How much MRR do failed payments cost a SaaS company?

A SaaS company with $100,000 of MRR that sees an assumed 2% of its monthly billings fail, and recovers 53% of them (the baseline rate in Recurly’s published enterprise analysis), loses $940 of MRR every month, or $11,280 of MRR by the end of one year. That is about 11% of its revenue run-rate gone without a single cancellation click.

Rules of thumb

• Leakage = MRR x monthly failure rate x (1 – recovery rate) x 12. At $100,000 MRR and 2% failing, every 1% of failed MRR you do not recover costs $240 of MRR a year (arithmetic from the table below).
• Each 15 points of recovery rate is worth $3,600 of MRR a year per $100,000 of MRR at a 2% failure rate, and double that at 4%.
• In Recurly’s analysis of one big-box retailer’s membership transactions, optimised retry strategies lifted recovery from about 53% to 71% (Recurly, April 2026); in the model, at a 2% failure rate, that is $4,320 of MRR a year kept per $100,000.
• Stripe data published by Churnkey puts SaaS involuntary churn at 8% of customers a year, 22% of all SaaS churn (Churnkey, November 2025). If your lost accounts are average size, that is $8,000 of MRR a year per $100,000.
• The 12-month MRR figure is a run-rate, not the revenue you missed: losing $940 a month costs $62,040 of collected revenue inside the year and leaves a $135,360 ARR hole behind it.

How much MRR do failed payments cost at different recovery rates?

At $100,000 of MRR, failed payments cost between $3,000 and $33,600 of MRR a year in this model, depending on the failure rate you assume (1%, 2% or 4% of billings a month) and the share you recover. The two Recurly rows use the recovery rates in Recurly’s April 2026 analysis: about 53% recovered by a big-box retailer’s existing retry logic on its membership subscriptions, and about 71% with optimised retries on the same transactions. That merchant is not a SaaS company, so treat both rows as reference points, not SaaS benchmarks. Recurly does not publish a SaaS first-attempt failure rate on the pages we checked, so the three failure columns are labelled assumptions: replace them with the failed-invoice total from your own billing export.

Worked row, 2% failing at 53% recovery: $100,000 x 0.02 = $2,000 of MRR fails each month; 47% of it is not recovered, so $940 of MRR is lost per month; x 12 = $11,280. The model treats each month’s failures as new accounts and ignores the small effect of a shrinking base, which is close enough below a few per cent a month.

FAQ

How much revenue do SaaS companies lose to failed payments?

Stripe data published by Churnkey puts SaaS involuntary churn at 8% of customers a year, 22% of all SaaS churn. For a $100,000 MRR company losing average-size accounts, that is about $8,000 of MRR a year, or $96,000 of ARR.

What is a good failed payment recovery rate?

Recurly’s April 2026 analysis says that if your current recovery rate is in the low-to-mid 50s, a rate in the high 60s to low 70s is an achievable target with the right retry strategy (Recurly). Measure it only on invoices whose dunning window has closed.

Is involuntary churn the same as failed payments?

No. A failed payment is a first-attempt decline; involuntary churn is the subset never recovered, where the subscription lapses. At a 2% failure rate and 53% recovery, $2,000 of MRR fails each month but only $940 becomes churn.

Should I model failed payments on MRR or on customer counts?

Use MRR when you have it. Published benchmarks are customer-count rates, and Recurly notes MRR churn and customer churn can diverge, citing 3% customer churn alongside 5% MRR churn when larger accounts leave.

Why is my Recurly churn benchmark so much lower than Stripe’s?

Recurly lists SaaS involuntary churn at 1.06% and labels it a median annual rate, while Stripe via Churnkey gives 8% a year. Recurly reports medians from its own network and Churnkey reports Stripe averages, so compare your figure with the source that matches your billing platform.

What do the published benchmarks say SaaS companies lose to failed payments?

In Stripe’s data published by Churnkey, 8% of SaaS customers are lost to involuntary churn each year, against 38% total annual churn, and B2B companies lose 6% a year (Churnkey, November 2025). The same report shows involuntary churn falling with price: 14% a year below $10 per order, 4% between $1,000 and $10,000.

Recurly’s benchmark page, updated with July 2026 network data, lists SaaS at 3.22% total churn, 2.16% voluntary and 1.06% involuntary, and labels these “median annual churn rates” (Recurly churn rate benchmarks). That is far below Stripe’s 38% annual figure, and neither page reconciles the two: Recurly reports medians from its own network, Churnkey reports Stripe averages, so compare each with your own platform’s data rather than with each other. Check the period of your own figure too: Recurly notes that a 2% monthly churn rate compounds to roughly 22% a year, not 24%. Here is what each published rate implies for a $100,000 MRR company, assuming the lost accounts are average size:

These are customer-count rates. Recurly’s own page warns that MRR churn can diverge from customer churn: its example is a 3% customer churn rate sitting alongside a 5% MRR churn rate when the accounts lost skew larger. Recurly’s own price-band data shows lower-priced plans churn involuntarily more often, so for many SaaS companies the failed-payment MRR rate can run below the customer rate.

Why does the 12-month MRR figure understate what failed payments cost?

A company losing $940 of MRR a month to failed payments misses $62,040 of collected revenue inside the first year and ends it with a $135,360 hole in ARR. The arithmetic: an account lost at the end of month 1 is missing for 11 months, one lost at the end of month 2 for 10, and so on, so the year-one revenue loss is $940 x (11 + 10 + … + 0) = $940 x 66 = $62,040.

That is why the recovery rate, not the failure rate, is the lever founders feel. Most failures are not a decision to leave: the customer wanted to stay and the card did not. The decline codes behind them are mapped cause by cause on our page on why SaaS involuntary churn is high; this page only prices them.

Which inputs move the failed-payment cost the most?

The recovery rate moves the cost most: in the table, going from 30% to 75% recovery cuts annual MRR loss by 64% at every failure rate ($16,800 to $6,000 at 2%), while halving the failure rate from 2% to 1% cuts it by 50%. Recurly’s published data points to four things that shift recovery:

• Retry logic. Recurly states that retries informed by network-level data improve recovery by 10 to 20 percentage points over single-merchant retry logic, and that 90% of recovered transactions happen within the first 10 days (Recurly, April 2026).
• Dunning window length. Recurly’s 2019 research found recovery rates were lowest for merchants with a dunning window of seven days or less, and its documentation says its default window is 10 days and that extending monthly plans to 28 days gives the retry engine more time (Recurly, 2019; Recurly docs). How long to run it is covered in how long a SaaS dunning period should be.
• Billing cycle. The same 2019 Recurly research found invoices on monthly plans had a higher probability of recovery than those on yearly plans. One failed annual invoice also carries 12 months of revenue, so annual plans change the size of each loss as well as the odds.
• Price point. Recurly’s July 2026 data shows involuntary churn of 1.30% for subscribers paying $10 to $25 a month and 0.18% above $250 a month (Recurly), so a company moving upmarket should expect a lower failure rate in the model.

How do you calculate your own failed-payment leakage in 4 steps?

Your own leakage needs 4 numbers from the billing system, all for the same 3 to 6 months: MRR at the start of each month, the MRR on invoices that failed at first attempt, the MRR on those invoices later paid, and the MRR on subscriptions cancelled for non-payment.

1. Monthly failure rate = failed-at-first-attempt MRR / starting MRR. Recurly’s documentation calls the invoice-count version the decline rate at renewal, with retries excluded.
2. Recovery rate = recovered MRR / failed MRR, counted only for invoices whose dunning window has closed.
3. Annual leakage = MRR x failure rate x (1 – recovery rate) x 12. Check it against step 4: the two should land within a few per cent.
4. Cross-check = MRR cancelled for non-payment over the same months, annualised. If it is far higher than step 3, cancellations are happening before retries finish.

Then run the step 3 result at your recovery rate plus 15 points. The gap is the most you should pay per year for better retries, card updating or outreach, before counting the year-one revenue effect above.

Is it cheaper to recover a failed payment or replace the lost MRR?

In the worked example, lifting recovery from 53% to 71% keeps $4,320 of MRR a year, which is $51,840 of ARR that does not need a new sales cycle to replace. Replacing it means new customers: count how many accounts at your average contract value make up $4,320 of MRR, then price each one with your own cost per booked demo and close rate, using a guide such as the cost per booked demo benchmarks for B2B SaaS. The recovered dollar needs no new sales cycle because the customer already chose you.

Doing the recovery work yourself costs engineering hours to configure retries and a card updater, plus someone to read the decline-code report monthly. The accounts that automation cannot clear (hard declines, unfinished authentication, a card that will not be replaced without a nudge) are where a person or an AI agent contacting the billing owner earns its cost; the method, scripts and consent points are on our page about failed payment recovery with call and SMS agents. Customers who have already cancelled are a different problem with different economics, covered in how to win back churned SaaS customers, and growth to replace lost MRR is covered on the SaaS lead generation page.

Sources

1. Recurly: Churn rate benchmarks (July 2026 network data)
2. Recurly: The case for smarter failed payment recovery (30 April 2026)
3. Recurly: Research gives insights on recurring invoice recoveries (23 July 2019)
4. Recurly documentation: Renewal and dunning benchmarks
5. Churnkey: Involuntary churn benchmarks and tactical advice (14 November 2025, Stripe data)