
ChartMogul, December 2025	About 2,700 B2B SaaS companies above $250k ARR	Median NRR 82%, top quartile 97%
Lenny Rachitsky, June 2020	20 growth leaders plus public company data	12-month NRR good / great: SMB and mid-market 90% / 110%, enterprise 110% / 130%
Clouded Judgement, 2 October 2026	Listed cloud software companies	Median NRR 110%

Part of the gap is the cut-off. ChartMogul’s sample starts at $250,000 of ARR and was measured through 2025; SaaS Capital’s starts at $1 million. Price point matters even more. In ChartMogul’s December 2025 analysis, products above $250 a month retained like classic B2B SaaS, while those under $50 a month retained ’20+ points worse than either B2B or B2C SaaS’.

Before you compare, convert a yearly figure back into months. A 91% GRR means losing about 0.8% of revenue a month. Skok’s alarm goes off when net revenue churn runs ‘above 2% per month’.

How investors read churn and NRR at seed and Series A

At seed you rarely have twelve months of history, so there’s no real annual NRR to show. Investors read cohorts instead: customers grouped by the month they signed, with the share still paying plotted over time. Andreessen Horowitz looks for ‘stabilization of retention in each cohort after a period such as 6 or 12 months’, and for ‘newer cohorts performing progressively better than older cohorts’.

At Series A the bar is written down. CRV’s March 2026 guide calls 100% NRR the baseline, 110% to 120% competitive and 120% or above premium, for B2B SaaS where a competitive round ‘generally starts at $2 million to $5 million in ARR’. High churn, it says, ‘raises immediate questions about product-market fit’. It also tells founders to separate failed-payment churn from real cancellations in the data room.

Churn also sets your other numbers. Average customer life is roughly 1 ÷ monthly churn, so 2% a month means about 50 months, and lifetime value (LTV) moves with it: see our guide to unit economics: CAC, LTV and payback. NRR also feeds what your ARR is worth in SaaS valuation multiples, and the seed checklist is in the startup metrics investors want at pre-seed and seed.

Why customers churn, and what brings the rate down

CRV’s split comes first. You fix a bounced card in billing; a customer who chose to leave is a different problem. Then go down the usual list:

• Failed payments. Cards expire and accounts run dry. Turn on automatic retries; Stripe’s recommended default for its Smart Retries is 8 tries within 2 weeks. Then email customers before their card expires.
• Customers who never got going. Do most cancellations come in the first three months? Then fix onboarding before anything else. Count how many new accounts reach real use in week one.
• Monthly plans. ChartMogul finds median NRR 10 to 20 percentage points higher on annual plans than on monthly ones, so make the annual option worth choosing.
• Pricing with nowhere to grow. Skok recommends ‘a pricing scheme that has a variable axis, such as the number of seats used’, so revenue grows with the customer. Our guide to SaaS pricing strategy compares the models.
• Customers you shouldn’t have signed. Some churn is a sales problem. MotorK chose to phase out legacy retail clients. Dropping a bad-fit account can be right, as long as your numbers say so.

Then put churn where it bites: the cash forecast. Every cancellation cuts the cash you collect, which raises net burn and shortens runway (our guide to burn rate and runway has the formula), and our guide to a startup financial model you can defend shows the structure. Prefer a template? We sell one for €129, the startup financial model with cap table. It’s an Excel file: five years, month by month, set up for SaaS, with runway, scenarios and a simulator for the round.

Checklist: your churn numbers before an investor meeting

1. Pull 12 to 24 months of subscription events out of your billing system. Every sign-up, upgrade and downgrade goes in, plus cancellations and reactivations.
2. Run the numbers month by month: customer churn, gross MRR churn, net MRR churn. Don’t change definitions halfway.
3. Take the customers you had twelve months ago and work out their NRR and GRR.
4. Keep failed payments apart from people who actually cancelled.
5. Build a cohort table by signup month. Do the curves flatten? Do newer cohorts hold up better?
6. Annual benchmark in front of you? Convert your monthly rate first, with 1 − (1 − m)^12.
7. Note why customers leave (the top three reasons will do) and what you’re changing about each one.